United States
Environmental Protection Region III Final Report
Agency Philadelphia, PA December 5,2000
&EPA Hydrodynamic and
Water Quality Model of
Christina River Basin
PA
MD
~ UJ
-------
Acknowledgments
Funding for this study was provided through the U.S. Environmental Protection Agency, Region 3, under
contracts #68-C7-0018 and #68-C-99-249. The EPA Work Assignment Manager was Mr. Leo
Essenthier. The EPA Region 3 TMDL Coordinator was Mr. Tom Henry. Technical direction was
provided by Mr. George Golliday and Mr. Larry Merrill, EPA Region 3. Coordination of the various
agencies involved in the study was provided by Mr. David Pollison, Mr. Thomas Fikslin, and Mr. Jason
Tsai of the Delaware River Basin Commission. Data and technical review were provided by Mr. Richard
Greene, Mr. Hassan Mirsajadi, and Ms. Xia Xie of Delaware Department of Natural Resources and
Environmental Control and by Mr. William Goman and Ms. Nancy Crickman of Pennsylvania
Department of Environmental Protection. Field monitoring data from August 1997 and solar radiation
data were provided by Dr. John Davis, Widener University. Discharge monitoring data for some
Pennsylvania NPDES discharges were provided by Mr. Bob Struble, Brandywine Valley Association.
ArcINFO GIS coverage for the NPDES point sources and HSPF watersheds as well as water withdrawal
data were provided by Mr. Jerry Kaufman, New Castle County Water Resources Agency.
This report should be cited as:
USEPA. 2000. Hydrodynamic and Water Quality Model of Christina River Basin. Final Report. United
States Environmental Protection Agency, Region III, Philadelphia, PA. December 5, 2000. 464 pp.
ii
-------
CONTENTS
Page
1 INTRODUCTION 1-1
1.1 Purpose of the Study 1-2
1.2 Scope of the Study 1-2
1.3 The EFDC Model Package 1-3
2 THE CHRISTINA RIVER BASIN SYSTEM 2-1
2.1 Physical Description 2-1
2.2 Hydrology 2-1
2.3 Eutrophication Processes 2-1
2.4 Sediment-Water Interactions 2-2
3 EFDC HYDRODYNAMIC MODEL 3-1
3.1 General 3-1
3.2 Hydrodynamics and Salinity and Temperature Transport 3-4
3.3 Sediment Transport 3-5
3.4 Water Quality and Eutrophication Simulation 3-5
3.5 Toxic Contaminant Transport and Fate 3-6
3.6 Finfish and Shellfish Transport 3-6
3.7 Near-field Discharge Dilution and Mixing Zone Analysis 3-7
3.8 Spill Trajectory and Search and Rescue Simulation 3-7
3.9 Wetland, Marsh, and Tidal Flat Simulation 3-7
3.10 Nearshore Wave-induced Currents and Sediment Transport 3-8
3.11 User Interface 3-8
3.12 Preprocessing Software 3-9
3.13 Program Configuration 3-9
3.14 Run-Time Diagnostics 3-9
3.15 Model Output Options 3-9
3.16 Postprocessing, Graphics and Visualization 3-10
3.17 Documentation 3-10
3.18 Computer Requirements 3-10
4 EFDC WATER QUALITY MODEL 4-1
4.1 Introduction 4-1
4.2 Conservation of Mass Equation 4-4
4.3 Algae 4-6
4.4 Organic Carbon 4-11
4.5 Phosphorus 4-16
4.6 Nitrogen 4-21
4.7 Silica 4-26
4.8 Chemical Oxygen Demand 4-28
4.9 Dissolved Oxygen 4-29
4.10 Total Active Metal 4-31
4.11 Fecal Coliform Bacteria 4-32
4.12 Method of Solution 4-32
4.13 Macroalgae (Periphyton) State Variable 4-33
iii
-------
5 EFDC SEDIMENT PROCESS MODEL 5-1
5.1 Depositional Flux 5-3
5.2 Diagenesis Flux 5-5
5.3 Sediment Flux 5-6
5.4 Silica 5-18
5.5 Sediment Temperature 5-19
5.6 Method of Solution 5-20
6 DATABASES 6-1
6.1 Introduction 6-1
6.2 Bathymetric and Stream Geometry Data 6-1
6.3 Tide Data 6-2
6.4 Climatology Data 6-2
6.5 Stream Flow Data 6-2
6.6 In-stream Water Quality Monitoring Data 6-3
6.7 Discharge Monitoring Data for Point Sources 6-3
7 LOADS TO TIIH SYSTEM 7-1
7.1 Nonpoint Source Loads 7-1
7.2 Point Source Loads 7-1
7.3 Water Withdrawals 7-2
7.4 Atmospheric Loads 7-2
8 DELAWARE RIVER BOUNDARY CONDITIONS 8-1
9 MODEL CALIBRATION 9-1
9.1 Computational Grid 9-1
9.2 Model Configuration 9-1
9.3 Calibration Period 9-2
9.4 Hydrodynamic and Hydraulic Calibration 9-3
9.5 Water Quality Calibration Results 9-6
9.6 Diel Dissolved Oxygen Calibration Results 9-11
9.7 Sediment Oxygen Demand and Benthic Nutrient Flux Rates 9-13
10 MODEL VALIDATION 10-1
10.1 Validation Period 10-1
10.2 Hydrodynamic and Hydraulic Validation 10-1
10.3 Water Quality Validation Results 10-3
10.4 Diel Dissolved Oxygen Validation Results 10-7
10.5 Sediment Oxygen Demand Rates 10-8
11 STATISTICAL SUMMARY OF CALIBRATION AND VALIDATION 11-1
11.1 Mean Error Statistic 11-1
11.2 Absolute Mean Error Statistic 11-2
11.3 Root-Mean-Square Error Statistic 11-2
11.4 Relative Error Statistic 11-2
11.5 Evaluation of Results 11-3
11.6 Comparison with Other Models 11-3
11.7 References for Section 11 11-15
iv
-------
12 SUMMARY AND CONCLUSIONS 12-1
12.1 Summary of EFDC Hydrodynamic and Water Quality Model Framework 12-1
12.2 Summary of Hydrodynamic Results 12-2
12.3 Summary of Water Quality Results 12-3
12.4 Sources of Uncertainty 12-3
12.5 Conclusions 12-5
13 REFERENCES 12-1
APPENDIX A - Water Quality Calibration Results: Transect Plots
APPENDIX B - Water Quality Calibration Results: Time-Series Plots
APPENDIX C - Sediment Flux Rates: Time-Series Plots
APPENDIX D - Sediment Flux Rates: Transect Plots
APPENDIX E - Water Quality Validation Results: Transect Plots
APPENDIX F - Water Quality Validation Results: Time-Series Plots
APPENDIX G - Listing of Input Data Files for the TMDL Permit Limits Simulation
APPENDIX H - Data Set for TMDL Simulations
APPENDIX I - Organic Carbon / CBOD5 Study
FIGURES
Figure 1-1 Christina River Basin study area 1-4
Figure 1-2 303(d) waters listed for nutrients, low DO (NPS and PS as cause) 1-5
Figure 1-3 303(d) waters listed for nutrients, low DO (PS as possible cause) 1-6
Figure 3-1 Primary modules of the EFDC model 3-2
Figure 3-2 Structure of the EFDC hydrodynamic model 3-2
Figure 3-3 Structure of the EFDC water quality model 3-2
Figure 3-4 Structure of the EFDC sediment transport model 3-3
Figure 3-5 Structure of the EFDC toxic model 3-3
Figure 4-1 Schematic diagram for the EFDC water column water quality model 4-2
Figure 4-2 Velocity limitation function for Monod equation and 5-parameter logistic function 4-36
Figure 5-1 Sediment layers and processes included in sediment process model 5-2
Figure 5-2 Schematic diagram for sediment process model 5-3
Figure 5-3 Benthic stress (a) and its effect on particle mixing (b) as a function of overlying
water column dissolved oxygen concentration 5-6
Figure 6-1 Locations of USGS stream gaging stations 6-9
Figure 6-2 Locations of STORET water quality stations 6-10
Figure 6-3 Locations of the 120 NPDES point source included in the model 6-11
Figure 6-4 Locations of NPDES point sources in August 1997 study (Davis 1998) 6-12
Figure 7-1 Watershed delineation for HSPF model of Christina River Basin 7-13
Figure 8-1 Boundary concentrations for CYA, DIA, and GRN algae 8-3
Figure 8-2 Boundary concentrations for RPC, LPC, and DOC 8-4
Figure 8-3 Boundary concentrations for RPP, LPP, and DOP 8-5
v
-------
Figure 8-4 Boundary concentrations for P4T, RPN, and LPN 8-6
Figure 8-5 Boundary concentrations for DON, NH4, and N03 8-7
Figure 8-6 Boundary concentrations for SUU, SAA, and COD 8-8
Figure 8-7 Boundary concentrations for DOO, TAM, and FCB 8-9
Figure 9-1 Schematic of EFDC model of Christina River Basin (point source locations) 9-15
Figure 9-2 Locations of consumptive use water withdrawals 9-16
Figure 9-3 Model-data comparison of tides at Port of Wilmington and Newport 9-17
Figure 9-4 Model-data hydrographs, Brandywine Creek and E.Br. Brandywine Creek 9-18
Figure 9-5 Model-data hydrographs, E. Br. Brandywine Creek and W. Br. Brandywine Creek .9-19
Figure 9-6 Model-data hydrographs, Christina River and White Clay Creek 9-20
Figure 9-7 Model-data hydrographs, Red Clay Creek 9-21
Figure 9-8 Monitoring stations used for model-data time-series comparisons 9-22
Figure 9-9 Diel dissolved oxygen at USGS monitoring stations 9-23
Figure 9-10 Water temperature at USGS monitoring stations 9-24
Figure 9-11 Periphyton biomass at USGS monitoring stations 9-25
Figure 9-12 Periphyton limitation factors (Modena gage, W. Br. Brandywine Cr.) 9-26
Figure 9-13 Periphyton limitation factors (Downingtown gage, E. Br. Brandywine Cr.) 9-27
Figure 9-14 Periphyton limitation factors (Chadds Ford gage, Brandywine Cr.) 9-28
Figure 9-15 Periphyton limitation factors (Smalley's Pond, Christina River) 9-29
Figure 9-16 Periphyton limitation factors (W. Br. Red Clay Creek) 9-30
Figure 9-17 Model-data diel DO comparison, Brandywine Creek East Branch 9-31
Figure 9-18 Model-data diel DO comparison, Brandywine Creek West Branch 9-32
Figure 9-19 Model-data diel DO comparison, Red Clay Creek West Branch 9-33
Figure 9-20 Model-data diel DO comparison, Red Clay Creek West Branch 9-34
Figure 9-21 Model-data diel DO comparison, White Clay Creek East Branch 9-35
Figure 9-22 Model-data diel DO comparison, White Clay Creek East Branch 9-36
Figure 10-1 Model-data comparison of tides at Port of Wilmington and Newport 10-9
Figure 10-2 Model-data hydrographs, Brandywine Creek and E. Br. Brandywine Creek 10-10
Figure 10-3 Model-data hydrographs, E. Br. Brandywine Creek and W. Br. Brandywine Creek 10-11
Figure 10-4 Model-data hydrographs, Christina River and White Clay Creek 10-12
Figure 10-5 Model-data hydrographs, Red Clay Creek 10-13
Figure 10-6 Diel dissolved oxygen at USGS monitoring stations 10-14
Figure 10-7 Water temperature at USGS monitoring stations 10-15
Figure 10-8 Periphyton biomass at USGS monitoring stations 10-16
Figure 11-1 Absolute mean error for several model studies 11-13
Figure 11-2 Mean dissolved oxygen error for several model studies 11-14
Figure 11-3 Relative error in dissolved oxygen for several water quality models 11-15
Figure 11-4 Relative error in chlorophyll for several water quality models 11-15
Figure 11-5 Relative error in total nitrogen for several water quality models 11-16
Figure 11-6 Relative error in total phosphorus for several water quality models 11-16
vi
-------
TABLES
Table 1-1 Stream reaches on the Pennsylvania 303(d) list cited for nutrients and low DO 1-7
Table 1-2 Stream reaches on the Delaware 303(d) list cited for nutrients and low DO 1-7
Table 4-1 EFDC model water quality state variables 4-1
Table 5-1 EFDC sediment process model state variables and flux terms 5-1
Table 6-1 Flow statistics for stream gages in Christina River Basin 6-5
Table 6-2 File structure of the water quality monitoring database 6-6
Table 6-3 File structure of the NPDES point source discharge database 6-7
Table 6-4 Characteristic (default) NPDES effluent concentrations 6-8
Table 6-5 Characteristic NPDES effluent parameter ratios 6-8
Table 7-1 Estimated flow rates for subwatersheds in Christina River Basin 7-3
Table 7-2 Estimated nonpoint source concentrations for HSPF watersheds in Christina Basin . 7-4
Table 7-3 Wastewater treatment plant monitoring, August 1997 7-5
Table 7-4 Methodology for developing EFDC point source loads from DMR data 7-6
Table 7-5 EFDC water quality parameter concentrations for WWTPs, August 1997 study .... 7-7
Table 7-6 Locations of NPDES point source discharges included in the model 7-8
Table 7-7 Locations of consumptive use water withdrawals included in the model 7-11
Table 7-8 Atmospheric dry deposition rates used in Christina River Basin EFDC Model .... 7-12
Table 7-9 Atmospheric wet deposition rates used in Christina River Basin EFDC Model .... 7-12
Table 8-1 Specified boundary condition parameters in EFDC water quality model 8-2
Table 9-1 Hydraulic control structures in Christina River Basin EFDC model 9-2
Table 9-2 Harmonic analysis of tides at Port of Wilmington and Newport 9-4
Table 9-3 Model-data comparison of velocity, flow, and geometry (August 1997 data) 9-5
Table 9-4 Stream reaches included in EFDC Christina River Basin water quality model 9-7
Table 9-5 Comparison of model periphyton biomass with 1985 measurements 9-12
Table 9-6 Model-data comparison of sediment oxygen demand rates 9-14
Table 10-1 Harmonic analysis of tides at Port of Wilmington and Newport 10-2
Table 10-2 Model-data comparison of sediment oxygen demand rates 10-8
Table 11-1 Statistical summary of Christina River Model 1997 calibration results 11-4
Table 11-2 Statistical summary of Christina River Model 1995 validation results 11-4
Table 11-3 Relative error of total nitrogen and total phosphorus for 1995 and 1997 periods .... 11-4
Table 11-4 Summary of various models in comparison group 11-6
vii
-------
This page intentionally left blank
viii
-------
1 - INTRODUCTION
Section 303(d) of the Federal Clean Water Act requires the states to identify waterbodies that
need Total Maximum Daily Loads (TMDLs) to assure compliance with water quality standards.
TMDLs, as defined by U.S. Environmental Protection Agency (EPA), are the sum of individual waste
load allocations (WLAs) for point sources and load allocations (LAs) for nonpoint sources of pollution.
Not all waterbodies require the development of a TMDL. Both Pennsylvania and Delaware have
developed 303(d) lists of impaired waters in the Christina River Basin. This study is concerned only
with the 303(d) waters listed for nutrients and low dissolved oxygen.
A hydrodynamic and water quality model of the Christina River Basin (see Figure 1-1) has been
developed for use in TMDL calculations. The Environmental Fluid Dynamics Code (EFDC)
hydrodynamic and water quality model has been selected for use in this study. The model includes a
portion of the tidal Delaware River, and tidal Christina River, as well as other nontidal streams, including
the upper Christina River, Brandywine Creek, East Branch Brandywine Creek, West Branch Brandywine
Creek, Buck Run, Red Clay Creek, and White Clay Creek. The model consists of 406 depth-averaged,
computational grid cells. The grid cells in the nontidal streams have lengths ranging from about 500 to
1,000 meters to provide sufficient longitudinal resolution.
The United States Geological Survey (USGS) is currently developing a watershed runoff model
for the Christina River Basin using the Hydrologic Simulation Program-Fortran (HSPF) model. The
HSPF model is scheduled to be completed approximately 2 years hence. When it is available, dynamic
nonpoint source flows and nutrient loadings computed by the HSPF model will be coupled with the
EFDC receiving water model to form a powerful tool for management of both point and nonpoint sources
in the basin. This report discusses the first phase of the TMDL for the Christina River Basin, namely, the
low-flow analysis for nutrients and dissolved oxygen. Since the HSPF model is not yet available to
provide dynamic, time-varying nonpoint source flows and loads, this phase of the TMDL will consider
only low-flow summer conditions. The flow rates and nutrient loading rates from the point and nonpoint
sources in the EFDC model for this phase were time-varying inputs. The point source loads generally
varied on a monthly time step based on discharge monitoring records (DMRs). The nonpoint source
loads varied on a daily time step as estimated from daily flow records and constant background
concentrations. The tidal boundary conditions were also configured as time-varying concentrations, as
were the atmospheric meteorological conditions.
7 - Introduction
1-1
-------
1.1 Purpose of the Study
The purpose of this project is to construct a numerical model of the major features of estuarine
and nontidal circulation and eutrophication processes for the Christina River Basin system. This
numerical model is a tool for the development of TMDLs for nutrients, dissolved oxygen, and zinc. The
objectives of the water quality model are as follows:
• Develop a water quality model that can be used to support development of TMDLs for
nutrients and dissolved oxygen for the Brandywine Creek and Christina River basins.
• Configure the EFDC model so that it is consistent with the HSPF watershed runoff loading
model being developed concurrently by the USGS.
• Develop and present a one-day workshop to transfer the water quality model technology to the
U.S. Environmental Protection Agency (EPA), Delaware Department of Natural Resources
and Environmental Control (DNREC), Pennsylvania Department of Environmental Protection
(PADEP), Delaware River Basin Commission (DRBC), and other interested parties.
• Document the development and calibration of the water quality model in the form of a
summary report
• Deliver the model source and executable code along with a low-flow data set that can be used
for TMDL analyses.
1.2 Scope of the Study
The scope of the study has been expanded beyond the scope of work described in the original
proposal for this project to include (1) the use of a multidimensional, finite-difference hydrodynamic and
water quality model (EFDC) instead of the DYNHYD/WASP framework; (2) the use of a sediment
process model to predict the response of benthic sediment nutrient fluxes and sediment oxygen demand
due to changes in loading to the system; and (3) the calibration of the model over a dynamic summer
period beginning in May and continuing through September 1997. The stream reaches cited for nutrients
and low dissolved oxygen on the Pennsylvania 303(d) list are shown in Table 1-1; those cited on the
Delaware 303(d) list are given in Table 1-2. The stream reaches listed for nutrients and low dissolved
oxygen where the potential causes are point sources, nonpoint sources, or other reasons are shown in
Figure 1-2. The stream reaches on the 303(d) lists cited for nutrients and low dissolved oxygen having
potential causes related to point sources only are shown in Figure 1-3. The stream segments shown in
Figure 1-3 will be the primary focus of this initial phase of the TMDL for low-flow conditions.
Calibration of the model was to be achieved using available data, and no field sampling to
support model development was included in the scope of this study. An extensive data monitoring
program for the Christina River Basin managed by DNREC and PADEP has been in place since at least
1995. Under this program various stations have been sampled generally on a monthly or bimonthly basis.
Data for calibration were obtained from various sources including STORET, USGS, DNREC, PADEP,
NOAA, and a special field sampling study conducted by Dr. John Davis in August 1997. The resulting
1-2
1 - Introduction
-------
data set provided reasonably complete coverage for all the stream reaches included in the model during
the summer 1997 calibration period.
1.3 The EFDC Model Package
The hydrodynamic and water quality model chosen for the study is EFDC, the Environmental
Fluid Dynamics Code (Hamrick 1992a; Park et al. 1995) because it contained features not available in
DYNHYD/WASP that were necessary for simulating the flow and eutrophication processes in the
Christina River Basin. A number of low-head dams and submerged weirs have been built in the streams
in the study area. DYNHYD does not have any means of handling flow through a hydraulic structure
whereas EFDC has that capability. The EFDC water quality module also contains a sediment process
submodel that is useful for determining the changes in sediment oxygen and nutrient flux rates due to
changes in external loadings to the system. The WASP model does not have this advanced capability.
The EFDC model was developed to comply with the requirements for the eutrophication model study of
the Christina River Basin. The EFDC hydrodynamic model produces three-dimensional (3-D)
predictions of velocity, diffusion, surface elevation, salinity, suspended sediment, and temperature on an
intratidal time scale (60-second time step). The water quality module, an adaptation of the Corps of
Engineers CE-QUAL-ICM (Cerco and Cole 1993), was integrated directly into EFDC and operates on a
time step which is double that of the hydrodynamic model. In addition, the water quality model
optionally interacts directly with a predictive sediment diagenesis model based on DiToro and Fitzpatrick
(1993). Point source and tributary (nonpoint source) loads were developed using water quality and flow
monitoring data.
7 - Introduction
1-3
-------
PA
MD
peake & Delaware Canal
Che'
76 00
75 45
75 30
75 15
Figure 1-1. Christina River Basin study area.
7-4
7 - Introduction
-------
/X
Bra
Run
PA
MD
Figure 1-2. 303(d) waters listed for nutrients, low DO (NPS and PS as cause)
/ - Introduction 1-5
-------
/X
Bra
Run
PA
MD
Figure 1-3. 303(d) waters listed for nutrients, low DO (PS as possible cause)
1-6 7 - Introduction
-------
Table 1-1. Stream reaches on the Pennsylvania 303(d) list cited for nutrients and low DO.
Watershed
Stream ID
Segment ID
Miles
Source of Impairment
Cause of Impairment
Brandywine Creek
00004
27
1.28
other
nutrients
Buck Run
00131
50
1.77
municipal point source
nutrients, low DO
Sucker Run
00202
970930-1437-GLW
6.78
agriculture
nutrients
W.Br. Brandywine Cr.
00085
various
16.05
agriculture
nutrients
Broad Run
00434
971209-1445-ACW
4.10
hydromodification,
agriculture
low DO,
nutrients
E.Br. Red Clay Creek
00413
various
11.62
agriculture
low DO
E.Br. White Clay Creek
00432
various
29.55
agriculture
nutrients, low DO
Egypt Run
00440
970508-1245-ACE
3.66
agriculture
low DO
Indian Run
00475
115
1.09
agriculture,
municipal point source
nutrients
Middle Br. White Clay
00462
115
17.33
agriculture,
municipal point source
nutrients
Red Clay Creek
00374
971203-1400-ACW
0.76
agriculture
low DO
Trout Run
00402
970506-1425-MRB
2.74
agriculture
nutrients
Walnut Run
00435
971209-1445-ACW
1.39
agriculture,
hydromodification
nutrients,
low DO
W.Br. Red Clay Creek
00391
971023-1145-MRB
4.58
agriculture
low DO
White Clay Creek
00373
971216-1230-GLW
1.13
agriculture
nutrients
Table 1-2. Stream reaches on the Delaware 303(d) list cited for nutrients and low DO.
Waterbody ID
Watershed Name
Segment
Miles
Pollutants/Stressors
Probable Sources
DE040-001
Brandywine Creek
Lower Brandywine
3.8
nutrients
PS, NPS, SF
DE040-002
Brandywine Creek
Upper Brandywine
9.3
nutrients
PS, NPS, SF
DE260-001
Red Clay Creek
Main stem
12.8
nutrients
PS, NPS, SF
DE260-002
Red Clay Creek
Burroughs Run
4.5
nutrients
NPS
DE320-001
White Clay Creek
Main stem
18.2
nutrients
PS, NPS
DE320-002
White Clay Creek
Mill Creek
16.6
nutrients
NPS
DE320-003
White Clay Creek
Pike Creek
9.4
nutrients
NPS
DE320-004
White Clay Creek
Muddy Run
5.8
nutrients
NPS
DEI 20-001
Christina River
Lower Christina River
1.5
nutrients, DO
NPS, SF
DEI 20-002
Christina River
Middle Christina River
7.5
nutrients
NPS, SF
DE120-003
Christina River
Upper Christina River
6.3
nutrients
NPS, SF
DE120-003-02
Christina River
Lower Christina Creek
8.4
nutrients
NPS
DE120-005-01
Christina River
West Branch
5.3
nutrients
NPS
DEI 20-006
Christina River
Upper Christina Creek
8.3
nutrients
NPS
DEI 20-007-01
Christina River
Little Mill Creek
12.8
nutrients, DO
NPS, SF
PS = point source; NPS = nonpoint source; SF = Superfund site
7 - Introduction
1-7
-------
This page intentionally left blank
1-8
1 - Introduction
-------
2 - THE CHRISTINA RIVER BASIN SYSTEM
2.1 Physical Description
The Christina River Basin covers an area of about 565 square miles in Chester County,
Pennsylvania; New Castle County, Delaware; and a small portion of Cecil County, Maryland
(Figure 1-1). The basin drains to the tidal Delaware River at Wilmington, Delaware. The major streams
in the watershed include the upper Christina River, Brandywine Creek, Red Clay Creek, and White Clay
Creek. The watershed is composed of diverse land uses and includes urban, rural, and agricultural areas.
The streams in the basin are used as municipal and industrial water supplies as well as for recreational
purposes.
2.2 Hydrology
The Christina River basin was delineated into 39 subwatersheds for the HSPF model. There are a
number of long-term stream gages in the basin maintained by the USGS. Data from these gaging stations
were used to determine the 7-day, 10-year (7Q10) low-flow discharge rates (see Section 6.5). These
same subwatersheds were used to determine daily nonpoint source inflows and loads to the EFDC model.
For model calibration, a daily-average flow rate per unit area (cfs/mi2) was estimated for each of the 39
basins and was then distributed to the appropriate EFDC model grid cell. The flow balance was checked
by comparing model results to the daily flow rates at selected USGS stream gage stations for the period
May 1 to September 21, 1997 (see Section 9.4 for details).
2.3 Eutrophication Processes
Algae in the water column eventually become deposited as organic matter and decay in the
bottom sediments, which contributes to oxygen demand. Nutrients in the estuary are taken up by algae
and predation as well as algal mortality which results in the transfer of nutrients to the benthic sediments.
In the summer, with increased temperature, the nutrients are mineralized in the sediments and released
back into the water column. Nutrients released from the sediments support the summer algal bloom.
Carbon produced by algae settles to bottom waters, decays, and consumes oxygen. Diminished oxygen
in bottom water enhances the release of sediment nutrients, especially ammonia. The nutrient release
continues the cycle of benthic release, algal production, and oxygen consumption. It is this cycle that is
simulated by the EFDC predictive sediment processes submodel.
The three major nutrients required by algae for growth are carbon, nitrogen, and phosphorus.
Diatoms also require silica from which they synthesize their distinctive skeletons. Algal production is
diminished or eliminated by the prolonged absence of one or more of the required nutrients. Nutrients
are supplied in various ratios from natural and anthropogenic sources. The ratio of nutrient utilization by
2 - The Christina River Basin System
2-1
-------
algae is within a limited range, however, largely determined by algal composition. The classic Redfield
ratios (Redfield et al. 1963) indicate that the ratio of carbon to nitrogen required by algae is 6 to 1 by
mass. The required ratio of carbon to phosphorus is 42 to 1, and the carbon to silica requirement is about
1.25 to 1 (Strickland 1960). The disparity in the ratio of nutrients supplied and nutrients required often
leads to depletion of one nutrient, due to algal uptake, while the others remain available. The depleted
nutrient is referred to as "limiting", since algal production is limited by the supply of this nutrient.
Inorganic carbon is seldom in short supply and is usually not considered in analyses of nutrient
limitations. Silica receives little emphasis in management studies since the supply from natural sources
is beyond control and usually abundant. The primary emphasis is placed on limitations by nitrogen and
phosphorus since the supply of these nutrients can be altered through the management of releases from
municipalities, industry, agriculture, and other sources.
A "rule of thumb" is that phosphorus is the limiting nutrient in freshwater systems (Hecky and
Kilham 1988) whereas nitrogen is limiting in estuarine and marine waters (Boynton et al. 1982). The
phosphorus limit in freshwater is influenced by the relative natural abundance of the two nutrients.
In downstream portions of estuaries and in coastal waters, the ratio of nitrogen to phosphorus is
altered from the ratio in runoff by internal recycling processes. Sediment-water interactions greatly
diminish the availability of nitrogen relative to phosphorus (Nixon 1981). Particulate organic nitrogen
and phosphorus enter the sediments roughly in Redfield proportions as organic matter. Within the
sediments, total phosphorus is conservative. The only pathways for removal are recycling of inorganic
phosphorus back to the water column or burial to deep, isolated sediments. On the other hand, total
nitrogen is a nonconservative parameter. A significant fraction may be lost through denitrification to
nitrogen gas. The nitrogen loss is such that the nitrogen returned to the water column is roughly half the
amount expected based on the nitrogen to phosphorus ratio of the incoming material. The reduced
nitrogen to phosphorus ratio of dissolved fluxes leaving the sediments, compared to particle fluxes
entering the sediments, acting over lengthy time scales, pushes the water column toward nitrogen rather
than phosphorus limitation.
2.4 Sediment-Water Interactions
Over time scales of years, benthic sediments act as sinks for oxygen, nitrogen, phosphorus, and
silica removed from the water column. Oxygen is consumed by the oxygenation of organic carbon and
by the nitrification of ammonia. Certain fractions of particulate nitrogen, phosphorus, and silica that
settle into bottom sediments are buried to deeper sediment layers from which recycling to the water
column is not possible.
2-2
2 - The Christina River Basin System
-------
Over seasonal time scales, sediments can be significant sources of dissolved nutrients to the
overlying water. The role of sediments in the systemwide nutrient budget is especially important during
the summer when seasonal low flows diminish tributary nutrient loads. During the summer, warm
temperature enhances biological processes in the sediments. The decay (i.e., diagenesis) of organic
matter produces phosphate, ammonia, and silica that are released into the overlying water.
A sediment process of potential importance to the management of eutrophication issues is the
coupled nitrification/denitrification sequence that occurs in bottom sediments. The nitrification reaction,
in which ammonia is oxidized to nitrate, requires oxygen. The denitrification reaction, in which nitrate is
reduced to nitrogen gas, normally takes place under anoxic conditions. Denitrification is therefore a
potential pathway for removal of nitrogen from the system. The primary source of nitrate for
denitrification is previously nitrified ammonia. The maximum denitrification occurs when oxygen is
available for nitrification. When oxygen is absent in the sediments, denitrification is diminished and
limited to the rate at which nitrate is supplied by diffusion from the water column.
When oxygen is freely available, a large fraction of ammonia produced in the sediments is
nitrified/denitrified to nitrogen gas, which is unavailable to algae. When oxygen is absent, virtually all
the ammonia produced is released to the overlying water and is available for algal consumption. The
coupling of nitrification and denitrification suggests the existence of an antagonistic cycle in the
eutrophication process. Conditions that lead to hypoxia diminish denitrification leading to increased
ammonia released to the water column that feeds algal production. Algal carbon settles to the bottom,
consumes oxygen, and further diminishes denitrification. This same mechanism suggests that a slight
improvement in bottom dissolved oxygen can start a positive feedback reaction which would promote
denitrification at the expense of ammonia release. The diminished ammonia release would limit algal
production and reduce the supply of carbon to the bottom sediments. The diminished carbonaceous
oxygen demand leads to increased dissolved oxygen, which leads to still more denitrification.
2 - The Christina River Basin System
2-3
-------
This page intentionally left blank
2-4 2 - The Christina River Basin System
-------
3 - EFDC HYDRODYNAMIC MODEL
Modeling the physics, chemistry, and biology of the receiving waters of streams, lakes, estuaries,
or coastal regions requires a model that incorporates all the major processes. Transport processes for this
study were simulated using the three-dimensional EFDC hydrodynamic model that includes temperature
transport. The EFDC hydrodynamic model was developed by Hamrick (1992a). The model formulation
was based on the principles expressed by the equations of motion, conservation of volume, and
conservation of mass. Quantities computed by the model included three-dimensional velocities, surface
elevation, vertical viscosity and diffusivity, temperature, salinity, and density.
3.1 General
The Environmental Fluid Dynamics Code is a general purpose modeling package for simulating
three-dimensional flow, transport, and biogeochemical processes in surface water systems including
rivers, lakes, estuaries, reservoirs, wetlands, and coastal regions. The EFDC model was originally
developed at the Virginia Institute of Marine Science for estuarine and coastal applications and is
considered public domain software. In addition to hydrodynamic and salinity and temperature transport
simulation capabilities, EFDC is capable of simulating cohesive and noncohesive sediment transport,
near field and far field discharge dilution from multiple sources, eutrophication processes, the transport
and fate of toxic contaminants in the water and sediment phases, and the transport and fate of various life
stages of finfish and shellfish. Special enhancements to the hydrodynamic portion of the code, including
vegetation resistance, drying and wetting, hydraulic structure representation, wave-current boundary
layer interaction, and wave-induced currents, allow refined modeling of wetland marsh systems,
controlled flow systems, and nearshore wave induced currents and sediment transport. The EFDC model
has been extensively tested and documented for more than 20 modeling studies. The model is presently
being used by a number of organizations including universities, governmental agencies, and
environmental consulting firms.
The structure of the EFDC model includes four major modules: (1) a hydrodynamic model, (2) a
water quality model, (3) a sediment transport model, and (4) a toxics model (see Figure 3-1). The EFDC
hydrodynamic model itself, which was used for this study, is composed of six transport modules
including dynamics, dye, temperature, salinity, near field plume, and drifter (see Figure 3-2). Various
products of the dynamics module (i.e., water depth, velocity, and mixing) are directly coupled to the
water quality, sediment transport, and toxics models as shown in the following figures. Schematic
diagrams for the water quality model, the sediment transport model, and the toxics model are shown in
Figures 3-3, 3-4, and 3-5, respectively.
3 - EFDC Hydrodynamic Model
3-1
-------
EFDC Mode
Hydrodynamics
Water
Quality
Sediment
Transport
Toxics
Figure 3-1. Primary modules of the EFDC model.
Hydrodynamics
'
Dynamics
(E, u, v, w, mixing)
Dye Temperature
„ .. .. Near Field 1
Salinity n.
3 Plume
l
Drifter
Figure 3-2. Structure of the EFDC hydrodynamic model.
Hydrodynamic
Model
DO
FCB
COD
TAM
Diatoms
Greens
Other
Algae
Organic
Carbon
Predicted Flux
Specified Flux
Nitrogen
Sediment
Diagenesis
Phosphorus
Water Quality
Figure 3-3. Structure of the EFDC water quality model.
3-2
3 - EFDC Hydrodynamic Model
-------
Hydrodynamic
Model
Water Column
Sediment Bed
Cohesive
Cohesive
Noncohesive
Noncohesive
Sediment Transport
Model
Figure 3-4. Structure of the EFDC sediment transport model.
Cohesive
Sediment
Hydrodynamic
Model
Sediment Transport
Model
Toxic Model
Figure 3-5. Structure of the EFDC toxics model.
3 - EFDC Hydrodynamic Model
3-3
-------
3.2 Hydrodynamics and Salinity and Temperature Transport
The physics of the EFDC model and many aspects of the computational scheme are equivalent to
the widely used Blumberg-Mellor model (Blumberg and Mellor 1987). The EFDC model solves the
three-dimensional, vertically hydrostatic, free surface, turbulent averaged equations of motions for a
variable density fluid. Dynamically coupled transport equations for turbulent kinetic energy, turbulent
length scale, salinity, and temperature are also solved. The two turbulence parameter transport equations
implement the Mellor-Yamada level 2.5 turbulence closure scheme (Mellor and Yamada 1982; Galperin
et al. 1988). The EFDC model uses a stretched or sigma vertical coordinate and Cartesian, or curvilinear,
orthogonal horizontal coordinates.
The numerical scheme employed in EFDC to solve the equations of motion uses second order
accurate spatial finite differencing on a staggered or C grid. The model's time integration employs a
second order accurate three-time level, finite difference scheme with an internal-external mode splitting
procedure to separate the internal shear or baroclinic mode from the external free surface gravity wave or
barotropic mode. The external mode solution is semi-implicit and simultaneously computes the
two-dimensional (2-D) surface elevation field by a preconditioned conjugate gradient procedure. The
external solution is completed by the calculation of the depth average barotropic velocities using the new
surface elevation field. The model's semi-implicit external solution allows large time steps that are
constrained only by the stability criteria of the explicit central difference or high order upwind advection
scheme (Smolarkiewicz and Margolin 1993) used for the nonlinear accelerations. Horizontal boundary
conditions for the external mode solution include options for simultaneously specifying the surface
elevation only, the characteristic of an incoming wave (Bennett and Mcintosh 1982), free radiation of an
outgoing wave (Bennett 1976; Blumberg and Kantha 1985), or the normal volumetric flux on arbitrary
portions of the boundary. The EFDC model's internal momentum equation solution, at the same time
step as the external solution, is implicit with respect to vertical diffusion. The internal solution of the
momentum equations is in terms of the vertical profile of shear stress and velocity shear, which results in
the simplest and most accurate form of the baroclinic pressure gradients and eliminates the over
determined character of alternate internal mode formulations. Time splitting inherent in the three-time-
level scheme is controlled by periodic insertion of a second order accurate two-time-level trapezoidal
step. EFDC is also readily configured as a 2-D model in either the horizontal or vertical planes.
The EFDC model implements a second order accurate in space and time, mass conservation
fractional step solution scheme for the Eulerian transport equations for salinity, temperature, suspended
sediment, water quality constituents, and toxic contaminants. The transport equations are temporally
integrated at the same time step or twice the time step of the momentum equation solution
(Smolarkiewicz and Margolin 1993). The advective step of the transport solution uses either the central
3-4
3 - EFDC Hydrodynamic Model
-------
difference scheme used in the Blumberg-Mellor model or a hierarchy of positive definite upwind
difference schemes. The highest accuracy upwind scheme, second order accurate in space and time, is
based on a flux-corrected transport version Smolarkiewicz's multidimensional positive-definite advection
transport algorithm (Smolarkiewicz and Clark, 1986; Smolarkiewicz and Grabowski 1990), which is
monotonic and minimizes numerical diffusion. The horizontal diffusion step, if required, is explicit in
time, whereas the vertical diffusion step is implicit. Horizontal boundary conditions include time variable
material inflow concentrations, upwind outflow, and a damping relaxation specification of climatological
boundary concentration. The NOAA Geophysical Fluid Dynamics Laboratory's atmospheric heat
exchange model (Rosati and Miyakoda 1988) is implemented for the temperature transport equation.
3.3 Sediment Transport
The EFDC code is capable of simulating the transport and fate of multiple size classes of
cohesive and noncohesive suspended sediment including bed deposition and resuspension. Water
column transport is based on the same high order advection-diffiision scheme used for salinity and
temperature. A number of options are included for the specification of settling velocities. For the
transport of multiple size classes of cohesive sediment, an optional flocculation model (Burban et al.
1989, 1990) can be activated. Sediment mass conservative deposited bed formulations are included for
both cohesive and noncohesive sediment. The deposited bed may be represented by a single layer or
multiple layers. The multiple bed layer option provides a time since deposition versus vertical position in
the bed relationship to be established. Water column/sediment bed interface elevation changes can be
optionally incorporated into the hydrodynamic continuity equation. An optional one-dimensional (1-D) in
the vertical, bed consolidation calculation can be performed for cohesive beds.
3.4 Water Quality and Eutrophication Simulation
The EFDC code includes two internal eutrophication submodels for water quality simulation
(Park et al. 1995). The simple or reduced eutrophication model is functionally equivalent to the WASP5
EUTRO model (Ambrose et al. 1993). The complex or full eutrophication model is functionally
equivalent to the CE-QUAL-ICM or Chesapeake Bay Water Quality model (Cerco and Cole 1993). Both
water column eutrophication models are coupled to a functionally equivalent implementation of the
CE-QUAL-ICM sediment diagenesis or biogeochemical processes model (DiToro and Fitzpatrick 1993).
The eutrophication models can be executed simultaneously with the hydrodynamic component of EFDC,
or EFDC simulated hydrodynamic transport fields can be saved, allowing the EFDC code to be executed
in a water quality only simulation mode.
The computational scheme used in the internal eutrophication models employs a fractional step
extension of the same advective and diffusive algorithms used for salinity and temperature, which
3 - EFDC Hydrodynamic Model
3-5
-------
guarantee positive constituent concentrations. A novel ordering of the reaction sequence in the reactive
source and sink fractional step allows the linearized reactions to be solved implicitly, further
guaranteeing positive concentrations. The eutrophication models accept an arbitrary number of point and
nonpoint source loadings as well as atmospheric and ground water loadings.
In addition to the internal eutrophication models, the EFDC model can be externally linked to the
WASP5 model. In the external linking mode, the EFDC model generates WASP5 input files describing
cell geometry and connectivity as well as advective and diffusive transport fields. For estuary simulation,
the transport fields may be intratidally time averaged or intertidally time averaged using the averaging
procedure described by Hamrick (1994).
3.5 Toxic Contaminant Transport and Fate
The EFDC code includes two internal submodels for simulating the transport and fate of toxic
contaminants. A simple, single contaminant submodel can be activated from the master input file. The
simple model accounts for water and suspended sediment phase transport with equilibrium partitioning
and a lumped first order reaction. Contaminant mass per unit area in the sediment bed is also simulated.
The second, more complex, submodel simulates the transport and fate of an arbitrary number of reacting
contaminants in the water and sediment phases of both the water column and sediment bed. In this mode,
the contaminant transport and fate simulation is functionally similar to the WASP5 TOXIC model
(Ambrose et al. 1993), with the added flexibility of simulating an arbitrary number of contaminants, and
the improved accuracy of utilizing more complex three-dimensional physical transport fields in a highly
accurate numerical transport scheme. Water-sediment phases interaction may be represented by
equilibrium or nonlinear sorption processes. In this mode, the multilayer sediment bed formulation is
active, with sediment bed water volume and dissolved contaminant mass balances activated to allow
contaminants to reenter the water column by sediment resuspension, pore water expulsion due to
consolidation, and diffusion from the pore water into the water column. The complex contaminant model
activates a subroutine describing reaction processes with appropriate reaction parameters provided by the
toxic reaction processes input file.
3.6 Finfish and Shellfish Transport
The EFDC code includes the capability of simulating the transport and fate of various life stages
of finfish and shellfish. In addition to advection and diffusion by the ambient flow, mortality, predation,
toxicity, and swimming behavior are simulated. Organism age and ambient environment queued vertical
and horizontal swimming and settling is simulated. Environmental queues include light intensity,
temperature, salinity, and tidal phases.
3-6
3 - EFDC Hydrodynamic Model
-------
3.7 Near-Field Discharge Dilution and Mixing Zone Analysis
In addition to the far-field transport and fate simulation capability incorporated into the EFDC
code's water quality and toxic contaminant modules, the code includes a near-field discharge dilution and
mixing zone module. The near field model is based on a Lagrangian buoyant jet and plume model (Frick
1984; Lee and Cheung 1990) and allows representation of submerged single and multiple port diffusers
and buoyant surface jets. The near field model provides analysis capabilities similar to CORMIX (Jirka
and Doneker 1991; Jirka and Akar 1991) while offering two distinct advantages. The first advantage is
that a more realistic representation of ambient current and stratification conditions, provided directly by
the EFDC hydrodynamic module, is incorporated into the analysis. The second advantage is that multiple
discharges and multiple near field analysis times may be specified to account for varying ambient current
and stratification conditions. For example, the analysis of 10 discharges under six ambient conditions
each would require 60 executions of CORMIX, while the entire analysis of the 60 situations would be
produced in a single EFDC simulation. The near-field simulation may be executed in two modes. The
first provides virtual source information for representing the discharges in a standard EFDC far field
transport and fate simulation. In the second mode the near-field and far-field transport are directly
coupled, using a virtual source formulation, to provide simultaneous near and far field transport and fate
simulation.
3.8 Spill Trajectory and Search and Rescue Simulation
In addition to the Eulerian transport equation formulation used for far field analysis and the
Lagrangian jet and plume module used for near field analysis, the EFDC code incorporates a number of
Lagrangian particle transport formulations based on an implicit trilinear interpolation scheme (Bennett
and Clites 1987). The first formulation allows release of neutrally buoyant or buoyant drifters at user
specified locations and times. This formulation is useful in simulating spill trajectories, search and rescue
operations, and oceanographic instrument drifters. The second formulation releases drifters in each
three-dimensional model cell at a specified sequence of times and calculates the generalized Lagrangian
mean velocity field (Andrews and Mclntyre 1978) relative to a user-specified averaging interval.
3.9 Wetland, Marsh, and Tidal Flat Simulation Extension
The EFDC model provides a number of enhancements for the simulation of flow and transport in
wetlands, marshes, and tidal flats. The code allows for drying and wetting in shallow areas by a mass
conservative scheme. The drying and wetting formulation is coupled to the mass transport equations in a
manner that prevents negative concentrations of dissolved and suspended materials. A number of
alternatives are in place in the model to simulate general discharge control structures such as weirs,
spillways, culverts, and water surface elevation activated pumps. The effect of submerged and emergent
plants is incorporated into the turbulence closure model and flow resistance formulation. Plant density
3 - EFDC Hydrodynamic Model
3-7
-------
and geometric characteristics of individual and composite plants are required as input for the vegetation
resistance formulation. A simple soil moisture model, allowing rainfall infiltration and soil water loss
due to evapotranspiration under dry conditions, is implemented. To represent narrow channels and
canals in wetland, marsh and tidal flat systems, a subgrid scale channel model is implemented. The
subgrid channel model allows a 1-D network in the horizontal channels to be dynamically coupled to the
two-dimensional horizontal grid representing the wetland, marsh, or tidal flat system. Volume and mass
exchanges between 2-D wetland cells and the 1-D channels are accounted for. The channels may
continue to flow when the 2-D wetland cells become dry.
3.10 Nearshore Wave-Induced Currents and Sediment Transport Extensions
The EFDC code includes a number of extensions for simulation of nearshore wave-induced
currents and noncohesive sediment transport. The extensions include a wave-current boundary layer
formulation similar to that of Grant and Madsen (1986); modifications of the hydrodynamic model's
momentum equations to represent wave period averaged Eulerian mean quantities; the inclusion of the
three-dimensional wave-induced radiation or Reynold's stresses in the momentum equations; and
modifications of the velocity fields in the transport equations to include advective transport by the wave-
induced Stake's drift. High frequency surface wave fields are provide by an external wave
refraction-diffraction model or by an internal mild slope equation submodel similar to that of Madsen and
Larsen (1987). The internal refraction-diffraction computation is executed on a refined horizontal grid
coincident with the main model's horizontal grid.
3.11 User Interface
The EFDC modeling package's user interface is based on text input file templates. This choice
was selected in the interest of maintaining model portability across a range of computing platforms and
readily allows the model user to modify input files using most text editing software. The text interface
also allows modification of model files on remote computing systems and in hetrogeneous network
environments. All input files have standard templates available with the EFDC code and in the digital
version of the user's manual. The file templates include extensive built-in documentation and an
explanation of numerical input data quantities. Actual numerical input data are inserted into the text
template in a flexible free format as internally specified in the file templates. Extensive checking of
input files is implemented in the code and diagnostic on screen messages indicate the location and nature
of input file errors. All input files involving dimensional data have unit conversion specifications for the
Meters-Kilograms-Seconds (MKS) international system of units used internally in the model.
3-8
3 - EFDC Hydrodynamic Model
-------
3.12 Preprocessing Software
The EFDC modeling package includes a grid generating preprocessor code, GEFDC, which is
used to construct the horizontal model grid, and interpolate bathymetry and initial fields such as water
surface elevation, salinity, to the grid cells. EFDC inputs files specifying the grid geometry and initial
fields are generated by the preprocessor. The preprocessor is capable of generating Cartesian and
curvilinear-orthogonal grids using a number of grid generation schemes (Mobley and Stewart 1980;
Ryskin and Leal 1983; Kang and Leal 1992).
3.13 Program Configuration
The EFDC code exists in only one generic version. A model application is specified entirely by
information in the input files. To minimize memory requirements for specific applications, an executable
file is created by adjusting the appropriate variable array size in the model's parameter file and compiling
the source code. The EFDC model can be configured to execute all or a portion of a model application in
reduced spatial dimension mode including 2-D depth or width averaged and 1-D cross section averaged.
The number of layers used in the 3-D mode or 2-D width averaged mode is readily changed by one line
of model input. Model grid sections specified as 2-D width averaged are allowed to have depth varying
widths to provide representations equivalent to those of 2-D width averaged estuarine and reservoir
models such as CE-QUAL-W2 (Cole and Buchak 1994).
3.14 Run-Time Diagnostics
The EFDC modeling package includes extensive built-in run-time diagnostics that may be
activated in the master input file by the model user. Representative diagnostics include records of
maximum CFL numbers, times and locations of negative depths, a variety of volume and mass balance
checks, and global mass and energy balances. An on screen print of model variables in a specified cell
can be activated during modeling execution. The model generates a number of log files that allow
additional diagnostics of any run-time problems encountered during the set-up of a new application.
3.15 Model Output Options
A wide variety of output options are available for the EFDC model, including (1) specification of
output files for horizontal plane and vertical plane transect plotting of vector and scalar field at a
specified time; (2) the generation of time series of model variables at selected locations and time
intervals; (3) grab sample simulation at specified times and locations; and (4) the specification of least
squares analysis of selected model variables at a defined location over a specified interval. A general
three-dimensional output option allows saving of all major model variables in a compressed-file format at
specified times. A restart file is generated at user-specified intervals during model execution.
3 - EFDC Hydrodynamic Model
3-9
-------
3.16 Postprocessing, Graphics, and Visualization
The generic model output files can be readily processed by a number of third party graphics and
visualization software packages, often without the need for intermediate processing (Rennie and Hamrick
1992). The availability of the source code to the user allows the code to be modified for specific output
options. Graphics and visualization software successfully used with EFDC output include: APE, AVS,
IDL, Mathematica, MatLab, NCAR Graphics, PV-Wave, Techplot, SiteView, Spyglass Transform and
Slicer, Voxelview, and GrADS. The model developer currently uses Spyglass and Voxelview and a
number of postprocessor applications are available for special image enhancement for these products.
3.17 Documentation
Extensive documentation of the EFDC model is available. Theoretical and computational
aspects of the model are described by Hamrick (1992a). The model user's manual (Hamrick 1996)
provides details on use of the GEFDC preprocessor and set-up of the EFDC input files. Input file
templates are also included. A number of papers describe model applications and capabilities (Hamrick
1992b; Hamrick 1994; Moustafa and Hamrick 1994; Hamrick and Wu 1996; and Wu et al. 1996).
3.18 Computer Requirements
The EFDC modeling system is written in FORTRAN 77. The few nonstandard VAX FORTRAN
language extensions in the code are supported by a wide variety of ANSI standard FORTRAN 77
compilers. The generic or universal source code has been compiled and executed on most UNIX
workstations (DEC Alpha, Hewlett-Packard, IBM RISC6000, Silicon Graphics, Sun and Sparc
compatibles) Cray and Convex supercomputers, and PC compatible and Macintosh personal computers.
Absoft, Lahey, and Microsoft compilers are supported on PC compatibles, while Absoft, Language
Systems, and Motorola compilers are supported on Macintosh and compatible systems.
3-10
3 - EFDC Hydrodynamic Model
-------
4 - EFDC WATER QUALITY MODEL
4.1 Introduction
The central issues in the water quality model are primary production of carbon by algae and
concentration of dissolved oxygen. Primary production provides the energy required by the ecosystem to
function. However, excessive primary production is detrimental since its decomposition in the water and
sediments consumes oxygen. Dissolved oxygen is necessary to support the life functions of higher
organisms and is considered an indicator of the health of estuarine systems. To predict primary
production and dissolved oxygen, a large suite of model state variables is necessary (Table 4-1). The
nitrate state variable in the model represents the sum of nitrate and nitrite nitrogen. The three variables
(salinity, water temperature, and total suspended solids) needed for computation of the above 21 state
variables are provided by the EFDC hydrodynamic model. The interactions among the state variables is
illustrated in Figure 4-1. The kinetic processes included in the EFDC water quality model are mostly
from the Chesapeake Bay three-dimensional water quality model, CE-QUAL-ICM (Cerco and Cole
1994). The kinetic sources and sinks, as well as the external loads for each state variable, are described
in Sections 4.3 to 4.11. The kinetic processes include the exchange of fluxes at the sediment-water
interface, including sediment oxygen demand, which are explained in Section 5 (EFDC Sediment Process
Model) of this report. The description of the EFDC water column water quality model in this section is
from Park et al. (1995).
Table 4-1. EFDC model water quality state variables.
(1) cyanobacteria
(12) labile particulate organic nitrogen
(2) diatom algae
(13) dissolved organic nitrogen
(3) green algae
(14) ammonia nitrogen
(4) refractory particulate organic carbon
(15) nitrate nitrogen
(5) labile particulate organic carbon
(16) particulate biogenic silica
(6) dissolved organic carbon
(17) dissolved available silica
(7) refractory particulate organic phosphorus
(18) chemical oxygen demand
(8) labile particulate organic phosphorus
(19) dissolved oxygen
(9) dissolved organic phosphorus
(20) total active metal
(10) total phosphate
(21) fecal coliform bacteria
(11) refractory particulate organic nitrogen
(22) macroalgae
4 - EFDC Water Quality Model
4-1
-------
i P04d
SAd
p04p
iSAp
photosynthesis
TSS'
light
reaeration
respiration
* TSS from hydrodynamic
model
FCB
— LPOP
LPOC
LPON
TAM
RPOC
RPON
RPOP
COD
N023
DOC
DO
DOP
DON
P04t
NH4
Bm
Figure 4-1. Schematic diagram for the EFDC water column water quality model.
4.1.1 Algae
Algae are grouped into four model classes: cyanobacteria, diatoms, greens, and macroalgae. The
grouping is based upon the distinctive characteristics of each class and upon the significant role the
characteristics play in the ecosystem. Cyanobacteria, commonly called blue-green algae, are
characterized by their abundance (as picoplankton) in saline water and by their bloom-forming char-
acteristics in fresh water. Cyanobacteria are unique in that some species fix atmospheric nitrogen,
although nitrogen fixers are not believed to be predominant in many river systems. Diatoms are distin-
guished by their requirement of silica as a nutrient to form cell walls. Diatoms are large algae
characterized by high settling velocities. Settling of spring diatom blooms to the sediments may be a
significant source of carbon for sediment oxygen demand. Algae that do not fall into the preceding two
groups are lumped into the heading of green algae. Green algae settle at a rate intermediate between
4-2
4 - EFDC Water Quality Model
-------
cyanobacteria and diatoms and are subject to greater grazing pressure than cyanobacteria. Macroalgae
are almost always attached to a stable substrate and are therefore most abundant in the areas of harbors
and near shore. The waters in many stream systems are characterized by various rooted macrophytes and
periphyton. All species of macroalgae in this study have been lumped into a single class of macroalgae.
Because of their attachment to the substrate, they are limited to growing in the bottom water-column
layer and are not subject to physical transport.
4.1.2 Organic Carbon
Three organic carbon state variables are considered: dissolved, labile particulate, and refractory
particulate. Labile and refractory distinctions are based upon the time scale of decomposition. Labile
organic carbon decomposes on a time scale of days to weeks whereas refractory organic carbon requires
more time. Labile organic carbon decomposes rapidly in the water column or the sediments. Refractory
organic carbon decomposes slowly, primarily in the sediments, and may contribute to sediment oxygen
demand years after deposition.
4.1.3 Nitrogen
Nitrogen is first divided into organic and mineral fractions. Organic nitrogen state variables are
dissolved organic nitrogen, labile particulate organic nitrogen, and refractory particulate organic
nitrogen. Two mineral nitrogen forms are considered: ammonium and nitrate. Both are utilized to satisfy
algal nutrient requirements, although ammonium is preferred from thermodynamic considerations. The
primary reason for distinguishing the two is that ammonium is oxidized by nitrifying bacteria into nitrate.
This oxidation can be a significant sink of oxygen in the water column and sediments. An intermediate in
the complete oxidation of ammonium, nitrite, also exists. Nitrite concentrations are usually much less
than nitrate, and for modeling purposes, nitrite is combined with nitrate. Hence the nitrate state variable
actually represents the sum of nitrate plus nitrite.
4.1.4 Phosphorus
As with carbon and nitrogen, organic phosphorus is considered in three states: dissolved, labile
particulate, and refractory particulate. Only a single mineral form, total phosphate, is considered. Total
phosphate exists as several states within the model ecosystem: dissolved phosphate, phosphate sorbed to
inorganic solids, and phosphate incorporated in algal cells. Equilibrium partition coefficients are used to
distribute the total among the three states.
4.1.5 Silica
Silica is divided into two state variables: available silica and particulate biogenic silica.
Available silica is primarily dissolved and can be utilized by diatoms. Particulate biogenic silica cannot
4 - EFDC Water Quality Model
4-3
-------
be utilized. In the model, particulate biogenic silica is produced through diatom mortality. Particulate
biogenic silica undergoes dissolution to available silica or else settles to the bottom sediments.
4.1.6 Chemical Oxygen Demand
In the context of this study, chemical oxygen demand is the concentration of reduced substances
that are oxidizable by inorganic means. The primary component of chemical oxygen demand is sulfide
released from sediments. Oxidation of sulfide to sulfate may remove substantial quantities of dissolved
oxygen from the water column.
4.1.7 Dissolved Oxygen
Dissolved oxygen is required for the existence of higher life forms. Oxygen availability
determines the distribution of organisms and the flows of energy and nutrients in an ecosystem.
Dissolved oxygen is a central component of the water quality model.
4.1.8 Total Active Metal
Both phosphate and dissolved silica sorb to inorganic solids, primarily iron and manganese.
Sorption and subsequent settling is one pathway for removal of phosphate and silica from the water
column. Consequently, the concentration and transport of iron and manganese are represented in the
model. Limited data do not allow a complete treatment of iron and manganese chemistry, however.
Rather, a single-state variable, total active metal, is defined as the total concentration of metals that are
active in phosphate and silica transport. Total active metal is partitioned between particulate and
dissolved phases by an oxygen-dependent partition coefficient.
4.1.9 Salinity
Salinity is a conservative tracer that provides verification of the transport component of the
model and facilitates examination of conservation of mass. Salinity also influences the dissolved oxygen
saturation concentration and is used in the determination of kinetics constants that differ in saline and
fresh water.
4.1.10 Temperature
Temperature is a primary determinant of the rate of biochemical reactions. Reaction rates
increase as a function of temperature, although extreme temperatures result in the mortality of organisms.
4.2 Conservation of Mass Equation
The governing mass-balance equation for each of the water quality state variables may be
expressed as:
4-4
4 - EFDC Water Quality Model
-------
dC + d(uC) + d(vQ + d(wQ
dt dx dy dz
—[ + A
x dx) By
K + A K^\ ~ Sc
y dy) 6z^ z dz)
(4-1)
C = concentration of a water quality state variable
u, v, w = velocity components in the x-, y-, and z-directions, respectively
Kx, Ky, Kz = turbulent diffiisivities in the x-, y-, and z-directions, respectively
Sc = internal and external sources and sinks per unit volume.
The last three terms on the left-hand side (LHS) of Eq. 4-1 account for the advective transport,
and the first three terms on the right-hand side (RHS) of Eq. 4-1 account for the diffusive transport.
These six terms for physical transport are analogous to, and thus the numerical method of solution is the
same as, those in the mass-balance equation for salinity in the hydrodynamic model (Hamrick 1992a).
The last term in Eq. 4-1 represents the kinetic processes and external loads for each of the state variables.
The present model solves Eq. 4-1 after decoupling the kinetic terms from the physical transport terms.
The solution scheme for both the physical transport (Hamrick 1992a) and the kinetic equations is second-
order accurate.
The governing mass-balance equation for water quality state variables (Eq. 4-1) consists of
physical transport, advective and diffusive, and kinetic processes. When solving Eq. 4-1, the kinetic
terms are decoupled from the physical transport terms. The mass-balance equation for physical transport
only, which takes the same form as the salt-balance equation, is:
d_
dx
dC + d(uC) + d(vC) + d(wC)
dt dx dy dz
k?£}
dz
The equation for kinetic processes only, which will be referred to as the kinetic equation, is:
dC
dt
= sr
(4-2)
(4-3)
which may be expressed as:
dC
dt
= KC + R
(4-4)
where K is kinetic rate (time1) and R is source/sink term (mass volume"1 time"1). Equation 4-4 is
obtained by linearizing some terms in the kinetic equations, mostly Monod type expressions. Hence, K
4 - EFDC Water Quality Model
4-5
-------
and R are known values in Eq. 4-4. Equation 4-2 is identical to, and thus its numerical method of
solution is the same as, the mass-balance equation for salinity (Hamrick 1992a).
The remainder of this chapter details the kinetics portion of the mass-conservation equation for
each state variable. Parameters are defined where they first appear. All parameters are listed, in
alphabetical order, in an appendix. For consistency with reported rate coefficients, kinetics are detailed
using a temporal dimension of days. Within the CE-QUAL-ICM computer code, kinetics sources and
sinks are converted to a dimension of seconds before employment in the mass-conservation equation.
4.3 Algae
Algae, which occupies a central role in the model (Figure 4-1), are grouped into three model state
variables: cyanobacteria (blue-green algae), diatoms, and green algae. The subscript, x, is used to denote
four algal groups: c for cyanobacteria, d for diatoms, g for green algae, and m for macroalgae. Sources
and sinks included in the model are
• growth (production)
• basal metabolism
• predation
• settling
• external loads
Equations describing these processes are largely the same for the four algal groups with differences in the
values of parameters in the equations. The kinetic equation describing these processes is:
Bx = algal biomass of algal group x(gC m"3)
t = time (day)
Px = production rate of algal group x (day1)
BMX = basal metabolism rate of algal group x (day1)
PR,, = predation rate of algal group x (day1)
WSX = settling velocity of algal group x (m day"1)
WBX = external loads of algal group x(gC day"1)
V = cell volume (m3).
The model simulates the total biomass of the macroalgae rather than the size of the macroalgae;
therefore, they can be treated as other groups of algae. Since macroalgae attach to the bottom, they are
limited to growing in the bottom layer only and are not be transported through water movement.
dB
in 8
— = (P - BM - PR)B + —(WS B ) +
Qj V X X X' X v X X'
WB
(4-5)
V
4-6
4 - EFDC Water Quality Model
-------
4.3.1 Production (Algal Growth)
Algal growth depends on nutrient availability, ambient light, and temperature. The effects of
these processes are considered to be multiplicative:
p, = PK-mmm (4-6)
PMX = maximum growth rate under optimal conditions for algal group x (day1)
fi (N) = effect of suboptimal nutrient concentration (0 < f, < 1)
f2(I) = effect of suboptimal light intensity (0 < f2 < 1)
f3(T) = effect of suboptimal temperature (0 < f3 < 1).
The freshwater cyanobacteria may undergo rapid mortality in salt water, e.g., freshwater
organisms in the Potomac River (Thomann et al. 1985). For the freshwater organisms, the increased
mortality may be included in the model by retaining the salinity toxicity term in the growth equation for
cyanobacteria:
Pc = PMC ft(N) f2(I) f,(T) ft(S) (4-7)
f4(S) = effect of salinity on cyanobacteria growth (0 < f4 < 1).
Activation of the salinity toxicity term, f4 (S), is an option in the source code.
4.3.2 Effect of Nutrients on Algal Growth
Using Liebig's "law of the minimum" (Odum 1971) that growth is determined by the nutrient in
least supply, the nutrient limitation for growth of cyanobacteria and green algae is expressed as:
\
f^N) = minimum
NH4 + N03 P04d
KHNx + NH4 + N03 KHPx + P04d
(4-8)
NH4 = ammonium nitrogen concentration (g N m"3)
NO 3 = nitrate nitrogen concentration (g N m"3)
KHNX = half-saturation constant for nitrogen uptake for algal group x (g N m"3)
P04d = dissolved phosphate phosphorus concentration (g P m"3)
KHPX = half-saturation constant for phosphorus uptake for algal group x(gP m"3).
Some cyanobacteria, e.g., Anabaena, can fix nitrogen from atmosphere and thus are not limited
by nitrogen. Hence, Eq. 4-8 is not applicable to the growth of nitrogen fixers.
Since diatoms require silica as well as nitrogen and phosphorus for growth, the nutrient
limitation for diatoms is expressed as:
4 - EFDC Water Quality Model
4-7
-------
/
\
f^N) = minimum
NH4 + N03
P04d
SAd
(4-9)
v
KHNd + NH4 + NO 3 ' KHP d + P04d ' KHS + SAd
/
SAd = concentration of dissolved available silica (g Si m"3)
KHS = half-saturation constant for silica uptake for diatoms (g Si m"3).
4.3.3 Effect of Light on Algal Growth
The daily and vertically integrated form of Steele's equation is:
exp (- Kess \H; + Az]j
(4-10)
(4-11)
(4-12)
FD = fractional daylength (0 < FD < 1)
Kess = total light extinction coefficient (m1)
Az = layer thickness (m)
I0 = daily total light intensity at water surface (langleys day"1)
(Is)x = optimal light intensity for algal group x (langleys day"1)
Ht = depth from the free surface to the top of the layer (m).
Light extinction in the water column consists of three fractions in the model: a background value
dependent on water color, extinction due to suspended particles, and extinction due to light absorption by
ambient chlorophyll:
Keb = background light extinction (m1)
KeTSS = light extinction coefficient for total suspended solid (m"1 per g m"3)
TSS = total suspended solid concentration (g m"3) provided from the hydrodynamic model
Kechl = light extinction coefficient for chlorophyll 'a' (m"1 per mg Chi m"3)
CChlx = carbon-to-chlorophyll ratio in algal group x(gC per mg Chi).
Since macroalgae only attach to the bottom, they are not included in computation of the light
extinction Self shading is not considered for macroalgae for the present model. For a model application
4-8 4 - EFDC Water Quality Model
(4-13)
-------
that does not simulate TSS, the KeTSS term may be set to zero and Keb may be estimated to include light
extinction due to suspended solid.
Optimal light intensity (Is) for photosynthesis depends on algal taxonomy, duration of exposure,
temperature, nutritional status, and previous acclimation. Variations in Is are largely due to adaptations
by algae intended to maximize production in a variable environment. Steel (1962) noted the result of
adaptations is that optimal intensity is a consistent fraction (approximately 50%) of daily intensity.
Kremer and Nixon (1978) reported an analogous finding that maximum algal growth occurs at a constant
depth (approximately 1 m) in the water column. Their approach is adopted so that optimal intensity is
expressed as:
(/) = maximum {(/) e (/) } (4-14)
v s'x (V o'avg ' v s>minj
(Dopt)x = depth of maximum algal growth for algal group x (m)
(Uavg= adjusted surface light intensity (langleys day"1).
A minimum, (Is)mm, in Eq. 4-14 is specified so that algae do not thrive at extremely low light
levels. The time required for algae to adapt to changes in light intensity is recognized by estimating (Is)x
based on a time-weighted average of daily light intensity:
(/) = CI I + CL L + CI L (4-15)
v o'avg a o b 1 c 2
I, = daily light intensity 1 day preceding model day (langleys day"1)
I2 = daily light intensity 2 days preceding model day (langleys day"1)
CIa, CIb, CIc = weighting factors for I0, I, and I2, respectively: CIa + CIb + CIc = 1.
4.3.4 Effect of Temperature on Algal Growth
A Gaussian probability curve is used to represent temperature dependency of algal growth:
UT) = exp(-KTGlx[T - TMJ2) if T < TMX
= exp(-KTG2x[IMx - if) if T > TMx (4-16)
T = temperature (°C) provided from the hydrodynamic model
TMX = optimal temperature for algal growth for algal group x (°C)
KTG1X = effect of temperature below TMX on growth for algal group x (°C2)
KTG2X = effect of temperature above TMX on growth for algal group x (°C 2).
4.3.5 Effect of Salinity on Growth of Freshwater Cyanobacteria
The growth of freshwater cyanobacteria in salt water is limited by:
4 - EFDC Water Quality Model
4-9
-------
_ STOX2
*4^) o o
STOX2 + S2
(4-17)
STOX = salinity at which Microcystis growth is halved (ppt)
S = salinity in water column (ppt) provided from the hydrodynamic model.
4.3.6 Algal Basal Metabolism
Algal biomass in the present model decreases through basal metabolism (respiration and
excretion) and predation. Basal metabolism in the present model is the sum of all internal processes that
decrease algal biomass and consists of two parts; respiration and excretion. In basal metabolism, algal
matter (carbon, nitrogen, phosphorus, and silica) is returned to organic and inorganic pools in the
environment, mainly to dissolved organic and inorganic matter. Respiration, which may be viewed as a
reversal of production, consumes dissolved oxygen. Basal metabolism is considered to be an
exponentially increasing function of temperature:
B M R = basal metabolism rate at TR for algal group x (day1)
KTBX = effect of temperature on metabolism for algal group x (°C_1)
TR,, = reference temperature for basal metabolism for algal group x (°C).
4.3.7 Algal Predation
The present model does not include zooplankton. Instead, a constant rate is specified for algal
predation, which implicitly assumes zooplankton biomass is a constant fraction of algal biomass. An
equation similar to that for basal metabolism (Eq. 4-18) is used for predation:
PRRX = predation rate at TRX for algal group x (day1).
The difference between predation and basal metabolism lies in the distribution of the end
products of the two processes. In predation, algal matter (carbon, nitrogen, phosphorus, and silica) is
returned to organic and inorganic pools in the environment, mainly to particulate organic matter. The
predation for macroalgae is a lumped parameter that includes losses due to grazing, frond breakage, and
other losses. This implicitly assumes that the losses are a fraction of the biomass.
4.3.8 Algal Settling
Settling velocities for four algal groups, WSC, WSd, WSg, and WSm, are specified as an input.
Seasonal variations in settling velocity of diatoms can be accounted for by specifying time-varying WSd.
BMx = RMRx ¦ exp (KTKx ['/' - TR,])
(4-18)
PR = PRR exp (KTB [T - TR ])
X X X X
(4-19)
4-10
4 - EFDC Water Quality Model
-------
4.4 Organic Carbon
The present model has three state variables for organic carbon: refractory particulate, labile
particulate, and dissolved.
4.4.1 Particulate Organic Carbon
Labile and refractory distinctions are based on the time scale of decomposition. Labile
particulate organic carbon with a decomposition time scale of days to weeks decomposes rapidly in the
water column or in the sediments. Refractory particulate organic carbon with a longer-than-weeks
decomposition time scale decomposes slowly, primarily in the sediments, and may contribute to sediment
oxygen demand years after decomposition. For labile and refractory particulate organic carbon, sources
and sinks included in the model are (Fig. 4-1):
• algal predation
• dissolution to dissolved organic carbon
• settling
• external loads.
The governing equations for refractory and labile particulate organic carbons are:
dRP0C = Y, FCRPPRxBx - Kw,or RP()C + —(WS^ RPOC) + (4-20)
Bt x=c,d,g,m X dz V
dLP0C = Y, FCLP PRxBx - KlpocLPOC + —(WSlpLPOC) + WLP0C (4-21)
dt x=c,d,g,m X dz V
RPOC = concentration of refractory particulate organic carbon (g C m"3)
LPOC = concentration of labile particulate organic carbon (g C m"3)
FCRP = fraction of predated carbon produced as refractory particulate organic carbon
FCLP = fraction of predated carbon produced as labile particulate organic carbon
Krpqc = dissolution rate of refractory particulate organic carbon (day1)
KLpoc = dissolution rate of labile particulate organic carbon (day1)
WSrp = settling velocity of refractory particulate organic matter (m day"1)
WSLP = settling velocity of labile particulate organic matter (m day"1)
WRPOC = external loads of refractory particulate organic carbon (g C day"1)
WLPOC = external loads of labile particulate organic carbon (g C day"1).
4.4.2 Dissolved Organic Carbon
Sources and sinks for dissolved organic carbon included in the model are (Fig. 4-1):
• algal excretion (exudation) and predation
• dissolution from refractory and labile particulate organic carbon
4 - EFDC Water Quality Model
4-11
-------
• heterotrophic respiration of dissolved organic carbon (decomposition)
• denitrification
• external loads
The kinetic equation describing these processes is:
\
(
dDOC
dt
x=c,d,g,m
KHR
FCD - (1 - FCD)-
KHR + DO
BM + FCDPPR
¦B
+ Kl(rorRPOC + KLpocLPOC - KhrDOC - DenitDOC + WI)^C (4-22)
DOC = concentration of dissolved organic carbon (g C m~3)
FCDX = fraction of basal metabolism exuded as dissolved organic carbon at infinite dissolved oxygen
concentration for algal group x
KHR,, = half-saturation constant of dissolved oxygen for algal dissolved organic carbon excretion for
group x (g 02 m"3)
DO = dissolved oxygen concentration (g 02 m"3)
FCDP = fraction of predated carbon produced as dissolved organic carbon
Khr = heterotrophic respiration rate of dissolved organic carbon (day1)
Denit = denitrification rate (day1) given in Eq. 4-34
WDOC = external loads of dissolved organic carbon (g C day"1).
The remainder of this section explains each term in Equations 4-20 to 4-22.
4.4.3 Effect of Algae on Organic Carbon
The terms within summation (V) in Equations 4-20 to 4-22 account for the effects of algae on
organic carbon through basal metabolism and predation.
4.4.3.1 Basal metabolism. Basal metabolism, consisting of respiration and excretion, returns
algal matter (carbon, nitrogen, phosphorus, and silica) back to the environment. Loss of algal biomass
through basal metabolism is (Eq. 4-18):
— = -BM B (4-23)
dt x x
which indicates that the total loss of algal biomass due to basal metabolism is independent of ambient
dissolved oxygen concentration. In this model, it is assumed that the distribution of total loss between
respiration and excretion is constant as long as there is sufficient dissolved oxygen for algae to respire.
Under that condition, the losses by respiration and excretion may be written as:
(1 - FCD ) BM B due to respiration (4-24)
4-12
4 - EFDC Water Quality Model
-------
FCD BM B
X XX
due to excretion
(4-25)
where FCDX is a constant of value between 0 and 1. Algae cannot respire in the absence of oxygen,
however. Although the total loss of algal biomass due to basal metabolism is oxygen-independent
(Eq. 4-23), the distribution of total loss between respiration and excretion is oxygen-dependent. When
oxygen level is high, respiration is a large fraction of the total. As dissolved oxygen becomes scarce,
excretion becomes dominant. Thus, Eq. 4-24 represents the loss by respiration only at high oxygen
levels. In general, Eq. 4-24 can be decomposed into two fractions as a function of dissolved oxygen
availability:
(1 - FCD ) BM B due to respiration (4-26)
x KHR + DO x x
X
KHR
(\ - FCD) BM B due to excretion (4-27)
x KHRx + DO x x
Equation 4-26 represents the loss of algal biomass by respiration, and Eq. 4-27 represents additional
excretion due to insufficient dissolved oxygen concentration. The parameter KHRX, which is defined as
the half-saturation constant of dissolved oxygen for algal dissolved organic carbon excretion in Eq. 4-22,
can also be defined as the half-saturation constant of dissolved oxygen for algal respiration in Eq. 4-26.
Combining Equations 4-25 and 4-27, the total loss due to excretion is:
BM B (4"28)
( \
KHR
FCD + (1 - FCD )
x. V r'
KHR + DO
X X
Equations 4-26 and 4-28 combine to give the total loss of algal biomass due to basal metabolism, BMX BX
(Eq. 4-23). The definition of FCDX in Eq. 4-22 becomes apparent in Eq. 4-28; i.e., fraction of basal
metabolism exuded as dissolved organic carbon at infinite dissolved oxygen concentration. At zero
oxygen level, 100% of total loss due to basal metabolism is by excretion regardless of FCDX. The end
carbon product of respiration is primarily carbon dioxide, an inorganic form not considered in the present
model, while the end carbon product of excretion is primarily dissolved organic carbon. Therefore, Eq.
4-28, that appears in Eq. 4-22, represents the contribution of excretion to dissolved organic carbon, and
there is no source term for particulate organic carbon from algal basal metabolism in Equations 4-20 and
4-21.
4.4.3.2 Predation. Algae produce organic carbon through the effects of predation. Zooplankton
take up and redistribute algal carbon through grazing, assimilation, respiration, and excretion. Since
4 - EFDC Water Quality Model
4-13
-------
zooplankton are not included in the model, routing of algal carbon through zooplankton predation is
simulated by empirical distribution coefficients in Equations 4-20 to 4-22; FCRP, FCLP, and FCDP. The
sum of these three predation fractions should be unity.
4.4.4 Heterotrophic Respiration and Dissolution
The second term on the RHS of Equations 4-20 and 4-21 represents dissolution of particulate to
dissolved organic carbon and the third term in the second line of Eq. 4-22 represents heterotrophic
respiration of dissolved organic carbon. The oxic heterotrophic respiration is a function of dissolved
oxygen: the lower the dissolved oxygen, the smaller the respiration term becomes. Heterotrophic
respiration rate, therefore, is expressed using a Monod function of dissolved oxygen:
K,,r = — Knoc (4-29)
"R KHORdo + DO DOC
KHORdo = oxic respiration half-saturation constant for dissolved oxygen (g 02 m"3)
KDOc = heterotrophic respiration rate of dissolved organic carbon at infinite dissolved oxygen
concentration (day1).
Dissolution and heterotrophic respiration rates depend on the availability of carbonaceous
substrate and on heterotrophic activity. Algae produce labile carbon that fuels heterotrophic activity:
dissolution and heterotrophic respiration do not require the presence of algae though, and may be fueled
entirely by external carbon inputs. In the model, algal biomass, as a surrogate for heterotrophic activity,
is incorporated into formulations of dissolution and heterotrophic respiration rates. Formulations of
these rates require specification of algal-dependent and algal-independent rates:
KrpoC (Krc + KRCalg £ Bx) ^f(KTHDR[T - TRhdr]) (4-30)
x-c,d,g
Klpoc = (Klc + KLCaig 52 Bx) ^V(KThdr[T - TRhdr]) (4-31)
x-c,d,g
K-DOC + ^DCalg E BJ ^p(KTMNLlT- TRmml\) (4-32)
x-c,d,g
Krc = minimum dissolution rate of refractory particulate organic carbon (day1)
Klc = minimum dissolution rate of labile particulate organic carbon (day1)
Kdc = minimum respiration rate of dissolved organic carbon (day1)
Kkr:iig, KLCalg = constants that relate dissolution of refractory and labile particulate organic carbon,
respectively, to algal biomass (day-1 per g C m"3)
KDcaig = constant that relates respiration to algal biomass (day-1 per g C m"3)
KThdr = effect of temperature on hydrolysis of particulate organic matter (°C_1)
4-14 4 - EFDC Water Quality Model
-------
TRhdr = reference temperature for hydrolysis of particulate organic matter (°C)
KTMvL = effect of temperature on mineralization of dissolved organic matter (°C_1)
TRV|,:l = reference temperature for mineralization of dissolved organic matter (°C).
Equations 4-30 to 4-32 have exponential functions that relate rates to temperature.
In the present model, the term "hydrolysis" is defined as the process by which particulate organic
matter is converted to dissolved organic form, and thus includes both dissolution of particulate carbon
and hydrolysis of particulate phosphorus and nitrogen. Therefore, the parameters, KTHDR and TRhdr, are
also used for the temperature effects on hydrolysis of particulate phosphorus (Equations 4-28 and 4-29)
and nitrogen (Equations 4-54 and 4-55). The term "mineralization" is defined as the process by which
dissolved organic matter is converted to dissolved inorganic form, and thus includes both heterotrophic
respiration of dissolved organic carbon and mineralization of dissolved organic phosphorus and nitrogen.
Therefore, the parameters, KTvr..:L and TRvr..:L. are also used for the temperature effects on mineralization
of dissolved phosphorus (Eq. 4-46) and nitrogen (Eq. 4-56).
4.4.5 Effect of Denitrification on Dissolved Organic Carbon
As oxygen is depleted from natural systems, organic matter is oxidized by the reduction of
alternate electron acceptors. Thermodynamically, the first alternate acceptor reduced in the absence of
oxygen is nitrate. The reduction of nitrate by a large number of heterotrophic anaerobes is referred to as
denitrification, and the stoichiometry of this reaction is (Stumm and Morgan 1981):
4 NOf +4 H+ + 5 CH20 - 2 N2 + 1H20 + 5 C02 (4"33)
The last term in Eq. 4-22 accounts for the effect of denitrification on dissolved organic carbon. The
kinetics of denitrification in the model are first-order:
KHORnn N()3
Denit = — — AANOXKnnr (4-34)
KHORdo + DO KHDNn + N03 DOC
KHDNn = denitrification half-saturation constant for nitrate (g N m"3)
AANOX = ratio of denitrification rate to oxic dissolved organic carbon respiration rate.
In Eq. 4-34, the dissolved organic carbon respiration rate, KD0C, is modified so that significant
decomposition via denitrification occurs only when nitrate is freely available and dissolved oxygen is
depleted. The ratio, AANOX, makes the anoxic respiration slower than oxic respiration. Note that KD0C,
defined in Eq. 4-32, includes the temperature effect on denitrification.
4 - EFDC Water Quality Model
4-15
-------
4.5 Phosphorus
The present model has four state variables for phosphorus: three organic forms (refractory
particulate, labile particulate, and dissolved) and one inorganic form (total phosphate).
4.5.1 Particulate Organic Phosphorus
For refractory and labile particulate organic phosphorus, sources and sinks included in the model
are (Fig. 4-1):
• algal basal metabolism and predation
• dissolution to dissolved organic phosphorus
• settling
• external loads.
The kinetic equations for refractory and labile particulate organic phosphorus are:
= E (FPRxBMx + FPRPPRJAPCB, - K^RPOP
vi x-c,d,g,m
+ —(WS -RPOF) + WRP0P (4-35)
dz V
= E (.FPL.-BM, ~ FPLPPRJAPCB, - KlpopLPOP
Ot x-c,d,g,m
+ JL(WSlpLPOP) + WLP0P (4-36)
dz V
RPOP = concentration of refractory particulate organic phosphorus (g P m"3)
LPOP = concentration of labile particulate organic phosphorus (g P m"3)
FPRX = fraction of metabolized phosphorus by algal group x produced as refractory particulate organic
phosphorus
FPLX = fraction of metabolized phosphorus by algal group x produced as labile particulate organic
phosphorus
FPRP = fraction of predated phosphorus produced as refractory particulate organic phosphorus
FPLP = fraction of predated phosphorus produced as labile particulate organic phosphorus
APC = mean algal phosphorus-to-carbon ratio for all algal groups (g P per g C)
Krpqp = hydrolysis rate of refractory particulate organic phosphorus (day1)
KLpop = hydrolysis rate of labile particulate organic phosphorus (day1)
WRPOP = external loads of refractory particulate organic phosphorus (g P day"1)
WLPOP = external loads of labile particulate organic phosphorus (g P day"1).
4-16
4 - EFDC Water Quality Model
-------
4.5.2 Dissolved Organic Phosphorus
Sources and sinks for dissolved organic phosphorus included in the model are (Fig. 4-1):
• algal basal metabolism and predation
• dissolution from refractory and labile particulate organic phosphorus
• mineralization to phosphate phosphorus
• external loads.
The kinetic equation describing these processes is:
dD0P = Y (FPD BM + FPDPPR )APCB
^ X X XJ X
Ot x-c,d,g,m
* KxpopRPOP * klpoplpop - Kdopdop * (4.37)
DOP = concentration of dissolved organic phosphorus (g P m"3)
FPDX = fraction of metabolized phosphorus by algal group x produced as dissolved organic phosphorus
FPDP = fraction of predated phosphorus produced as dissolved organic phosphorus
Kdop = mineralization rate of dissolved organic phosphorus (day1)
WDOP = external loads of dissolved organic phosphorus (g P day"1).
4.5.3 Total Phosphate
For total phosphate that includes both dissolved and sorbed phosphate (Section 4.5.4), sources
and sinks included in the model are (Fig. 4-1):
• algal basal metabolism, predation, and uptake
• mineralization from dissolved organic phosphorus
• settling of sorbed phosphate
• sediment-water exchange of dissolved phosphate for the bottom layer only
• external loads.
The kinetic equation describing these processes is:
dP04t = Y (FPI BM + FPIP PR - P )APCB + KnnpDOP
j XX X X X
vl x-c,d,g,m
5 /ri/c ,)/) i , BFP04d WP04t (A 00,
+ —(wSTSsp04p) + Az + —— ( )
P04t = total phosphate (g P m"3) = P04d + P04p (4-39)
P04d = dissolved phosphate (g P m"3)
P04p = particulate (sorbed) phosphate (g P m"3)
FPIX = fraction of metabolized phosphorus by algal group x produced as inorganic phosphorus
4 - EFDC Water Quality Model
4-17
-------
FPIP = fraction of predated phosphorus produced as inorganic phosphorus
WSTSS = settling velocity of suspended solid (m day"1), provided by the hydrodynamic model
BFP04d = sediment-water exchange flux of phosphate (g P m"2 day"1), applied to the bottom layer only
WP04t = external loads of total phosphate (g P day"1).
In Eq. 4-38, if total active metal is chosen as a measure of sorption site, the settling velocity of
total suspended solid, WSTSS, is replaced by that of particulate metal, WSs (Sections 4.5.4 and 4.10). The
remainder of this section explains each term in Equations 4-35 to 4-38, except BFP04d (benthic flux of
dissolved orthophosphate), which is described in Chapter 5.
4.5.4 Total Phosphate System
Suspended and bottom sediment particles (clay, silt, and metal hydroxides) adsorb and desorb
phosphate in river and estuarine waters. This adsorption-desorption process has been suggested to buffer
phosphate concentration in water column and to enhance the transport of phosphate away from its
external sources (Carritt and Goodgal 1954; Froelich 1988; Lebo 1991). To ease the computational
complication due to the adsorption-desorption of phosphate, dissolved and sorbed phosphate are treated
and transported as a single state variable. Therefore, the model phosphate state variable, total phosphate,
is defined as the sum of dissolved and sorbed phosphate (Eq. 4-39), and the concentrations for each
fraction are determined by equilibrium partitioning of their sum.
In CE-QUAL-ICM, sorption of phosphate to particulate species of metals including iron and
manganese was considered based on a phenomenon observed in the monitoring data from the mainstem
of the Chesapeake Bay: phosphate was rapidly depleted from anoxic bottom waters during the autumn
reaeration event (Cerco and Cole 1993). Their hypothesis was that reaeration of bottom waters caused
dissolved iron and manganese to precipitate, and phosphate sorbed to newly formed metal particles and
rapidly settled to the bottom. One state variable, total active metal, in CE-QUAL-ICM was defined as the
sum of all metals that act as sorption sites, and the total active metal was partitioned into particulate and
dissolved fractions via an equilibrium partitioning coefficient (Section 4.10). Then phosphate was
assumed to sorb to only the particulate fraction of the total active metal.
In the treatment of phosphate sorption in CE-QUAL-ICM, the particulate fraction of metal
hydroxides was emphasized as a sorption site in bottom waters under anoxic conditions. Phosphorus is a
highly particle-reactive element, and phosphate in solution reacts quickly with a wide variety of surfaces,
being taken up by and released from particles (Froelich 1988). The present model has two options, total
suspended solid and total active metal, as a measure of a sorption site for phosphate, and dissolved and
4-18
4 - EFDC Water Quality Model
-------
sorbed fractions are determined by equilibrium partitioning of their sum as a function of total suspended
solid or total active metal concentration:
P04p = Kp°4p TSS P04t or P04p = Kp°4p TAMP P04t (4-40)
1 + Kp04pTSS 1 + Kp04pTAMP
P04d = P04t or P04d = P04t
1 - Kp04pTSS 1 - Kp04pTAMP
= P04t - P04p (4-4!)
Kp04p = empirical coefficient relating phosphate sorption to total suspended solid (per g m"3) or
particulate total active metal (per mol m~3) concentration
TAMp = particulate total active metal (mol m"3).
Dividing Eq. 4-40 by Eq. 4-41 gives:
K = P®4P 1 or K P04p_ 1 (4-42)
P°4p pQ4d TSS P04p p(J4d TM/[p
where the meaning of Kp04p becomes apparent, i.e., the ratio of sorbed to dissolved phosphate per unit
concentration of total suspended solid or particulate total active metal (i.e., per unit sorption site
available).
4.5.5 Algal Phosphorus-to-Carbon Ratio (APC)
Algal biomass is quantified in units of carbon per volume of water. In order to express the
effects of algal biomass on phosphorus and nitrogen, the ratios of phosphorus-to-carbon and nitrogen-to-
carbon in algal biomass must be specified. Although global mean values of these ratios are well known
(Redfield et al. 1963), algal composition varies especially as a function of nutrient availability. As
phosphorus and nitrogen become scarce, algae adjust their composition so that smaller quantities of these
vital nutrients are required to produce carbonaceous biomass (DiToro 1980; Parsons et al. 1984).
Examining the field data from the surface of upper Chesapeake Bay, Cerco and Cole (1993) showed that
the variation of nitrogen-to-carbon stoichiometry was small and thus used a constant algal nitrogen-to-
carbon ratio, ANCX. Large variations, however, were observed for algal phosphorus-to-carbon ratio
indicating the adaptation of algae to ambient phosphorus concentration (Cerco and Cole 1993): algal
phosphorus content is high when ambient phosphorus is abundant and is low when ambient phosphorus
is scarce. Thus, a variable algal phosphorus-to-carbon ratio, APC, is used in model formulation. A mean
4 - EFDC Water Quality Model
4-19
-------
ratio for all algal groups, APC, is described by an empirical approximation to the trend observed in field
data (Cerco & Cole 1994):
APC = [CP , + CP ,-exp [-CP qP04d\Xl (4-43
\ prml prmz ~ l prm.5 JJ
CPprmi = minimum carbon-to-phosphorus ratio (g C per g P)
CPprm2 = difference between minimum and maximum carbon-to-phosphorus ratio (g C per g P)
CPPrm3 = effect of dissolved phosphate concentration on carbon-to-phosphorus ratio (per g P m"3).
4.5.6 Effect of Algae on Phosphorus
The terms within summation (V) in Equations 4-35 to 4-38 account for the effects of algae on
phosphorus. Both basal metabolism (respiration and excretion) and predation are considered, and thus
formulated, to contribute to organic and phosphate phosphorus. That is, the total loss by basal
metabolism (BMX Bx in Eq. 4-5) is distributed using distribution coefficients; FPRX, FPLX, FPDX, and
FPIX. The total loss by predation (PRX BX in Eq. 4-5), is also distributed using distribution coefficients;
FPRP, FPLP, FPDP, and FPIP. The sum of four distribution coefficients for basal metabolism should be
unity, and so is that for predation. Algae take up dissolved phosphate for growth, and algae uptake of
phosphate is represented by (- £ Px APC Bx) in Eq. 4-38.
4.5.7 Mineralization and Hydrolysis
The third term on the RHS of Equations 4-35 and 4-36 represents hydrolysis of particulate
organic phosphorus, and the last term in Eq. 3-7 represents mineralization of dissolved organic
phosphorus. Mineralization of organic phosphorus is mediated by the release of nucleotidase and
phosphatase enzymes by bacteria (Chrost and Overbek 1987) and algae (Boni et al. 1989). Since the
algae themselves release the enzymes and bacterial abundance is related to algal biomass, the rate of
organic phosphorus mineralization is related to algal biomass in model formulation. Another mechanism
included in model formulation is that algae stimulate production of an enzyme that mineralizes organic
phosphorus to phosphate when phosphate is scarce (Chrost and Overbek 1987; Boni et al. 1989). The
formulations for hydrolysis and mineralization rates including these processes are:
Krpof - (Krp ~ KHpKfFF04dKmahY.d K) ^(kthdr\J - TR„dr]) (4-44)
Kuvp - (Klp ~ KHpTp04dK^MgB,)tXV(KTHDR[T ~ TR"Dr]) <4"45)
Kdop " (Kdf * KHpTp04dKDF°*J?d.BJ eXP(KTMNL[T ' TRmml]) <4"4<>)
KkP = minimum hydrolysis rate of refractory particulate organic phosphorus (day1)
4-20 4 - EFDC Water Quality Model
-------
Klp = minimum hydrolysis rate of labile particulate organic phosphorus (day1)
Kdp = minimum mineralization rate of dissolved organic phosphorus (day1)
Kki,:iig. KLPalg = constants that relate hydrolysis of refractory and labile particulate organic phosphorus,
When phosphate is abundant relative to KHP, the rates become close to the minimum values with
little influence from algal biomass. When phosphate becomes scarce relative to KHP, the rates increase
with the magnitude of increase depending on algal biomass. Equations 4-44 to 4-46 have exponential
functions that relate rates to temperature.
4.6 Nitrogen
The present model has five state variables for nitrogen: three organic forms (refractory
particulate, labile particulate, and dissolved) and two inorganic forms (ammonium and nitrate). The
nitrate state variable in the model represents the sum of nitrate and nitrite.
4.6.1 Particulate Organic Nitrogen
For refractory and labile particulate organic nitrogen, sources and sinks included in the model are
(Figure 4-1):
• algal basal metabolism and predation
• dissolution to dissolved organic nitrogen
• settling
• external loads.
The kinetic equations for refractory and labile particulate organic nitrogen are:
respectively, to algal biomass (day-1 per g C m~3)
KDpaig = constant that relates mineralization to algal biomass (day-1 per g C m"3)
KHP = mean half-saturation constant for algal phosphorus uptake (g P m"3).
X
(4-47)
dRPON = ^2
E 0FNR BMx+FNRPPR)ANCxBx - K^RPON
x-c,d,g,m
+ ^(WS^RPON) +
oz
WRPON
V
(4-48)
dLPON
dt
x-c,d,g,m
£ (FNLxBMx +FNLPPR_f)ANCxBx - KLm,LPON
4 - EFDC Water Quality Model
4-21
-------
+ — (WSlpLPON) + WLP0N
dz V
(4-49)
RPON = concentration of refractory particulate organic nitrogen (g N m"3)
LPON = concentration of labile particulate organic nitrogen (g N m"3)
FNRX = fraction metabolized nitrogen by algal group x as refractory particulate organic nitrogen
FNLX = fraction of metabolized nitrogen by algal group x produced as labile particulate organic nitrogen
FNRP = fraction of predated nitrogen produced as refractory particulate organic nitrogen
FNLP = fraction of predated nitrogen produced as labile particulate organic nitrogen
ANCX = nitrogen-to-carbon ratio in algal group x (g N per g C)
Krpon = hydrolysis rate of refractory particulate organic nitrogen (day1)
KLpon = hydrolysis rate of labile particulate organic nitrogen (day1)
WRPON = external loads of refractory particulate organic nitrogen (g N day"1)
WLPON = external loads of labile particulate organic nitrogen (g N day"1).
4.6.2 Dissolved Organic Nitrogen
Sources and sinks for dissolved organic nitrogen included in the model are (Fig. 4-1):
• algal basal metabolism and predation
• dissolution from refractory and labile particulate organic nitrogen
• mineralization to ammonium
• external loads.
The kinetic equation describing these processes is:
dD0N = V (FN/) BM + FNDPPR )ANC B
j ^ JC JC X X X
vi x-c,d,g,m
* Krpon'^PON + K^LPON - K^DON + (4-50)
DON = concentration of dissolved organic nitrogen (g N m"3)
FNDX = fraction of metabolized nitrogen by algal group x produced as dissolved organic nitrogen
FNDP = fraction of predated nitrogen produced as dissolved organic nitrogen
Kdon = mineralization rate of dissolved organic nitrogen (day1)
WDON = external loads of dissolved organic nitrogen (g N day"1).
4.6.3 Ammonium Nitrogen
Sources and sinks for ammonia nitrogen included in the model are (Fig. 4-1):
• algal basal metabolism, predation, and uptake
• mineralization from dissolved organic nitrogen
4-22
4 - EFDC Water Quality Model
-------
• nitrification to nitrate
• sediment-water exchange for the bottom layer only
• external loads.
The kinetic equation describing these processes is:
= E (FNIx-BMx + FNIPPR, - PNXPJANCX-BX + K^DON
vi x-c,d,g,m
- m NH4 ~ + ™2± (4-51)
A z V
FNIX = fraction of metabolized nitrogen by algal group x produced as inorganic nitrogen
FNIP = fraction of predated nitrogen produced as inorganic nitrogen
PNX = preference for ammonium uptake by algal group x (0 < PNX < 1)
Nit = nitrification rate (day1) given in Eq. 4-59
BFNH4 = sediment-water exchange flux of ammonium (g N m"2 day"1), applied to the bottom layer only
WNH4 = external loads of ammonium (g N day"1).
4.6.4 Nitrate Nitrogen
Sources and sinks for nitrate nitrogen included in the model are (Fig. 4-1):
• algal uptake
• nitrification from ammonium
• denitrification to nitrogen gas
• sediment-water exchange for the bottom layer only
• external loads.
The kinetic equation describing these processes is:
dN03 = - V (1 - PN)P ANC B + Nit NH4 - ANDCDenit DOC
j JC jc jc oc
vl x-c,d,g,m
+ BFNQ3 + WNQ3 (4_52)
Az + V
ANDC = mass of nitrate nitrogen reduced per mass of dissolved organic carbon oxidized (0.933 g N per
g C from Eq. 4-33)
BFN03 = sediment-water exchange flux of nitrate (g N m"2 day"1), applied to the bottom layer only
WN03 = external loads of nitrate (g N day"1).
The remainder of this section explains each term in Equations 4-48 to 4-52, except BFNH4 and
BFN03 which are described in Chapter 5.
4 - EFDC Water Quality Model
4-23
-------
4.6.5 Effect of Algae on Nitrogen
The terms within summation (V) in Equations 4-48 to 4-52 account for the effects of algae on
nitrogen. As in phosphorus, both basal metabolism (respiration and excretion) and predation are
considered, and thus formulated, to contribute to organic and ammonium nitrogen. That is, algal nitrogen
released by both basal metabolism and predation are represented by distribution coefficients; FNRX,
FNLX, FNDX, FNIX, FNRP, FNLP, FNDP, and FNIP. The sum of four distribution coefficients for basal
metabolism should be unity; the sum of the predation distribution coefficients should also be unity.
Algae take up ammonium and nitrate for growth, and ammonium is preferred from
thermodynamic considerations. The preference of algae for ammonium is expressed as:
N() 7 KHN
PN = NH4 — + NH4 ^ (4-53)
(KHN + NH4)(KHN +N03) (NH4 + NO 3) (KHN +N03)
This equation forces the preference for ammonium to be unity when nitrate is absent, and to be zero
when ammonium is absent.
4.6.6 Mineralization and Hydrolysis
The third term on the RHS of Equations 4-48 and 4-49 represents hydrolysis of particulate
organic nitrogen and the last term in Eq. 4-50 represents mineralization of dissolved organic nitrogen.
Including a mechanism for accelerated hydrolysis and mineralization during nutrient-limited conditions
(Section 4.5.7), the formulations for these processes are:
' (K•» * KHNZN4
-------
= I V KHN (4-57)
3^™^™ JC
x-c,d,g
Equations 4-54 to 4-56 have exponential functions that relate rates to temperature.
4.6.7 Nitrification
Nitrification is a process mediated by autotrophic nitrifying bacteria that obtain energy through
the oxidation of ammonium to nitrite and of nitrite to nitrate. The stoichiometry of complete reaction is
(Bowie et al. 1985):
NH4+ + 2 02 - NOf + H20 + 2 H+ (4-58)
The first term in the second line of Eq. 4-51 and its corresponding term in Eq. 4-52 represent the effect of
nitrification on ammonium and nitrate, respectively. The kinetics of complete nitrification process are
formulated as a function of available ammonium, dissolved oxygen and temperature:
Ar, DO NH4 ,r f ,rp. .m
Nit = Nit jmJT) (4-59)
KHNitDO + DO KHNitN + NH4 m '"
fmt{T) = exp(-KNitI[T - TNitf) if T < TNit
= exp(-KNit2[TNit - 7]2) if T> TNit (4"6°)
KHNitD0 = nitrification half-saturation constant for dissolved oxygen (g 02 m"3)
KHNitN = nitrification half-saturation constant for ammonium (g N m"3)
Nitm = maximum nitrification rate at TNit (g N m"3 day"1)
TNit = optimum temperature for nitrification (°C)
KNitl = effect of temperature below TNit on nitrification rate (°C-2)
KNit2 = effect of temperature above TNit on nitrification rate (°C-2).
The Monod function of dissolved oxygen in Eq. 4-59 indicates the inhibition of nitrification at
low oxygen level. The Monod function of ammonium indicates that when ammonium is abundant, the
nitrification rate is limited by the availability of nitrifying bacteria. The effect of suboptimal temperature
is represented using Gaussian form.
4.6.8 Denitrification
The effect of denitrification on dissolved organic carbon was described in Section 4.4.5.
Denitrification removes nitrate from the system in stoichiometric proportion to carbon removal as
determined by Eq. 4-33. The last term in the first line of Eq. 4-52 represents this removal of nitrate.
4 - EFDC Water Quality Model
4-25
-------
4.7
silica.
Silica
The present model has two state variables for silica: particulate biogenic silica and available
4.7.1 Particulate Biogenic Silica
Sources and sinks for particulate biogenic silica included in the model are (Fig. 4-1):
• diatom basal metabolism and predation
• dissolution to available silica
• settling
• external loads
The kinetic equation describing these processes is:
^ = (FSPdBMd ~ FSPPPRd)ASCdBd - KsmSU ~ JL(WSdSU) * (4-61)
ot oz V
SU = concentration of particulate biogenic silica (g Si m"3)
FSPd = fraction of metabolized silica by diatoms produced as particulate biogenic silica
FSPP = fraction of predated diatom silica produced as particulate biogenic silica
ASCd = silica-to-carbon ratio of diatoms (g Si per g C)
Ksua = dissolution rate of particulate biogenic silica (day1)
WSU = external loads of particulate biogenic silica (g Si day"1).
4.7.2 Available Silica
Sources and sinks for available silica included in the model are (Fig. 4-1):
• diatom basal metabolism, predation, and uptake
• settling of sorbed (particulate) available silica
• dissolution from particulate biogenic silica
• sediment-water exchange of dissolved silica for the bottom layer only
• external loads.
The kinetic equation describing these processes is:
= (FSIjBMj + FSIPPRd - Pd)ASCdBd + KsmSU + jL(WSTSS SAp)
BFSAd WSA
+ +
Az V
(4-62)
SA = concentration of available silica (g Si m"3) = SAd + SAp (4-63)
SAd = dissolved available silica (g Si m"3)
SAp = particulate (sorbed) available silica (g Si m"3)
4-26 4 - EFDC Water Quality Model
-------
FSId = fraction of metabolized silica by diatoms produced as available silica
FSIP = fraction of predated diatom silica produced as available silica
BFSAd = sediment-water exchange flux of available silica (g Si m"2 day"1), applied to bottom layer only
WSA = external loads of available silica (g Si day"1).
In Eq. 4-62, if total active metal is chosen as a measure of sorption site, the settling velocity of
total suspended solid, WSTSS, is replaced by that of particulate metal, WSs (Sections 4.7.3 and 4.10).
4.7.3 Available Silica System
Analysis of Chesapeake Bay monitoring data indicates that silica shows similar behavior as
phosphate in the adsorption-desorption process (Cerco and Cole 1993). As in phosphate, therefore,
available silica is defined to include both dissolved and sorbed fractions (Eq. 4-63). Treatment of
available silica is the same as total phosphate, and the same method to partition dissolved and sorbed
phosphate is used to partition dissolved and sorbed available silica:
KK1 TSS Kka TAMp
SAp = SA or SAp = — SA (4-64)
1 + KSApTSS 1 + KSApTAMp
SAd = 1 SA or SAd = l- SA
1 - KSApTSS 1 - KSApTAMp
= SA - SAp (4"65)
KSAp = empirical coefficient relating available silica sorption to total suspended solid (per g m"3) or
particulate total active metal (per mol m"3) concentration.
As in KP04p in Section 4.5.4, KSAp is the ratio of sorbed to dissolved available silica per unit
sorption site available.
4.7.4 Effect of Diatoms on Silica
In Equations 4-62 and 4-63, those terms expressed as a function of diatom biomass (Bd) account
for the effects of diatoms on silica. As in phosphorus and nitrogen, both basal metabolism (respiration
and excretion) and predation are considered, and thus formulated, to contribute to particulate biogenic
and available silica. That is, diatom silica released by both basal metabolism and predation are
represented by distribution coefficients; FSPd, FSId, FSPP, and FSIP. The sum of two distribution
coefficients for basal metabolism should be unity and so is that for predation. Diatoms require silica as
well as phosphorus and nitrogen, and diatom uptake of available silica is represented by
(- PdASCdBd) in Eq. 4-63.
4 - EFDC Water Quality Model
4-27
-------
4.7.5 Dissolution
The term (- Ksua SU) in Eq. 4-62 and its corresponding term in Eq. 4-63 represent dissolution of
particulate biogenic silica to available silica. The dissolution rate is expressed as an exponential function
of temperature:
Ksm = Ksv-^P&T^T - TRsm]) (4-66)
Ksu = dissolution rate of particulate biogenic silica at TRSUA (day1)
KTsua = effect of temperature on dissolution of particulate biogenic silica (°C_1)
TRsua = reference temperature for dissolution of particulate biogenic silica (°C).
4.8 Chemical Oxygen Demand
In the present model, chemical oxygen demand is the concentration of reduced substances that
are oxidizable through inorganic means. The source of chemical oxygen demand in saline water is
sulfide released from sediments. A cycle occurs in which sulfate is reduced to sulfide in the sediments
and reoxidized to sulfate in the water column. In fresh water, methane is released to the water column by
the sediment process model. Both sulfide and methane are quantified in units of oxygen demand and are
treated with the same kinetic formulation. The kinetic equation, including external loads, if any, is:
dSOD . DO KC0D.C0D + BFCOD f WCOD
at khcod * do az v
COD = concentration of chemical oxygen demand (g 02-equivalents m"3)
KHcod = half-saturation constant of dissolved oxygen required for oxidation of chemical oxygen
demand (g 02 m"3)
KCOD = oxidation rate of chemical oxygen demand (day1)
BFCOD = sediment flux of chemical oxygen demand (g 02-equivalents m"2 day"1), applied to bottom
layer only
WCOD = external loads of chemical oxygen demand (g 02-equivalents day"1).
An exponential function is used to describe the temperature effect on the oxidation rate of
chemical oxygen demand:
KCOD = KCDexp(KTCOD[T - TRCOD]) (4-68)
Kcd = oxidation rate of chemical oxygen demand at TRrOL) (day1)
KTcod = effect of temperature on oxidation of chemical oxygen demand (°C_1)
TRcqd = reference temperature for oxidation of chemical oxygen demand (°C).
4-28
4 - EFDC Water Quality Model
-------
4.9 Dissolved Oxygen
Sources and sinks of dissolved oxygen in the water column included in the model are (Fig. 4-1):
algal photosynthesis and respiration
nitrification
heterotrophic respiration of dissolved organic carbon
oxidation of chemical oxygen demand
surface reaeration for the surface layer only
sediment oxygen demand for the bottom layer only
external loads.
The kinetic equation describing these processes is:
dDO
dt
£
x-c,d,g,m
(1.3 - 03PN)Px - (1 - FCD )
, BM
x KHR + DO x
\
AOCRB
AONTNUNH4 - AOCRKhrDOC -
DO
KHcod + DO
-KCODCOD
+ K (DO - DO) +
A z
V
(4-69)
AONT =
AOCR =
Kr =
DOs =
SOD =
WDO =
mass of dissolved oxygen consumed per unit mass of ammonium nitrogen nitrified (4.33 g 02
per g N; see Section 4.9.2)
dissolved oxygen-to-carbon ratio in respiration (2.67 g 02 per g C; see Section 4.9.1)
reaeration coefficient (day1): the reaeration term is applied to the surface layer only
saturated concentration of dissolved oxygen (g 02 m"3)
sediment oxygen demand (g 02 m"2 day"1), applied to the bottom layer only; positive is to the
water column
external loads of dissolved oxygen (g 02 day"1).
The two sink terms in Eq. 4-69, heterotrophic respiration and chemical oxygen demand, are
explained in Section 4.4.4 (Eq. 4-29) and Section 4.8 (Eq. 4-67), respectively. The remainder of this
section explains the effects of algae, nitrification, and surface reaeration.
4.9.1 Effect of Algae on Dissolved Oxygen
The first line on the RHS of Eq. 4-69 accounts for the effects of algae on dissolved oxygen.
Algae produce oxygen through photosynthesis and consume oxygen through respiration. The quantity
produced depends on the form of nitrogen utilized for growth. Equations describing production of
dissolved oxygen are (Morel 1983):
4 - EFDC Water Quality Model
4-29
-------
106C02 + 16NH4+ + H^o; + 106H20 - protoplasm + 106O2 + 15H+ (4"7°)
106C02 + 16NO; + H2P()4 + 122H20 + 17// - protoplasm + 1380, (4"71)
When ammonium is the nitrogen source, one mole of oxygen is produced per mole of carbon dioxide
fixed. When nitrate is the nitrogen source, 1.3 moles of oxygen are produced per mole of carbon dioxide
fixed. The quantity, (1.3 - 0.3 PNX), in the first term of Eq. 4-69 is the photosynthesis ratio and
represents the molar quantity of oxygen produced per mole of carbon dioxide fixed. It approaches unity
as the algal preference for ammonium approaches unity.
The last term in the first line of Eq. 4-69 accounts for the oxygen consumption due to algal
respiration (Eq. 4-26). A simple representation of respiration process is:
CH20 + 02 = C02 + H20 (4-72)
from which, AOCR = 2.67 g 02 per g C.
4.9.2 Effect of Nitrification on Dissolved Oxygen
The stoichiometry of nitrification reaction (Eq. 4-58) indicates that two moles of oxygen are
required to nitrify one mole of ammonium into nitrate. However, cell synthesis by nitrifying bacteria is
accomplished by the fixation of carbon dioxide so that less than two moles of oxygen are consumed per
mole ammonium utilized (Wezernak and Gannon 1968), i.e., AONT = 4.33 g 02 per g N.
4.9.3 Effect of Surface Reaeration on Dissolved Oxygen
The reaeration rate of dissolved oxygen at the air-water interface is proportional to the oxygen
gradient across the interface, (DOs - DO), when assuming the air is saturated with oxygen. The saturated
concentration of dissolved oxygen, which decreases as temperature and salinity increase, is specified
using an empirical formula (Genet et al. 1974):
DOs = 14.5532 - 0.38217 T + 5.4258 x 10"3 • T2
- CL (1.665 xlO"4 - 5.866x 10~6 T + 9.796 x 10"8 • T2) (4-73)
CL = chloride concentration (mg/L) = S/l.80655.
The reaeration coefficient includes the effect of turbulence generated by bottom friction
(O'Connor and Dobbins 1958) and that by surface wind stress (Banks and Herrera 1977):
K
K
u
-fl + W
h
eq
— KTTr " 20
Az
(4-74)
4-30
4 - EFDC Water Quality Model
-------
Kro = proportionality constant = 3.933 in MKS unit
ueq = weighted velocity over cross-section (m sec"1) = V(ukVk)/V(Vk)
heq = weighted depth over cross-section (m) = £(Vk)/Bt)
Bt) = width at the free surface (m)
Wrea = wind-induced reaeration (m day"1)
= 0.728 U 1/2 - 0.317 U + 0.0372 U 2 (4"75)
WWW
Uw = wind speed (m sec"1) at the height of 10 m above surface
KTr = constant for temperature adjustment of DO reaeration rate.
4.10 Total Active Metal
The present model requires simulation of total active metal for adsorption of phosphate and silica
if that option is chosen (Fig. 4-1). The total active metal state variable is the sum of iron and manganese
concentrations, both particulate and dissolved. In the model, the origin of total active metal is benthic
sediments. Since sediment release of metal is not explicit in the sediment model (see Chapter 5), release
is specified in the kinetic portion of the water column model. The only other term included is settling of
the particulate fraction. Then the kinetic equation for total active metal, including external loads, if any,
may be written as:
dTAM = KHbmf BFTAMcKtam(T _ Ttam) + JAMp) + (4-76)
dt KHbmf + DO Az dz s V
TAM = total active metal concentration (mol m"3) = TAMd + TAMp (4-77)
TAMd = dissolved total active metal (mol m"3)
TAMp= particulate total active metal (mol m"3)
KHbmf = dissolved oxygen concentration at which total active metal release is half the anoxic release
rate (g 02 m"3)
BFTAM = anoxic release rate of total active metal (mol m"2 day"1), applied to the bottom layer only
Ktam = effect of temperature on sediment release of total active metal (°C_1)
Ttam = reference temperature for sediment release of total active metal (°C)
WSs = settling velocity of particulate metal (m day"1)
WTAM = external loads of total active metal (mol day"1).
In estuaries, iron and manganese exist in particular and dissolved forms depending on dissolved
oxygen concentration. In the oxygenated water, most of the iron and manganese exist as particulate
while under anoxic conditions, large fractions are dissolved, although solid-phase sulfides and carbonates
exist and may predominate. The partitioning between particulate and dissolved phases is expressed using
4 - EFDC Water Quality Model
4-31
-------
a concept that total active metal concentration must achieve a minimum level, which is a function of
dissolved oxygen, before precipitation occurs:
TAMd = minimum {TAMdmxexpi - Kdotam 1)0) , 7'AM} (4-78)
TAMp = TAM - TAMd (4"79)
TAMdmx = solubility of total active metal under anoxic conditions (mol m"3)
Kdotam = constant that relates total active metal solubility to dissolved oxygen (per g 02 m"3).
4.11 Fecal Coliform Bacteria
Fecal coliform bacteria are indicative of organisms from the intestinal tract of humans and other
animals and can be used as an indicator bacteria as a measure of public health (Thomann and Mueller
1987). In the present model, fecal coliform bacteria have no interaction with other state variables, and
have only one sink term, die-off The kinetic equation, including external loads, may be written as:
dI'CB = - KFCB TFCB7 ~ 20FCB + WFCB (4-80)
dt V
FCB = bacteria concentration (MPN per 100 ml)
KFCB = first order die-off rate at 20 °C (day1)
TFCB = effect of temperature on decay of bacteria (°C_1)
WFCB = external loads of fecal coliform bacteria (MPN per 100 ml m3 day"1).
4.12 Method of Solution
The kinetic equations for the 21 state variables in the EFDC water column water quality model
can be expressed in a 21 x 21 matrix after linearizing some terms, mostly Monod type expressions:
i-[C] = IK] [C] + [if] (4-81)
ot
where [C] is in mass volume"1, [K] is in time"1, and [R] is in mass volume"1 time"1. Since the settling of
particulate matter from the overlying cell acts as an input for a given cell, when Eq. 4-81 is applied to a
cell of finite volume, it may be expressed as:
| [C]<- = [Kl\ [C]t ~ A [K]t [C]t„ ~ [fi]t (4-82)
where the four matrices [C], [Kl], [K2], and [R] are defined in Appendix A of Park et al. (1995). The
subscript k designates a cell at the kth vertical layer. The layer index k increases upward with KC vertical
layers; k = 1 is the bottom layer and k = KC is the surface layer. Then X = 0 for k = KC; otherwise,
4-32
4 - EFDC Water Quality Model
-------
A = 1. The matrix [K2] is a diagonal matrix, and the non-zero elements account for the settling of
particulate matter from the overlying cell.
Equation 4-82 is solved using a second-order accurate trapezoidal scheme over a time step of 0,
which may be expressed as:
[ctf = (m - 1 ([Cf +1{[»]? [C]° + i[K2fk [ct] + («3)
where 0 = 2 m At is the time step for the kinetic equations; [I] is a unit matrix; [C]A = [C]N + [C]°; the
superscripts O and N designate the variables before and after being adjusted for the relevant kinetic
processes. Since Eq. 4-83 is solved from the surface layer downward, the term with [C]k+1A is known for
the klh layer and thus placed on the RHS. In Eq. 4-83, inversion of a matrix can be avoided if the 21 state
variables are solved in a proper order. The kinetic equations are solved in the order of the variables in
the matrix [C] defined in Appendix A of Park et al. (1995).
4.13 Macroalgae (Periphyton) State Variable
The EFDC water quality model was augmented to represent benthic attached algae (often
referred to as macroalgae in estuarine waters and periphyton in fresh waters) using the existing
framework for phytoplankton growth kinetics. Mathematical relationships based on the impacts of
temperature, available light, available nutrients, stream velocity, and density-dependent interactions were
incorporated into the algae growth kinetics framework within EFDC. The major differences between
modeling techniques for attached and free-floating algae are: (1) attached algae are expressed in terms of
areal densities rather than volumetric concentrations; (2) attached algae growth can be limited by the
availability of bottom substrate; (3) the availability of nutrients to the macroalgae matrix can be
influenced by stream velocity; and (4) macroalgae are not subject to hydrodynamic transport. A good
description of periphyton kinetics as it relates to water quality modeling can be found in Warwick et al.
(1997) and has been used to develop this section of the report.
A mass-balance approach was used to model macroalgae growth, with carbon serving as the
measure of standing crop size or biomass. For each model grid cell the equation for macroalgae growth
is slightly different than the one for free-floating algae (Eq. 4-6):
p„ = PM„ Am m u1) ur) fm (4-84)
where
Pmm = maximum growth rate under optimal conditions for macroalgae
f,(N) = effect of suboptimal nutrient concentration (0 < fj < 1)
4 - EFDC Water Quality Model
4-33
-------
f2(I) = effect of suboptimal light intensity (0 < f2 < 1)
f3(T) = effect of suboptimal temperature (0 < f3 < 1)
f4(V) = velocity limitation factor (0 < f4 < 1)
f5(D) = density-dependent growth rate reduction factor (0 < f5 < 1).
The basic growth kinetics for macroalgae were developed from those supplied by EFDC and
others developed by Runke (1985). The macroalgae population as a whole is characterized by the total
biomass present without considering the different species and their associated environmental processes.
The optimum growth for the given temperature is adjusted for light, nutrients, velocity, and density-
dependent limitations. Each growth limitation factor can vary from 0 to 1. A value of 1 indicates the
factor does not limit growth, and a value of 0 means the factor is so severely limiting that growth is
stopped entirely (Bowie et al. 1985).
Stream velocity has a twofold effect on periphyton productivity in freshwater streams: velocity
increases to a certain level to enhance biomass accrual, but further increases result in substantial scouring
(Horner et al. 1990). A benthic algal population is typified as a plant community with an understory and
an overstory. The entire community is called a matrix. As the matrix develops, the periphyton
community is composed of an outer layer of photosynthetically active cells and inner layers of senescent
and decomposing cells. Layering can also develop among different species of periphyton.
Environmental conditions within the matrix are altered by the physical structure of the periphyton. This
influences nutrient uptake and primary production rates of the algae (Sand-Jensen 1983). Above a
certain level, current has a simulating effect on periphyton metabolism by mixing the overlying waters
with nutrient-poor waters that develop around cells (Whitford and Schumacher 1964). The physical
structure of the periphyton community and nutrient uptake by periphyton interfere with nutrient flux
through the microbial matrix (Stevenson and Glover 1993).
Current is constantly scouring periphyton from its substrate. At high enough velocities, shear
stress can result in substantial biomass reduction. Even at low velocities, sudden increases in velocity
raise instantaneous loss rates substantially, but these high rates persist only briefly (Horner et al. 1990).
An increase in velocity above that to which benthic algae are accustomed leads to increased loss rates
and temporarily reduced biomass. However, recolonization and growth after biomass reduction are
usually rapid. The effects of suboptimal velocity upon growth rate are represented in the model by a
velocity limitation function. Two options are available in the model for specifying the velocity
limitation: (1) a Michaelis-Menton (or Monod) equation (4-85) and (2) a five-parameter logistic function
(4-86). The Monod equation limits macroalgae growth due to low velocities, whereas the five-parameter
logistic function can be configured to limit growth due to either low or high velocities (Figure 4-2).
4-34
4 - EFDC Water Quality Model
-------
Velocity limitation option 1, the Michaelis-Menton equation, is written as follows:
m
u
(4-85)
KMV + U
where
U = stream velocity (m/sec)
KMV = half-saturation velocity (m/sec)
Velocity limitation option 2, the five-parameter logistic function is as follows:
m = d - —^—
[ i + (-? r
(4-86)
c
where
U = stream velocity (m/sec)
a = asymptote at minimum x
b = slope after asymptote a
c = x-translation
d = asymptote at maximum x
e = slope before asymptote d
The half-saturation velocity in Eq. 4-85 is the velocity at which half the maximum growth rate occurs.
This effect is analogous to the nutrient limitation because the effect of velocity at suboptimal levels on
periphyton growth is due to increasing the exchange of nutrients between the algal matrix and the
overlying water (Runke 1985). However, this formula can be too limiting at low velocities. This
function does not allow periphyton growth in still waters, but periphyton does grow in still waters such as
lakes. Therefore, the function is applied only at velocities above a minimum threshold level (KMVmin).
When velocities are at or below this lower level, the limitation function is applied at the minimum level.
Above this velocity, the current produces a steeper diffusion gradient around the periphyton (Whitford
and Schumacher 1964). A minimum formulation is used to combine the limiting factors for nitrogen,
phosphorus, and velocity. The most severely limiting factor alone limits periphyton growth. Note that
Eq. 4-86 can be configured so that low velocities are limiting by setting parameter d greater than
parameter a, and viceversa to limit growth due to high velocities. In waters that are rich in nutrients, low
velocities will not limit growth. However, high velocities may cause scouring and detachment of the
macroalgae, resulting in a reduction in biomass. The five-parameter logistic function can be configured
to approximate this reduction by limiting growth at high velocities.
Macroalgae (periphyton) growth can also be limited by the availability of suitable substrate
(Ross 1983). Macroalgae communities reach maximum rates of primary productivity at low levels of
4 - EFDC Water Quality Model
4-35
-------
biomass (Mclntire 1973; Pfeifer and McDiffett 1975). The relationship between standing crop and
production employs the Michaelis-Menton kinetic equation:
/(£>) = — (4-87)
5 KBP + P
m
where
KBP = half-saturation biomass level (g C/m2)
Pm = macroalgae biomass level (g C/m2).
The half-saturation biomass level (KBP) is the biomass at which half the maximum growth rate occurs.
Caupp et al. (1991) used a KBP value of 5.0 g C/m2 (assuming 50% of ash free dry mass is carbon) for a
region of the Truckee River system in California. The function in Eq. 4-87 allows maximum rates of
primary productivity at low levels of biomass with decreasing rates of primary productivity as the
community matrix expands.
Velocity Limitation Function
for Control of Periphyton Growth
1.0 -
0.9
o 08 _
l2 o.7
0 0.6 -
1 0.5
E
Zi 0.4
0.3 -
o
o3 0.2
>
0.1 -
0.0
0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00
Velocity (ft/sec)
Logistic function Monod Function
Figure 4-2. Velocity limitation function for (Option 1) the Monod equation where KMV = 0.25 m/sec and
KMVmin=0.15 m/sec, and (Option 2) the five-parameter logistic function where a= 1.0, b=12.0,
c=0.3, cl=0.35. and c=3.0 (high velocities are limiting).
4-36
4 - EFDC Water Quality Model
-------
5 - EFDC SEDIMENT PROCESS MODEL
A sediment process model developed by DiToro and Fitzpatrick (1993; hereinafter referred to as
D&F) and was coupled with CE-QUAL-ICM for Chesapeake Bay water quality modeling (Cerco and
Cole 1993). The sediment process model was slightly modified and incorporated into the EFDC water
quality model to simulate the processes in the sediment and at the sediment-water interface. The
description of the EFDC sediment process model in this section is from Park et al. (1995). The sediment
process model has 27 water-quality related state variables and fluxes (Table 5-1).
(14) ammonia nitrogen in layer 2
The nitrate state variables, (15), (16), and (22), in the model represent the sum of nitrate and
nitrite nitrogen. The three G classes for particulate organic matter (POM) in Layer 2 and the two layers
for inorganic substances are described below.
In the sediment model, benthic sediments are represented as two layers (Fig. 5-1). The upper
layer (Layer 1) is in contact with the water column and may be oxic or anoxic depending on dissolved
oxygen concentration in the overlying water. The lower layer (Layer 2) is permanently anoxic. The
upper layer depth, which is determined by the penetration of oxygen into the sediments, is at its
maximum only a small fraction of the total depth. Because Hj (~ 0.1 cm) « H2,
Table 5-1. EFDC sediment process model state variables and flux terms
1) particulate organic carbon G1 class in layer 2 (15) nitrate nitrogen in layer 1
2) particulate organic carbon G2 class in layer 2 (16) nitrate nitrogen in layer 2
3) particulate organic carbon G3 class in layer 2 (17) phosphate phosphorus in layer 1
4) particulate organic nitrogen G1 class in layer 2 (18) phosphate phosphorus in layer 2
5) particulate organic nitrogen G2 class in layer 2 (19) available silica in layer 1
6) particulate organic nitrogen G3 class in layer 2 (20) available silica in layer 2
7) particulate organic phosphorus G1 class in layer 2 (21) ammonia nitrogen flux
8) particulate organic phosphorus G2 class in layer 2 (22) nitrate nitrogen flux
9) particulate organic phosphorus G3 class in layer 2 (23) phosphate phosphorus flux
10) particulate biogenic silica in layer 2
11) sulfide/methane in layer 1
12) sulfide/methane in layer 2
13) ammonia nitrogen in layer 1
(24) silica flux
(25) sediment oxygen demand
(26) release of chemical oxygen demand
(27) sediment temperature
H = Hx + H2 ~ H2
(5-1)
5 - EFDC Sediment Process Model
5-1
-------
WATER COLUMN
DEPOSITION
0£
LLI
>-
<
LU
Q
LU
CO
CM
0£
LU
>-
<
SURFACE MASS TRANSFER (s): _r fdj
PARTITIONING: fd
REACTIONS:
SEDIMENTATION (W):
PARTICLE
MIXING A
J2L
J
DIAGENESIS: J
PARTITIONING: fd,
t
REACTIONS: ><2
SEDIMENTATION (W):
1
fPi
DIFFUSION
. (KL)
ct
H
Figure 5-1. Sediment layers and processes included in sediment process model.
where H is the total depth (approximately 10 cm), H, is the upper layer depth and H2 is the lower layer
depth.
The model incorporates three basic processes (Fig. 5-2): (1) depositional flux of POM, (2) the
diagenesis of POM, and (3) the resulting sediment flux. The sediment model is driven by net settling of
particulate organic carbon, nitrogen, phosphorus, and silica from the overlying water to the sediments
(depositional flux). Because of the negligible thickness of the upper layer (Eq. 5-1), deposition is
considered to proceed from the water column directly to the lower layer. Within the lower layer, the
model simulates the diagenesis (mineralization or decay) of deposited POM, which produces oxygen
demand and inorganic nutrients (diagenesis flux). The third basic process is the flux of substances
produced by diagenesis (sediment flux). Oxygen demand, as sulfide (in salt water) or methane (in fresh
water), takes three paths out of the sediments: (1) oxidation at the sediment-water interface as sediment
oxygen demand, (2) export to the water column as chemical oxygen demand, or (3) burial to deep,
5-2
5 - EFDC Sediment Process Model
-------
inactive sediments. Inorganic nutrients produced by diagenesis take two paths out of the sediments: (1)
release to the water column or (2) burial to deep, inactive sediments (Fig. 5-2).
NET SETTLING
OF POM
WATER
NUTRIENT
RELEASE
COD
RELEASE
SEDIMENT
OXYGEN
DEMAND
SEDIMENT
EXPORTED
DIAGENESIS
(DECAY)
OF POM
OXYGEN
DEMAND
~ EXERTED
INORGANIC
NUTRIENTS
^ BURIAL
Figure 5-2. Schematic diagram for sediment process model.
This section describes the three basic processes with reactions and sources/sinks for each state
variable. The method of solution includes finite difference equations, solution scheme, boundary, and
initial conditions. Complete model documentation can be found in D&F (1993).
5.1 Depositional Flux
Deposition is one process that couples the water column model with the sediment model.
Consequently, deposition is represented in both the water column and sediment models. In the water
column model, the governing mass-balance equations for the following state variables contain settling
terms, which represent the depositional fluxes:
• three algal groups, cyanobacteria, diatoms and green algae (Eq. 4-5)
• refractory and labile particulate organic carbon (Equations 4-20 and 4-21)
• refractory and labile particulate organic phosphorus (Equations 4-35 and 4-36) and total
phosphate (Eq. 4-38)
• refractory and labile particulate organic nitrogen (Equations 4-48 and 4-49)
• particulate biogenic silica (Eq. 4-61) and available silica (Eq. 4-62).
5 - EFDC Sediment Process Model
5-3
-------
The sediment model receives these depositional fluxes of particulate organic carbon (POC),
particulate organic nitrogen (PON), particulate organic phosphorus (POP), and particulate biogenic silica
(PSi). Because of the negligible thickness of the upper layer (Eq. 5-1), deposition is considered to
proceed from the water column directly to the lower layer. Since the sediment model has three G classes
of POM, G; (i = 1, 2, or 3), depending on the time scales of reactivity (Section 5.2), the POM fluxes from
the water column should be mapped into three G classes based on their reactivity. Then the depositional
fluxes for the ifc G class (i = 1, 2, or 3) may be expressed as:
Jpoc,i = FCLPi WSu,LPOCn + FCRPi WSrpRPOCn + £ FCBx.WSxBxn (5-2)
x-c,d,g
JPON, = FNLPiWSlpLPONn + FNRPiWSrpRPONn+ £ FNBx,iANCxWSxBx (5-3)
x-c,d,g
Jpop,r = FPLPt • WSjj, LP OP N + FPRPj ¦ WS^ RPOP N + £ FPBx,iAPCWSxBx
x-c,d,g
+ y. ¦ WStssP04P N (5-4)
Jps, - WSJSU*' ~ ~ WSm-SAp" <5"5)
Jpom,i = depositional flux of POM (M = C, N or P) routed into the ith G class (g m"2 day"1)
JPSl = depositional flux of PSi (g Si m"2 day"1)
FCLPj, FNLPj, FPLPj = fraction of water column labile POC, PON, and POP, respectively, routed into
the ith G class in sediment
FCRP, FNRP,. FPRPj = fraction of water column refractory POC, PON, and POP, respectively, routed
into the ith G class in sediment
FCBX1, FNBX1, FPBX1 = fraction of POC, PON, and POP, respectively, in the algal group x routed into
the ith G class in sediment
Y; = 1 for i = 1
0 for i = 2 or 3.
In the source code, the sediment process model is solved after the water column water quality model, and
the calculated fluxes using the water column conditions at t = tn are used for the computation of the water
quality variables at t = tn+ 0. The superscript N indicates the variables after being updated for the kinetic
processes, as defined in Eq. 4-82.
The settling of sorbed phosphate is considered to contribute to the labile G, pool in Eq. 5-4, and
settling of sorbed silica contributes to JPSl in Eq. 5-5 to avoid creation of additional depositional fluxes
5-4
5 - EFDC Sediment Process Model
-------
for inorganic particulates. The sum of distribution coefficients should be unity: FCLP = V| FN LP, =
Xi FPLP = FCRP = FNRP = FPRP = FCBX I = FNBX = FPB = 1. The settling
velocities, WSLP, WSkP. WSX, and WSTSS, as defined in the EFDC water column model (Section 4), are
net settling velocities. If total active metal is selected as a measure of sorption site, WSTSS is replaced by
WSs in Equations 5-4 and 5-5 (see Sections 4.5 and 4.7).
5.2 Diagenesis Flux
Another coupling point of the sediment model to the water column model is the sediment flux,
which is described in Section 5.3. The computation of sediment flux requires that the magnitude of the
diagenesis flux be known. The diagenesis flux is explicitly computed using mass-balance equations for
deposited POC, PON, and POP. (Dissolved silica is produced in the sediments as the result of the
dissolution of PSi. Since the dissolution process is different from the bacterial-mediated diagenesis
process, it is presented separately in Section 5.4.) In the mass-balance equations, the depositional fluxes
of POM are the source terms and the decay of POM in the sediments produces the diagenesis fluxes. The
integration of the mass-balance equations for POM provides the diagenesis fluxes that are the inputs for
the mass-balance equations for ammonium, nitrate, phosphate, and sulfide/methane in the sediments
(Section 5.3).
The difference in decay rates of POM is accounted for by assigning a fraction of POM to various
decay classes (Westrisch and Berner 1984). POM in the sediments is divided into three G classes, or
fractions, representing three scales of reactivity. The G, (labile) fraction has a half life of 20 days, and
the G2 (refractory) fraction has a half life of one year. The G3 (inert) fraction is nonreactive, i.e., it
undergoes no significant decay before burial into deep, inactive sediments. The varying reactivity of the
G classes controls the time scale over which changes in depositional flux will be reflected in changes in
diagenesis flux. If the G, class would dominate the POM input into the sediments, then there would be
no significant time lag introduced by POM diagenesis and any changes in depositional flux would be
readily reflected in diagenesis flux.
Because the upper layer thickness is negligible (Eq. 5-1) and thus depositional flux is considered
to proceed directly to the lower layer (Equations 5-2 to 5-5), diagenesis is considered to occur in the
lower layer only. The mass-balance equations are similar for POC, PON, and POP, and for different G
classes. The mass-balance equation in the anoxic lower layer for the ith G class (i = 1, 2, or 3) may be
expressed as:
5 - EFDC Sediment Process Model
5-5
-------
U POM,i _ _ y . C\T 20 ^ JJ _ ___ J
n 2 POM,i vPOM,i POM,i 12 2 n ^POMJ JPOM,i
Gpovi. = concentration of POM (M = C, N, or P) in the ilh G class in Layer 2 (g m"3)
KpQMi = decay rate of the ith G class POM at 20 °C in Layer 2 (day1)
0poM,i = constant for temperature adjustment for KP0M l
T = sediment temperature (°C)
W = burial rate (m day"1).
Since the G3 class is inert, KP0H3 = 0.
Once the mass-balance equations for Gpom,i and GP0M,2are solved, the diagenesis fluxes are
computed from the rate of mineralization of the two reactive G classes:
2
T = V Y A7 " 20 r M (5-'
M 2s POM,i POM,i POM,i 12 2
2 = 1
JM = diagenesis flux (g m"2 day"1) of carbon (M = C), nitrogen (M = N), or phosphorus (M = P).
5.3 Sediment Flux
The mineralization of POM produces soluble intermediates, which are quantified as diagenesis
fluxes in the previous section. The intermediates react in the oxic and anoxic layers, and portions are
returned to the overlying water as sediment fluxes. Computation of sediment fluxes requires mass-
balance equations for ammonium, nitrate, phosphate, sulfide/methane, and available silica. This section
describes the flux portion for ammonium, nitrate, phosphate, and sulfide/methane of the model.
Available silica is described in Section 5.4.
In the upper layer, the processes included in the flux portion are (Fig. 5-1)
• exchange of dissolved fraction between Layer 1 and the overlying water
• exchange of dissolved fraction between Layer 1 and 2 via diffusive transport
• exchange of particulate fraction between Layer 1 and 2 via particle mixing
• loss by burial to the lower layer (Layer 2)
• removal (sink) by reaction
• internal sources.
Since the upper layer is quite thin, Hj ~ 0.1 cm (Eq. 5-1) and the surface mass transfer coefficient (5) is
on the order of 0.1 m day"1, then the residence time in the upper layer is H/s ~ 10"2 days. Hence, a
5-6 5 - EFDC Sediment Process Model
-------
steady-state approximation is made in the upper layer. Then the mass-balance equation for ammonium,
nitrate, phosphate, or sulfide/methane in the upper layer is:
dCt
Hl = 0 = s(fdQ Ct0 - fdl Ctl) + KL(fd2 Ct2 - fdl Ctl)
dt
2
+ Z)(fp2 Ct2 ~ fpx Ct^) - WCtl - —Ctl + Jx
Ct, & Ct2 = total concentrations in Layer 1 and 2, respectively (g m"3)
Ct0 = total concentration in the overlying water (g m"3)
5 = surface mass transfer coefficient (m day"1)
KL = diffusion velocity for dissolved fraction between Layer 1 and 2 (m day"1)
£) = particle mixing velocity between Layer 1 and 2 (m day"1)
fd0 = dissolved fraction of total substance in the overlying water (0 < fd0 < 1)
fd, = dissolved fraction of total substance in Layer 1 (0 < fd, < 1)
fpi = particulate fraction of total substance in Layer 1 (= 1 - fd,)
fd2 = dissolved fraction of total substance in Layer 2 (0 < fd2 < 1)
fp2 = particulate fraction of total substance in Layer 2 (= 1 - fd2)
Kj = reaction velocity in Layer 1 (m day"1)
Jj = sum of all internal sources in Layer 1 (g m"2 day"1).
The first term on the RHS of Eq. 5-8 represents the exchange across sediment-water interface.
Then the sediment flux from Layer 1 to the overlying water, which couples the sediment model to the
water column model, may be expressed as:
<5-9>
Jaq = sediment flux of ammonium, nitrate, phosphate, or sulfide/methane to the overlying water (g m"2
day"1).
The convention used in Eq. 5-9 is that positive flux is from the sediment to the overlying water.
In the lower layer, the processes included in the flux portion are (Fig. 5-1)
• exchange of dissolved fraction between Layer 1 and 2 via diffusive transport
• exchange of particulate fraction between Layer 1 and 2 via particle mixing
• deposition from Layer 1 and burial to the deep inactive sediments
• removal (sink) by reaction
• internal sources including diagenetic source.
5 - EFDC Sediment Process Model
5-7
-------
The mass-balance equation for ammonium, nitrate, phosphate or sulfide/methane in the lower layer is:
dCU
H2 ^ = " KL{fd2Ct2 - fdxCt^ - Za(jp2-Ct2 - fpxCt^
at
+ W{Ctl - Ct2) - K2 Ct2 + J2 (5-10)
k2 = reaction velocity in Layer 2 (m day"1)
J2 = sum of all internal sources including diagenesis in Layer 2 (g m"2 day"1).
The substances produced by mineralization of POM in sediments may be present in both
dissolved and particulate phases. This distribution directly affects the magnitude of the substance that is
returned to the overlying water. In Equations 5-8 to 5-10, the distribution of a substance between the
dissolved and particulate phases in a sediment is parameterized using a linear partitioning coefficient.
The dissolved and particulate fractions are computed from the partitioning equations:
fdl = fP\ = 1 " fd\ (5"U)
1 + ml-nl 1 1
fd2 = i 1 fP2 = 1 " fd2 (5"12)
1 + m2 ¦ ti2
m,. m2 = solid concentrations in Layer 1 and 2, respectively (kg L"1)
ti1, 7z2 = partition coefficients in Layer 1 and 2, respectively (per kg L"1).
The partition coefficient is the ratio of particulate to dissolved fraction per unit solid concentration (i.e.,
per unit sorption site available).
All terms, except the last two terms, in Equations 5-8 and 5-10 are common to all state variables
and are described in Section 5.3.1. The last two terms represent the reaction and source/sink terms,
respectively. These terms, which take different mathematical formulations for different state variables,
are described in Sections 5.3.2 to 5.3.5 for ammonium, nitrate, phosphate, and sulfide/methane,
respectively.
5.3.1 Common Parameters for Sediment Flux
Parameters that are needed for the sediment fluxes are s, 6), KL, W, H2, m,. m2, ti,, tt2, k,. k2, J,,
and J2 in Equations 5-8 to 5-12. Of these, k,. k2, Jb and J2 are variable-specific. Among the other
common parameters, W, H2, mh and m2, are specified as input. The modeling of the remaining three
parameters, s, 6), and KL, is described in this section.
5-8
5 - EFDC Sediment Process Model
-------
5.3.1.1 Surface mass transfer coefficient. Owing to the observation that the surface mass
transfer coefficient, 5, can be related to the sediment oxygen demand, SOD (DiToro et al. 1990), 5 can be
estimated from the ratio of SOD and overlying water oxygen concentration:
Di SOD
s
Hl DO,
(5-13)
o
D, = diffusion coefficient in Layer 1 (m2 day"1).
Knowing 5, it is possible to estimate the other model parameters.
5.3.1.2 Particulate phase mixing coefficient. The particle mixing velocity between Layer 1
and 2 is parameterized as:
= ^p®dp ^poc,\ DOq (5-14)
^2 GPoc,r KMDp + DO0
Dp = apparent diffusion coefficient for particle mixing (m2 day"1)
0Dp = constant for temperature adjustment for Dp
GP0Cjr = reference concentration for GP0C,i (g C m"3)
KMDp = particle mixing half-saturation constant for oxygen (g 02 m"3).
The enhanced mixing of sediment particles by macrobenthos (bioturbation) is quantified by estimating
Dp. The particle mixing appears to be proportional to the benthic biomass (Matisoff 1982), which is
correlated to the carbon input to the sediment (Robbins et al. 1989). This is parameterized by assuming
that benthic biomass is proportional to the available labile carbon, Gp0C,b and Gpoc,r is the reference
concentration at which the particle mixing velocity is at its nominal value. The Monod-type oxygen
dependency accounts for the oxygen dependency of benthic biomass.
It has been observed that a hysteresis exists in the relationship between the bottom water oxygen
and benthic biomass. Benthic biomass increases as the summer progresses. However, the occurrence of
anoxia/hypoxia reduces the biomass drastically and also imposes stress on benthic activities. After full
overturn, the bottom water oxygen increases, but the population does not recover immediately. Hence,
the particle mixing velocity, which is proportional to the benthic biomass, does not increase in response
to the increased bottom water oxygen. Recovery of benthic biomass following hypoxic events depends
on many factors including severity and longevity of hypoxia, constituent species, and salinity (Diaz and
Rosenberg 1995).
5 - EFDC Sediment Process Model
5-9
-------
This phenomenon of reduced benthic activities and hysteresis is parameterized based on the idea
of stress that low oxygen imposes on the benthic population. It is analogous to the modeling of the toxic
effect of chemicals on organisms (Mancini 1983). A first order differential equation is employed, in
which the benthic stress (1) accumulates only when overlying oxygen is below KMDp and (2) is
dissipated at a first order rate (Fig. 5-3a):
— = - KstST +
dt
dST_
A. rrrp )J i
dt ST
DO„
KM,
'DP.
if DOr. < KM
Dp
(5-15)
if DO0 > KM,
Dp
ST = accumulated benthic stress (day)
Kst = first order decay rate for ST (day1).
The behavior of this formulation can be understood by evaluating the steady-state stresses at two extreme
conditions of overlying water oxygen, DO0:
as DO0 = 0 Kst ST=1 f(ST) = (1 - KstST) = 0
as DO0 > KMDp Kst ST = 0 f(ST) = (1 - Kst ST) = 1
The dimensionless expression, f(ST) = 1 - KST ST, appears to be the proper variable to quantify the effect
of benthic stress on benthic biomass and thus particle mixing (Fig. 5-3b).
The final formulation for the particle mixing velocity, including the benthic stress, is:
Z) = ,)p Q"p GPOC+ D°0 + ^min (5-16)
h2 gpoqr kmDp - do0jk h2
Dpmin = minimum diffusion coefficient for particle mixing (m2 day"1).
The reduction in particle mixing due to the benthic stress, f(ST), is estimated by employing the following
procedure. The stress, ST, is normally calculated with Eq. 5-15. Once DO0 drops below a critical
concentration, DOST c, for NChypoxm consecutive days or more, the calculated stress is not allowed to
decrease until tV|,;s days of DO0 > DOST c. That is, only when hypoxic days are longer than critical
hypoxia days (NChypoxm), the maximum stress, or minimum (1 - Kst ST), is retained for a specified period
(Imbs days) after DO0 recovery (Fig. 5-3). No hysteresis occurs if DO0 does not drop below DOST c or if
hypoxia lasts less than NChypoxm days. When applying maximum stress for tM,;s days, the subsequent
hypoxic days are not included in t^g. This parameterization of hysteresis essentially assumes seasonal
5-10
5 - EFDC Sediment Process Model
-------
hypoxia, i.e., one or two major hypoxic events during summer, and might be unsuitable for systems with
multiple hypoxic events throughout a year.
$ 15'
Ld
a:
i—
(/>
O 1°-
KM„P = 4.0 g 02 m~3
DO5T.C = 3.0 g 02 m~3
Ks, = 0.03 day"1
< (MBS ~
(a)
/ \
hysteresis \
\ \
\ \
\ \
\ \
\ \
\ \
D00 (g rrf3) /
\ \
\ \
\ \
\ \
\
\
S
\ /
/ \ \
/ s \
\ / ¦/ ^ \
7 / ^
-------
(bio-irrigation). This is modeled by increasing the diffusion coefficient relative to the molecular
diffusion coefficient:
KL . z±_^£i— + nBiBTt0 (5-.7)
"2
Dd = diffusion coefficient in pore water (m2 day"1)
0Dd = constant for temperature adjustment for Dd
RBi bt = ratio of bio-irrigation to bioturbation.
The last term in Eq. 5-17 accounts for the enhanced mixing by organism activities.
5.3.2 Ammonia Nitrogen
Diagenesis is assumed not to occur in the upper layer because of its shallow depth, and
ammonium is produced by diagenesis in the lower layer:
Jxm4 = 0 J%NH4 = JN (from Eq. 5-7) (5-18)
Ammonium is nitrified to nitrate in the presence of oxygen. A Monod-type expression is used for the
ammonium and oxygen dependency of the nitrification rate. Then the oxic layer reaction velocity in
Eq. 5-8 for ammonium may be expressed as:
2 _ DO0 KMNH4 2 rJ- 20
1 ,nh4 ~ , 777T "7777 77777" knh4 vnh4 v '
2 'KMNH4,02 + DO0 KMNH4 + NH4\
and then the nitrification flux becomes:
j = Ki,nh4 (5-20)
J Nit
S
KMv||4 o2 = nitrification half-saturation constant for dissolved oxygen (g 02 m"3)
NH4, = total ammonium nitrogen concentration in Layer 1 (g N m"3)
KM,, 14 = nitrification half-saturation constant for ammonium (g N m"3)
Kv,|4 = optimal reaction velocity for nitrification at 20 °C (m day"1)
0NH4 = constant for temperature adjustment for Kv,,,
JNlt = nitrification flux (g N m"2 day"1).
Nitrification does not occur in the anoxic lower layer:
K =0 (5-21)
k2,NH4 v v '
5-12
5 - EFDC Sediment Process Model
-------
Once Equations 5-8 and 5-10 are solved for NH4j and NH42, the sediment flux of ammonium to
the overlying water, JaqNH4, can be calculated using Eq. 5-9. Note that it is not NH4j and NH42 that
determine the magnitude of JaqNH4 (Section X-B-2 in D&F 1993). The magnitude is determined by
(1) the diagenesis flux, (2) the fraction that is nitrified, and (3) the surface mass transfer coefficient (5)
that mixes the remaining portion.
5.3.3 Nitrate Nitrogen
Nitrification flux is the only source of nitrate in the upper layer, and there is no diagenetic source
for nitrate in both layers:
J\,N03 = JNit {fl'<)m E(l- 5"19) J2,NOS = 0 (5"22)
Nitrate is present in sediments as dissolved substance, i.e., *n;1=N03 = ^.ncb = making fdljN03 = fd2jN03 = 1
(Equations 5-11 and 5-12): it also makes £) meaningless, hence 6) = 0. Nitrate is removed by
denitrification in both oxic and anoxic layers with the carbon required for denitrification supplied by
carbon diagenesis. The reaction velocities in Equations 5-8 and 5-10 for nitrate may be expressed as:
2 2 aT - 20 f5-23>>
K1 ,N03 ~ KN03,l'®N03
k = k -a7" 20 (5-24)
K2,N03 KN03,2 N03
and the denitrification flux out of sediments as a nitrogen gas becomes:
2
j _ Ki,M93 Kip? + K NO3 (5-25)
N2{g) J\UJl 2JV03 2
kN03 i = reaction velocity for denitrification in Layer 1 at 20 °C (m day"1)
kN03 2 = reaction velocity for denitrification in Layer 2 at 20 °C (m day"1)
0NO3 = constant for temperature adjustment for kN03j1 and kN03 2
JN2(g) = denitrification flux (g N m"2 day"1)
N03j = total nitrate nitrogen concentration in Layer 1 (g N m"3)
N032 = total nitrate nitrogen concentration in Layer 2 (g N m"3).
Once Equations 5-8 and 5-10 are solved for N03j and N032, the sediment flux of nitrate to the
overlying water, JaqN03, can be calculated using Eq. 5-9. The steady-state solution for nitrate showed that
the nitrate flux is a linear function of NO30 (Eq. 111-15 in D&F 1993): the intercept quantifies the amount
of ammonium in the sediment that is nitrified but not denitrified (thus releases as JaqN03), and the slope
quantifies the extent to which overlying water nitrate is denitrified in the sediment. It also revealed that
5 - EFDC Sediment Process Model 5-13
-------
if the internal production of nitrate is small relative to the flux of nitrate from the overlying water, the
normalized nitrate flux to the sediment, - JaqNO3/NO30, is linear in 5 for small 5 and constant for large .v
(Section III-C in D&F 1993). For small 5 (~ 0.01 m day"1), Hj is large (Eq. 5-13) so that oxic layer
denitrification predominates and JaqN03 is essentially zero independent of NO30 (Fig. III-4 in D&F 1993).
5.3.4 Phosphate Phosphorus
Phosphate is produced by the diagenetic breakdown of POP in the lower layer:
J\,P04 = 0 J2,P04 = JP Vr0m E(l- 5"7) (5"26)
A portion of the liberated phosphate remains in the dissolved form and a portion becomes particulate
phosphate, either via precipitation of phosphate-containing minerals (Troup 1974), e.g., vivianite,
Fe3(P04)2(s), or by partitioning to phosphate sorption sites (Lijklema 1980; Barrow 1983; Giordani and
Astorri 1986). The extent of particulate formation is determined by the magnitude of the partition
coefficients, tc1jPo4 and h2,po4, in Equations 5-11 and 5-12. Phosphate flux is strongly affected by DO0,
the overlying water oxygen concentration. As DO0 approaches zero, the phosphate flux from the
sediments increases. This mechanism is incorporated by making tt1)P04 larger, under oxic conditions, than
tc2,po4- hi the model, when DO0 exceeds a critical concentration, (DOi:i)CIIL|„,4. sorption in the upper layer
is enhanced by an amount A^p^:
^\J-04 = ll2,P04'^yKP04,\) DO0 > (PO0)critj,O4 (5-27)
When oxygen falls below (DO0)cntpO4, then:
_ = _ ,/a_ DO < (DO) „ (5-28)
1 ,P04 2,P04 ^ PO4,0 0 0 'crit,P04
which smoothly reduces tt1)P04 to Tt: p04 as DO0 goes to zero. There is no removal reaction for phosphate
in both layers:
K = K =0 (5 -29)
K1 ,P04 k2 ,P04 w v '
Once Equations 5-8 and 5-10 are solved for P04j and P042, the sediment flux of phosphate to
the overlying water, JaqP04, can be calculated using Eq. 5-9.
5.3.5 Sulfide/Methane and Oxygen Demand
5.3.5.1 Sulfide. No diagenetic production of sulfide occurs in the upper layer. In the lower
layer, sulfide is produced by carbon diagenesis (Eq. 5-7) decremented by the organic carbon consumed
by denitrification (Eq. 5-25). Then:
5-14
5 - EFDC Sediment Process Model
-------
/ =0 / = a ¦ J - a J (5-30)
J\,H2S u 2,H2S 02,C JC U02JJ03 JN2(g) v '
ao2,c = stoichiometric coefficient for carbon diagenesis consumed by sulfide oxidation (2.6667 g 02-
equivalents per g C)
a02,N03 = stoichiometric coefficient for carbon diagenesis consumed by denitrification (2.8571 g 02-
equivalents per g N).
A portion of the dissolved sulfide that is produced in the anoxic layer reacts with the iron to form
particulate iron monosulfide, FeS(s) (Morse et al. 1987). The particulate fraction is mixed into the oxic
layer where it can be oxidized to ferric oxyhydroxide, Fe203(s). The remaining dissolved fraction also
diffuses into the oxic layer where it is oxidized to sulfate. Partitioning between dissolved and particulate
sulfide in the model represents the formation of FeS(s), which is parameterized using partition
coefficients, tt1H2s and n2s, in Equations 5-11 and 5-12.
The present sediment model has three pathways for sulfide, the reduced end product of carbon
diagenesis: (1) sulfide oxidation, (2) aqueous sulfide flux, and (3) burial. The distribution of sulfide
among the three pathways is controlled by the partitioning coefficients and the oxidation reaction
velocities (Section V-E in D&F 1993). Both dissolved and particulate sulfide are oxidized in the oxic
layer, consuming oxygen in the process. In the oxic upper layer, the oxidation rate that is linear in
oxygen concentration is used (Cline and Richards 1969; Millero 1986; Boudreau 1991). In the anoxic
lower layer, no oxidation can occur. Then the reaction velocities in Equations 5-8 and 5-10 may be
expressed as:
2 _ I 2 ,, 2 r - 20 £^0 in
K\,H2S ~ ViH2S,dl'Ja\,H2S + KH2S,pl'JPlJ{2S K3H2S - ™ 4 U U
2 • KMH2S02
K2 ,H2S = 0 ('5"32')
KH2s,di = reaction velocity for dissolved sulfide oxidation in Layer 1 at 20 °C (m day"1)
kH2s,pi = reaction velocity for particulate sulfide oxidation in Layer 1 at 20°C (m day"1)
0H2S = constant for temperature adjustment for KH2Sdl and kH2S pl
KMH2S 02 = constant to normalize the sulfide oxidation rate for oxygen (g 02 m"3).
The constant, KMH2S 02, which is included for convenience only, is used to scale the oxygen
concentration in the overlying water. At DO0 = KMH2Sj02, the reaction velocity for sulfide oxidation rate
is at its nominal value.
5 - EFDC Sediment Process Model
5-15
-------
The oxidation reactions in the oxic upper layer cause oxygen flux to the sediment, which exerts
SOD. By convention, SOD is positive: SOD = -Jaq02- The SOD in the model consists of two
components, carbonaceous sediment oxygen demand (CSOD) due to sulfide oxidation and nitrogenous
sediment oxygen demand (NSOD) due to nitrification:
2
SOD = CSOD + NSOD = H2Sl + aQ2NH4 'JNit (5~3
H2S, = total sulfide concentration in Layer 1 (g 02-equivalents m"3)
ao2,NH4 = stoichiometric coefficient for oxygen consumed by nitrification (4.33 g 02 per g N).
Equation 4-29 is nonlinear for SOD because the RHS contains 5 (= SOD/DO0) so that SOD appears on
both sides of the equation: note that JNlt (Eq. 5-20) is also a function of 5. A simple back substitution
method is used, as explained in Section 5.6.1.
If the overlying water oxygen is low, then the sulfide that is not completely oxidized in the upper
layer can diffuse into the overlying water. This aqueous sulfide flux out of the sediments, which
contributes to the chemical oxygen demand in the water column model, is modeled using
- s(fdhH2S H2S, - COD) (5-34)
The sulfide released from the sediment reacts very quickly in the water column when oxygen is available,
but can accumulate in the water column under anoxic conditions. The COD, quantified as oxygen
equivalents, is entirely supplied by benthic release in the water column model (Eq. 3-16). Since sulfide
also is quantified as oxygen equivalents, COD is used as a measure of sulfide in the water column in
Eq. 5-34.
5.3.5.2 Methane. When sulfate is used up, methane can be produced by carbon diagenesis and
methane oxidation consumes oxygen (DiToro et al. 1990). Owing to the abundant sulfate in the
saltwater, only the aforementioned sulfide production and oxidation are considered to occur in the
saltwater. Since the sulfate concentration in fresh water is generally insignificant, methane production is
considered to replace sulfide production in fresh water. In fresh water, methane is produced by carbon
diagenesis in the lower layer decremented by the organic carbon consumed by denitrification, and no
diagenetic production of methane occurs in the upper layer (Eq. 5-30):
J\,CH4 ~ 0 Jl,CH4 ~ a02,C ^C ~ a02,N03'^N2{g) (5-35_
5-16
5 - EFDC Sediment Process Model
-------
The dissolved methane produced takes two pathways: (1) oxidation in the oxic upper layer causing
CSOD or (2) escape from the sediment as aqueous flux or as gas flux:
J2 ,CH4 = CSOD + Jaq£H4 + JcH4(g) (5-36)
Jaq CH4 = aqueous methane flux (g 02-equivalents m"2 day"1)
JCH4(g) = gaseous methane flux (g 02-equivalents m"2 day"1).
A portion of dissolved methane that is produced in the anoxic layer diffuses into the oxic layer
where it is oxidized. This methane oxidation causes CSOD in the freshwater sediment (DiToro et al.
1990):
CSOD - CSODmax
I \
r\T - 20
1 - sech\ K' :"4 CH4 ]
(5-37)
CSODmax = minimum\j2 KL ¦ CH4mt -J%CH4, J2 r/I4} (5"38)
1.02420 ~ T (5~39)
CH4 , = 100
sat
h +
1 + ^
10
CSODmax = maximum CSOD occurring when all the dissolved methane transported to the oxic layer is
oxidized
kCH4 = reaction velocity for dissolved methane oxidation in Layer 1 at 20°C (m day"1)
6ch4 = constant for temperature adjustment for kCH4
CH4sat = saturation concentration of methane in the pore water (g 02-equivalents m"3).
The term, (h + H2)/10 where h and H2 are in meters, in Eq. 5-39 is the depth from the water surface that
corrects for the in situ pressure. Equation 5-39 is accurate to within 3% of the reported methane
solubility between 5 and 20°C (Yamamoto et al. 1976).
If the overlying water oxygen is low, the methane that is not completely oxidized can escape the
sediment into the overlying water either as aqueous flux or as gas flux. The aqueous methane flux, which
contributes to the chemical oxygen demand in the water column model, is modeled using (DiToro et al.
1990):
A7 " 20
J^ch, = CSODm^ech[Kcm cm ] = CSOD^ - CSOD <5"40>
5 - EFDC Sediment Process Model
5-17
-------
Methane is only slightly soluble in water. If its solubility, CH4sat given by Eq. 5-39, is exceeded in the
pore water, it forms a gas phase that escapes as bubbles. The loss of methane as bubbles, i.e., the
gaseous methane flux, is modeled using Eq. 5-36 with J2 CH4 from Eq. 5-35, CSOD from Eq. 5-37, and
JaqCH4 from Eq. 5-40 (DiToro et al. 1990).
5.4 Silica
The production of ammonium, nitrate, and phosphate in sediments is the result of the
mineralization of POM by bacteria. The production of dissolved silica in sediments is the result of the
dissolution of particulate biogenic or opaline silica, which is thought to be independent of bacterial
processes.
The depositional flux of particulate biogenic silica from the overlying water to the sediments is
modeled using Eq. 5-5. With this source, the mass-balance equation for particulate biogenic silica may
be written as:
"'Ss'"2 ~ WPSi + Jps< " Jds< <5"41)
PSi = concentration of particulate biogenic silica in the sediment (g Si m"3)
SSl = dissolution rate of PSi in Layer 2 (g Si m"3 day"1)
JPSl = depositional flux of PSi (g Si m"2 day"1) given by Eq. 5-5
JDSl = detrital flux of PSi (g Si m"2 day"1) to account for PSi settling to the sediment that is not associated
with the algal flux of biogenic silica.
The processes included in Eq. 5-41 are dissolution (i.e., production of dissolved silica), burial, and
depositional and detrital fluxes from the overlying water. Equation 5-41 can be viewed as the analog of
the diagenesis equations for POM (Eq. 5-6). The dissolution rate is formulated using a reversible
reaction that is first order in silica solubility deficit and follows a Monod-type relationship in particulate
silica:
ss, --
PSi + KMpSi
KSl = first order dissolution rate for PSi at 20°C in Layer 2 (day1)
0Sl = constant for temperature adjustment for KSl
KMPSl = silica dissolution half-saturation constant for PSi (g Si m"3)
Sisat = saturation concentration of silica in the pore water (g Si m"3).
5-18
5 - EFDC Sediment Process Model
-------
The mass-balance equations for mineralized silica can be formulated using the general forms,
Equations 5-8 and 5-10. There is no source/sink term and no reaction in the upper layer:
Jl,S: = Kl,S: = 0 (5"43)
In the lower layer, silica is produced by the dissolution of particulate biogenic silica, which is modeled
using Eq. 5-42. The two terms in Eq. 5-42 correspond to the source term and reaction term in Eq. 5-10:
J2S. = Ks • 0J" 20 — Si t H2 (5-44)
5' 5' PSi + KMm sat 2
k2,=K,-QT~2° — fw'H, (5-45)
s' Sl PSi + KMpSi 2
A portion of silica dissolved from particulate silica sorbs to solids and a portion remains in the
dissolved form. Partitioning using the partition coefficients, 7t1=si and 7t2 Si, in Equations 5-11 and 5-12
controls the extent to which dissolved silica sorbs to solids. Since silica shows similar behavior as
phosphate in the adsorption-desorption process, the same partitioning method as applied to phosphate
(Section 5.3.4) is used for silica. That is, when DO0 exceeds a critical concentration, (DO0)cnt Sl, sorption
in the upper layer is enhanced by an amount Att:su:
71
I Si
n2Si'(^nsui) DO, > (DO0)crjSj (5-46)
When oxygen falls below (DO0)cntSl, then:
Ku, = DO, <- (D0„)cn,si (5-47)
which smoothly reduces 7t1>si to 7t2=si as DO0 goes to zero.
Once Equations 5-8 and 5-10 are solved for Si2 and Si2, the sediment flux of silica to the overlying
water, Jaq Sl, can be calculated using Eq. 5-9.
5.5 Sediment Temperature
All rate coefficients in the aforementioned mass-balance equations are expressed as a function of
sediment temperature, T. The sediment temperature is modeled based on the diffusion of heat between
the water column and sediment:
— = —{Tw - T) (5-48)
dt H2
Dt = heat diffusion coefficient between the water column and sediment (m2 sec"1)
5 - EFDC Sediment Process Model
5-19
-------
Tw = temperature in the overlying water column (°C) calculated by Eq. 4-82.
The model application in D&F and Cerco and Cole (1993) used DT = 1.8 x 10 m sec .
5.6 Method of Solution
5.6.1 Finite-Difference Equations and Solution Scheme
An implicit integration scheme is used to solve the governing mass-balance equations. The finite
difference form of Eq. 5-8 may be expressed as:
0 = stfd,-al -fdxa[) * KL(fd2a{ - fdxa[) * o2' o,')
/ Ki / /
w ct[ - —ct[ + jI
s
(5-49)
where the primed variables designate the values evaluated at i $and the unprimed variables are those at
t, where 0 is defined in Eq. 4-82. The finite difference form of Eq. 5-10 may be expressed as:
0 = - Kl,(jd2Gi!1 - fdxCt[) - Zi()fp2Ct2 ~ fpxCt[) + W(Ci[ - Ct2)
Hi /
(k2 + ^Ct2 +
/ 2
j' + — a,
2 0 2
\
(5-50)
The two terms, - (H2/0)Ct2' and (H2/0)Ct2, are from the derivative term, H2(c'Ct2/ct) in Eq. 5-10, each of
which simply adds to the Layer 2 removal rate and the forcing function, respectively. Setting these two
terms equal to zero results in the steady-state model. The two unknowns, Ct/ and Ct2', can be calculated
at every time step using:
\
2
1
sfdl + ax + —
-a.
-a„
a2 + W + k2 +
0j
' A
Ct[
Ct'
Jx + sfd0Ct0
H,
A +
0
-Cl
(5-51)
ax = KLfdx + 1a-fpx + W
an
KL-fd, + ^fp2
(5-52)
The solution of Eq. 5-51 requires an iterative method since the surface mass transfer coefficient, s, is a
function of the SOD (Eq. 5-13), which is also a function of s (Eq. 5-33). A simple back substitution
method is used:
(1) Start with an initial estimate of SOD: for example, SOD = a02 C'Jc or the previous time step SOD.
(2) Solve Eq. 5-51 for ammonium, nitrate, and sulfide/methane.
5-20
5 - EFDC Sediment Process Model
-------
(3) Compute the SOD using Eq. 5-33.
(4) Refine the estimate of SOD: a root finding method (Brent's method in Press et al. 1986) is used
to make the new estimate.
(5) Go to (2) if no convergence.
(6) Solve Eq. 5-51 for phosphate and silica.
For the sake of symmetry, the equations for diagenesis, particulate biogenic silica and sediment
temperature are also solved in implicit form. The finite difference form of the diagenesis equation
(Eq. 5-6) may be expressed as:
G
/
POMJ
r JL t
POMJ + it POMJ
"2
A V1
1 A TS~ aT ~ 20 V/ jjr
1 + d-KpOMj-dPOm + — W
"2
(5-53)
The finite difference form of the PSi equation (Eq. 5-41) may be expressed as:
PSi' =
PSi + „ (Jpsi + ^ds)
H2
\
1 + ®KSldSl
T- 20 Si sat fd2,Si'Sil
+ — W
PSi + KMpSi H2
-1
(5-54)
using Eq. 5-36 for the dissolution term, in which PSi in the Monod-type term has been kept at time level t
to simplify the solution. The finite difference form of the sediment temperature equation (Eq. 5-48) may
be expressed as:
( r, \{ \ -1
r =
T +
_0_
H2
D T
^t 1 w
1 +
_0_
H7
Dn
(5-55)
5.6.2 Boundary and Initial Conditions
The above finite difference equations constitute an initial boundary-value problem. The boundary
conditions are the depositional fluxes (JPOm,i and JPSl) and the overlying water conditions (Ct0 and Tw) as a
function of time, which are provided from the water column water quality model. The initial conditions
are the concentrations at t = 0, GPOM,i(0), PSi(O), Ct,(0). Ct2(0), and T(0), to start the computations.
Strictly speaking, these initial conditions should reflect the past history of the overlying water conditions
and depositional fluxes, which is often impractical because of lack of field data for these earlier years.
5 - EFDC Sediment Process Model
5-21
-------
This page intentionally left blank
5-22
5 - EFDC Sediment Process Model
-------
6 - DATABASES
6.1 Introduction
In general, historical water quality data sets for specific waterbodies are typically scattered in
among numerous federal and state government agencies, universities, and the private sector. Since
specific monitoring programs often have differing objectives and sources of funding, coordination of the
results of research and monitoring programs into centralized computer databases for on-line retrieval and
analysis generally does not exist. If historical data sets are not readily accessible in database
management systems, however, comprehensive integrating analyses, hypothesis testing, and model-
building efforts using existing data are very costly, if not impossible, tasks. In the absence of
environmental evaluations based on such comprehensive data sets, resource and management decisions,
with a larger degree of uncertainty, are often made using only a limited portion of the existing available
data.
The increasing scientific complexity and public cost implications of aquatic resource
management issues related to waste disposal practices in freshwater and estuarine ecosystems thus
require the use of large data sets and sophisticated methods, including models, for credible scientific
evaluations of public policy options. The technical credibility of a site-specific modeling framework is,
in fact, largely determined by the extent of the agreement between model simulation results and observed
data sets. Comparison of model results with the observed database provides the only benchmark that is
available to test the model under existing loading conditions. Following satisfactory testing of the model
against existing hydrologic, hydrographic, and loading conditions, a properly validated model can then be
used to evaluate the probable water quality impact based on management controls that alter external
loading conditions. This section of the report documents the sources of historical data used in the
development of the Christina River Basin hydrodynamic and water quality model.
6.2 Bathymetric and Stream Geometry Data
The Christina River Basin EFDC model incorporates both tidal and nontidal waterbodies into its
framework. The tidal portions of the model include the Delaware River from Reedy Point on the south to
Marcus Hook on the north, as well as tidal portions of the Christina River, lower Brandywine Creek, and
the lower White Clay Creek. Bathymetric data were available for the Delaware River and the mouth of
Christina River from the NOAA Geophysical Data System for Hydrographic Survey Data (GEODAS)
CD-ROM. GEODAS contains historical hydrographic survey data from 1850 to 1994. Bathymetry data
in the tidal Christina River were estimated from depth contours shown on the 7.5-minute USGS
quadrangle map for Wilmington South. Shoreline information was obtained from the 1:100,000 Digital
Line Graph CD-ROM published by the USGS.
6 - Databases
6-1
-------
Cross-section and bottom elevation data for the nontidal streams in the EFDC model were
obtained from several sources, including the Federal Emergency Management Agency (FEMA), New
Castle County Water Resources Agency, and 7.5-minute quadrangle maps of the area. Input data sets for
HEC-2 model runs used to develop flood insurance rate maps were obtained from FEMA. Unfortunately,
these data were available only in hard copy "print out" format. The bottom elevations and channel
widths from the HEC-2 printouts were used to develop channel geometry for certain stream reaches. For
other stream reaches lacking HEC-2 information, estimates of channel geometry and bottom elevation
were made from the 7.5-minute USGS quadrangle maps.
6.3 Tide Data
Long-term tide measurements at 15-minute intervals were recorded by the USGS at the Port of
Wilmington (station 01481602) near the mouth of the Christina River and at Newport (station 01480065)
about 11.4 km (7.0 miles) upstream from the mouth of Christina River. Time-series data files including
the calibration period May 1997 to September 1997 were provided by DNREC for both tide gage
locations. These time-series data were used to assess the hydrodynamic calibration of the tidal estuary
portion of the EFDC model.
6.4 Climatology Data
Meteorological data for Wilmington (WBAN station 13781) were obtained from the National
Climatic Data Center. These data included both daily and hourly summaries of atmospheric pressure, air
temperature, relative humidity, wind speed, wind direction, and precipitation. In addition, daily values of
maximum, minimum, and average air temperature, rainfall, and solar radiation measured at the Stroud
Water Research Center near Avondale were provided by Dr. John Davis. The daily solar radiation data
were used to develop hourly values assuming a sine function distribution from sunrise to sunset. The
climatology data were used to develop the meteorological conditions for the study time period (May 1 to
September 21, 1997).
6.5 Stream Flow Data
A number of long-term USGS stream gaging stations are found in the Christina River Basin (see
Figure 6-1). Statistical analyses were performed on these stations to determine the average flow,
harmonic mean flow, and 7Q10 flow rates. The 7Q10 flow rate will be used for the low-flow TMDL
analysis. A summary of the flow statistics is provided in Table 6-1. The daily average flow rates from
the gages were used to estimate flow contributions from each of the 39 HSPF watersheds in the study
area. The daily flows from each watershed were then distributed to the appropriate EFDC model grid
cell.
6-2
6 - Databases
-------
6.6 In-stream Water Quality Monitoring Data
The primary source of water-column water quality data was from EPA's STORET system, which
contains data collected and archived by various agencies including the USGS, Delaware DNREC, and
Pennsylvania DEP. Water quality monitoring data for the basin were downloaded from STORET for the
period 1980 to 1998. Characterization of both the estuary and stream water quality was required for the
development of a model that could be used for TMDL analysis. In August 1997, a detailed water quality
field survey was undertaken by Dr. John Davis for PADEP and DNREC at four locations: (1) Red Clay
Creek near Kennett Square, (2) White Clay Creek near Avondale, (3) East Branch Brandywine Creek
near Downingtown, and (4) West Branch Brandywine Creek near Coatesville. A detailed description of
the August 1997 field surveys can be found in Davis (1998).
Data from STORET were compiled into a comprehensive and flexible database (dBASE III
format) to characterize the spatial and temporal water quality trends of the Christina River Basin study
area. The water quality database was used to (1) prepare tributary loads (nonpoint source loads),
(2) develop boundary conditions for the ocean boundary of the model, and (3) compile time-series and
longitudinal transect data sets for comparison to model results.
The locations of the water quality stations obtained from STORET are shown in Figure 6-2. A
computer data management system was developed for analyzing the observed water quality information
and for comparison to model results. The approach was based on the use of a common reference to
latitude, longitude, time, and depth in both the real-world observations and the model results. The data
management system provided a common link between the field monitoring data and the EFDC model
output that facilitated the automation of model-data comparisons. A total of 44 parameter fields were
included in the Christina River Basin database file (see Table 6-2). The database consisted of more than
40,000 records.
6.7 Discharge Monitoring Data for Point Sources
Discharge Monitoring Records (DMRs) for various point sources in the Brandywine Creek
watershed were provided in hard copy form by the Brandywine Valley Association. Other DMRs were
provided in electronic format by PADEP and DNREC. The hard-copy data sheets covered the period
1993 to 1997. The hard-copy data were keypunched and the electronic data were reformatted into a
database file for use in developing point source loads for the water quality model. The database file
(dBASE III format) contained 30 fields as described in Table 6-3. A list of all 120 NPDES discharges
included in the model is given in Section 7 (see Table 7-6) and the locations are shown in Figure 6-3.
The August 1997 study (Davis 1998) included seven NPDES discharges that were monitored for flow
and water quality parameters (see Figure 6-4). Loading values for the various water quality constituents
6 - Databases
6-3
-------
were computed based on the flow rates and concentrations provided on the DMRs or measured during the
August 1997 study.
The NPDES discharges included 19 single residence discharges (SRD) that are not required to
submit DMR data. For purposes of model calibration, it was assumed that these SRD discharges
operated at their permit discharge limits. Characteristic concentrations for the various water quality
parameters were then assigned to the NPDES source based on the type of discharge, and the loading in
kg/day for each constituent was computed for input to the EFDC model. The characteristic effluent
concentrations used for this study are listed in Table 6-4, and the characteristic effluent parameter ratios
are listed in Table 6-5. The characteristic effluent concentrations and parameter ratios were derived from
effluent monitoring data collected by Davis (1998) in August 1998 and from literature values reported in
the Technical Guidance Manual for Developing TMDLs (USEPA 1995).
6-4
6 - Databases
-------
Table 6-1. Flow statistics for stream gages in Christina River Basin (cfs).
USGS
Gage ID
Drainage
Area (mi2)
Years of
Record
Average
Flow
Harmonic
Mean
7Q10
Flow
1Q10
Flow
7Q1
Flow
1Q1
Flow
01478000
20.5
1944-94
28.21
8.31
1.53
0.54
3.79
1.83
01478500
66.7
1952-79
85.91
47.10
11.00
10.15
24.05
22.38
01478650
1994
38.66
01479000
89.1
1932-94
114.65
62.19
15.60
14.04
31.23
28.45
01479820
1989-96
24.69
01480000
47.0
1944-94
63.39
36.51
10.25
8.91
18.38
16.37
01480015
1990-94
41.08
01480300
18.7
1961-96
26.25
12.83
3.40
3.01
6.62
6.19
01480500
45.8
1944-96
66.33
34.64
8.24
7.34
15.41
14.21
01480617
55.0
1970-96
91.31
52.79
19.02
15.54
24.84
21.63
01480650
6.2
1967-68
6.00
3.51
01480665
33.4
1967-68
36.36
23.45
01480700
60.6
1966-96
93.46
50.53
13.86
12.17
21.84
19.87
01480800
81.6
1959-68
86.63
44.81
12.56
11.86
20.57
18.81
01480870
89.9
1972-96
153.43
87.17
28.44
23.62
37.66
34.63
01481000
287.0
1912-96
395.13
234.13
70.63
65.04
117.01
107.14
01481500
314.0
1947-94
477.01
266.73
78.13
71.96
123.45
113.32
6 - Databases
6-5
-------
Table 6-2. File structure of the water quality monitoring database.
Number of data records: 30422
Field Field Name Type Width
1
SOURCE
Character
10
2
OTHERID
Character
10
3
STATION
Character
15
4
DATE
Date
8
5
TIME
Character
4
6
LAT
Numeric
10
7
LON
Numeric
10
8
BOTTOM
Numeric
7
9
DEPTH
Numeric
7
10
TEMP
Numeric
8
11
FLOW
Numeric
8
12
OXY
Numeric
8
13
BOD5
Numeric
8
14
BOD2 0
Numeric
8
15
CBOD5
Numeric
8
16
CBOD2 0
Numeric
8
17
COD
Numeric
8
18
PH
Numeric
8
19
ALK
Numeric
8
20
ACID
Numeric
8
21
TN
Numeric
8
22
TON
Numeric
8
23
NH3
Numeric
8
24
N02
Numeric
8
25
N03
Numeric
8
26
N02 3
Numeric
8
27
TKN
Numeric
8
28
TP
Numeric
8
29
DISSP
Numeric
8
30
0P04T
Numeric
8
31
0P04D
Numeric
8
32
TOC
Numeric
8
33
DOC
Numeric
8
34
DIC
Numeric
8
35
TC
Numeric
8
36
CHLORIDE
Numeric
8
37
ZINCDISS
Numeric
8
38
ZINCTOT
Numeric
8
39
FEC_COLI
Numeric
8
40
CHLA
Numeric
8
41
PHEOPHYTN
Numeric
8
42
TOTRESIDUE
Numeric
8
43
NON
Numeric
8
44
DOP
Numeric
8
Total **
354
Dec Description
Agency identifier
Other identifier
Station name
calendar date YYYY/MM/DD
24-hr clock time (EST) hhmm
6 latitude (decimal degrees)
6 longitude (decimal degrees)
2 station bottom depth (ft)
2 station sample depth (ft)
2 water temperature (deg C)
3 flow rate (cfs)
2 dissolved oxygen (mg/L)
2 BOD 5-day (mg/L)
2 BOD 2 0-day (mg/L)
2 CBOD 5-day (mg/L)
2 CBOD 2 0-day (mg/L)
2 chemical oxygen demand (mg/L)
2 pH (standard units)
3 total alkalinity (mg/L as CaC03)
3 total acidity (mg/L as CaC03)
3 total nitrogen (mg/L as N)
3 total organic nitrogen (mg/L as N)
3 ammonia nitrogen (mg/L as N)
3 nitrite nitrogen (mg/L as N)
3 nitrate nitrogen (mg/L as N)
3 nitrite + nitrate nitrogen (mg/L as N)
3 total Kjeldahl nitorgen (mg/L as N)
3 total phosphorus (mg/L as P)
3 dissolved phosphorus (mg/L as P)
3 total orthophosphate (mg/L as P)
3 dissolved orthophosphate (mg/L as P)
3 total organic carbon (mg/L as C)
3 dissolved organic carbon (mg/L as C)
3 dissolved inorganic carbon (mg/L as C)
3 total carbon (mg/L as C)
3 total chloride (mg/L)
3 dissolved zinc (ug/L as Zn)
3 total zinc (ug/L as Zn)
1 fecal coliform bacteria (MPN/lOOmL)
2 chlorophyll-a (ug/L
2 pheophyton-a (ug/L)
2 total nonfilterable residue (mg/L)
3 computed: N02+N03 nitrogen (mg/L)
3 computed: dissolved organic phosphorus (mg/L)
6-6
6 - Databases
-------
Table 6-3. File structure of the NPDES point source discharges database.
Number of data records: 348
Field Field Name Type Width
1
SOURCE
Character
10
2
OTHERID
Character
10
3
STATION
Character
15
4
DATE
Date
8
5
TIME
Character
4
6
LAT
Numeric
10
7
LON
Numeric
10
8
BOTTOM
Numeric
7
9
DEPTH
Numeric
7
10
FLOW
Numeric
10
11
TEMP
Numeric
10
12
BOD
Numeric
10
13
SS
Numeric
10
14
NH3
Numeric
10
15
P
Numeric
10
16
DO
Numeric
10
17
FCB
Numeric
10
18
CL
Numeric
10
19
PHMAX
Numeric
10
20
PHMIN
Numeric
10
21
ZN
Numeric
10
22
CU
Numeric
10
23
FE
Numeric
10
24
FEDISS
Numeric
10
25
AL
Numeric
10
26
MN
Numeric
10
27
CR
Numeric
10
28
NI
Numeric
10
29
PB
Numeric
10
30
OIL_GREASE
Numeric
10
Total **
292
Description
Agency identifier
Other identifier
Station name
calendar date YYYY/MM/DD
24-hr clock time (EST) hhmm
latitude (decimal degrees)
longitude (decimal degrees)
(not used)
(not used)
flow rate (MGD)
temperature (deg C)
BOD (mg/L)
total suspended solids (mg/L)
ammonia nitrogen (mg/L as N)
total phosphorus (mg/L as P)
dissolved oxygen (mg/L)
fecal coliform bacteria (MPN/lOOmL)
chloride (mg/L)
maximum pH
minimum pH
total zinc (ug/L as Zn)
total copper (ug/L as Cu)
total iron (ug/L as Fe)
dissolved iron (ug/L as Fe)
total aluminum (ug/L as Al)
total manganese (ug/L as Mn)
total chromium (ug/L as Cr)
total nickel (ug/L as Ni)
total lead (ug/L as Pb)
oil Sc. grease
Dec
6
6
1
1
6
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
Databases
-------
Table 6-4. Characteristic (default) NPDES effluent concentrations.
Discharge Type
Code
Characteristic Concentration
NH3-N
(mg/L)
CBOD5
(mg/L)
TP
(mg/L)
DO
(mg/L)
TSS
(mg/L)
CLO
(mg/L)
FCB
(cfu/lOOmL)
Municipal WWTP
MUN
1.5
15.0
2.0
5.0
20.0
90
200
Small STP
STP
1.5
15.0
2.0
5.0
20.0
90
200
Advanced Secondary Treatment 1
ATP1
0.1
2.0
1.2
5.0
10.0
70
200
Advanced Secondary Treatment 2
ATP2
0.1
3.0
1.0
5.0
10.0
90
200
Single Residence Discharge
SRD
1.5
10.0
2.0
6.0
20.0
90
200
Water Filtration Plant
WFP
0.1
2.0
0.1
5.0
20.0
90
2
Industrial Treatment Discharge
IND
0.5
30.0
0.3
5.0
30.0
90
200
Noncontact Cooling Water
NCW
0.1
2.0
0.1
5.0
10.0
90
2
Stormwater Runoff
SWR
1.5
15.0
2.0
5.0
100.0
90
200
Groundwater Cleanup discharge
GWC
0.035
2.0
0.11
5.0
10.0
50
2
Exceptions:
PA0024058
STP
25.0
7.5
3.0
PA0025488
ATP2
4.0
106
DE0000451-002
NCW
4.0
Table 6-5. Characteristic NPDES effluent parameter ratios.
Discharge Type
Code
Characteristic Ratio
CBOD5:
BOD5
CBODu:
COBD5
DOC:TOC
TN:NH3
N03:NH3
N02:NH3
OP04:TP
Municipal WWTP
MUN
1.0
2.84
0.5
2.42
0.84
0.0300
0.92
Small STP
STP
1.0
2.84
0.5
2.42
0.84
0.0300
0.92
Advanced Secondary Treatment 1
ATP1
1.0
2.84
0.5
444.6
314.3
0.3333
0.94
Advanced Secondary Treatment 2
ATP2
1.0
2.84
0.5
222.3
157.2
0.2962
0.92
Single Residence Discharge
SRD
1.0
2.84
0.5
2.42
0.84
0.0300
0.92
Water Filtration Plant
WFP
1.0
2.84
0.5
2.42
0.84
0.0300
0.92
Industrial Treatment Discharge
IND
1.0
2.84
0.5
9.30
0.087
0.0435
0.33
Noncontact Cooling Water
NCW
1.0
2.84
0.5
2.42
0.84
0.0300
0.92
Stormwater Runoff
SWR
1.0
2.84
0.5
2.42
0.84
0.0300
0.92
Groundwater Cleanup discharge
GWC
1.0
2.84
0.5
60.0
40.0
10.000
0.92
Exceptions:
PA0024058
STP
0.75
PA0025488
ATP2
3.38
0.75
55.14
30.14
0.1429
6-8
6 - Databases
-------
/¦*
01480650
0148030/
S06(j
01480700
'1 4808(^0
01480500*
1870.x
(b 61(7
iOO
*480000
PA
MD
01478!
d/47800tf
Canal
Chesapeake «
Figure 6-1. Locations of USGS stream gaging stations.
6 - Data Bases
6-9
-------
RingF&m,
PA
MD
Canal
Chesapeake «
Figure 6-2. Locations of STORET water quality stations.
6-10
6 - Data Bases
-------
f\^s, ^.v
A0027987
r~ PA00S7339
PA0Q54
s> A0026018
PA0053821. PA0° l>568-001
036161
PAOOZ'6859 "-^» M \p
?S5056120 'v-%/
\ PA0055085
0 PA004
PA0052663
PA0047252
_ v PA0055484
un ^~t~PA0055476
OPAOO
PA0030848 Rin
South
3rook
PA004
PA00530
PA0057720
pA905
0056898
002548
PA0027103
PA0012637
&0Q52.019
0050067
DE0Q2W0
DE0000653X
045
27545
DE0DOO451
PA00537
DE0021768
A00570
ONJ0004286
NJ0024635
DEE0000566
DE0000051
DE0050
0000221
NJ0024023
/ ONJ0021601
ONJ0005100
Z>
ODE0000191
DEQ051004
MD00€5(f45
MD0022o41
Zj
)
NJ0021598
ONJ0024856
DE0050911
DE0020001° G
DE0000612
DE0000256
DE0021555
Ke & Delaware
Chesape0
Figure 6-3. Locations of the 120 NPDES point sources included in the model.
6 - Data Bases 6-11
-------
/¦*
PA001281
PAb02653'
^PAOQ26859
PA
MD
Canal
Chesapeake «
Figure 6-4. Locations of NPDES point sources in August 1997 study (Davis 1998).
6-12 6 - Data Bases
-------
7 - LOADS TO THE SYSTEM
External loads of nutrients and oxygen demand were divided into four classes: (1) nonpoint source
loads (i.e., diffuse sources) including tributary sources and groundwater sources, (2) point-source loads, (3)
water withdrawals, and (4) atmospheric loads. Nonpoint source loads were carried by freshwater flows and
groundwater entering the main stream reaches. Point-source loads were discharges from the sewage
treatment plants in the study area. Consumptive use water withdrawals were removed from the model system
at the appropriate grid cell. Atmospheric loads were transfers from the atmosphere to the water surface via
rainfall (wet deposition) and other processes (dry deposition). Atmospheric deposition is not a significant
source in the narrow stream channels, but may be more important in the open estuary waterbodies in the
lower Christina River and Delaware River because of the larger water surface area in those regions.
7.1 Nonpoint Source Loads
Carbon, nitrogen, phosphorus, dissolved oxygen, and floating algae were treated as mass loads in the
model. Nonpoint sources were defined by the delineation of subwatersheds in the HSPF model as shown in
Figure 7-1. Ideally, nonpoint source loads are generated by a watershed runoff model to provide predictive
nutrient loads to the receiving waters reflective of climatological (rainfall-runoff) characteristics. However,
the HSPF watershed loading model for the Christina River Basin is not scheduled to be completed for a few
more years. Instead, for this low-flow study, monitoring data in STORET were used to develop tributary
loads for the HSPF subwatersheds. Based on water-quality data collected by the USGS (Reif 1999) and
baseflow samples collected in 1997 (Senior 1999), estimates of nonpoint source concentrations for sub-
basins Bl, B5, B6, B8, and B13 were made for nitrite+nitrate nitrogen, ammonia nitrogen, and ortho-
phosphate. Estimates of nonpoint source concentrations for other sub-basins were based on instream water
quality stations within or downstream of the sub-basin. The lowest flow rates during the calibration period
were approximately equal to the 7Q10 flow on certain days in September. The nonpoint source loads
(kg/day) for model calibration were computed by multiplying the flow rates for the HSPF subwatersheds by
the characteristic concentration for a given water quality constituent. For the low-flow TMDL data set,
nonpoint source loads were computed using the daily average flow rates and the characteristic low-flow (or
background) concentrations. The estimated 7Q10 flow rates for each subwatershed are listed in Table 7-1.
The estimated nonpoint source (background) concentrations for each of the water quality constituents and for
each of the 39 subwatersheds are provided in Table 7-2. After the HSPF model is completed, the dynamic
nonpoint source loads and flow rates will be computed by HSPF and coupled to the EFDC receiving water
model. The HSPF model of the Christina River Basin includes a total of 39 subwatersheds.
7.2 Point Source Loads
For model calibration, a time-series of monthly average loads for the 1997 simulation period was
developed for nutrients, dissolved oxygen, chlorides, and total suspended solids at the point sources using the
DMR database. The only nutrients reported on the DMR records were ammonia nitrogen and total
7 - Loads to the System 7-1
-------
phosphorus, if any were reported at all. Thus, a method was developed to estimate the various species of
nitrogen, phosphorus, and carbon needed by the EFDC model from the sparse data provided on the DMRs.
Fortunately, detailed monitoring of several wastewater treatment plants was conducted in August 1997 by
Davis (1998). These plants included Kennett Square, Sunoco, DARA, Broad Run, South Coatesville, and
Coatesville (see Table 7-5). The point source monitoring included additional nutrient species for phosphorus
and nitrogen not reported on the DMRs. The ratio of each nitrogen species to ammonia nitrogen was
computed, as was the ratio of the phosphorus species to total phosphorus (see Table 6-5). These ratios were
then used to develop the loadings for each water quality parameter required by the EFDC model according to
the rules listed in Table 7-4. The loading rates for the discharges lacking DMR data (mainly the single
residence discharges) were estimated using the permit flow limit and the characteristic concentrations shown
in Table 6-4 for the associated discharge type. The locations of all 120 NPDES point source discharges
included in the model are provided in Table 7-6 arranged according to major stream reach. The river miles
are referenced to the mouth of the Christina River (river mile 74.9 according to EPA Reach File 1). The flow
limit listed in Table 7-6 is the permit limit for each discharge. The ratios for converting CBOD5 to organic
carbon were determined based on data collected during a special study conducted in August-September 1999
from several of the larger WWTPs in the basin. A discussion of the results of the special study is given in
Appendix I.
7.3 Water Withdrawals
There are a number of water withdrawals in the Christina River Basin. Only the 28 consumptive use
withdrawals were included in the model. For model calibration, the withdrawal rates were held constant for
all but four of the locations (DE-02, DE-04, DE-05, and DE-15) where daily withdrawal rates were used in
the model. For the other 24 locations, the withdrawal rates were set either to the safe yield rate or to 75% of
the pump capacity if the safe yield was not available. The locations of the water withdrawals included in the
model are listed in Table 7-7 arranged according to stream reach.
7.4 Atmospheric Loads
Atmospheric loads are typically divided into wet and dry deposition. Wet deposition is associated
with dissolved substances in rainfall. The settling of particulate matter during non-rainfall events contributes
to dry deposition. Observations of concentrations in rainwater are frequently available, and dry deposition is
usually estimated as a fraction of the wet deposition. The atmospheric deposition rates reported in the Long
Island Sound Study (Hydro Qual 1991) and the Chesapeake Bay Model Study (Cerco and Cole 1993) as well
as information provided by DNREC for Lewes, Delaware, were used to develop both dry and wet deposition
loads for the EFDC model of the Christina River Basin. The dry atmospheric deposition rates are presented
in Table 7-8, and the wet deposition concentrations are shown in Table 7-9. The loading rate for wet
deposition of nutrients was computed internally by the model by multiplying the rainfall rate times the
nutrient concentration during each model time step.
7-2
7 - Loads to the System
-------
Table 7-1. Estimated 7Q10 flow rates for watersheds in Christina River Basin.
WSID
Watershed Description
Area
(so.mi.)
7Q10 unit
(cfs/sq.mi.)
7Q10 Flow
Ccfs)
Brandywine Creek Watershed:
B1
Upper West Br. at Honeybrook
18.40
0.3351
6.17
B2
Upper West Br. at Hibernia
27.04
0.3352
9.06
B3
Upper West Br. at Coatesville
17.65
0.2809
4.96
B4
Upper West Br. at Embreeville
17.10
0.1708
2.92
B5
Buck Run
27.51
0.1708
4.70
B6
Doe Run
22.58
0.1708
3.86
B7
Broad Creek
6.44
0.1707
1.10
B8
Upper East Br. at Struble Lake
33.02
0.3765
12.43
B9
Upper East Br. at Shamona Creek
10.02
0.3015
3.02
BIO
Lower East Branch
20.93
0.1908
3.99
Bll
Marsh Creek
19.98
0.2816
5.62
B12
Beaver Creek
18.09
0.2815
5.09
B13
Valley Creek
20.65
0.1708
3.53
B14
Main Stem above Chadds Ford
24.43
0.1708
4.17
B15
Pocopson Creek
9.20
0.1708
1.57
B16
Main Stem below Chadds Ford
26.55
0.1700
4.51
B17
Main Stem through Wilmington
6.06
0.1801
1.09
Christina River Watershed:
CI
Main Stem above Cooches Bridge
14.31
0.0419
0.60
Clwb
West Branch
6.73
0.0223
0.15
C2
Muddy Run
8.67
0.0750
0.65
C3
Belltown Run
6.43
0.0746
0.48
C4
Little Mill Creek
9.23
0.5090
4.70
C5
Main Stem above Smalley's Pond
10.67
0.0749
0.80
C6
Main Stem Lower Tidal
21.20
0.0750
1.59
FF/zj'te Cfay Creek Watershed
W1
West Branch
10.21
0.2302
2.35
W2
Middle Branch
15.87
0.2306
3.66
W3
East Branch above Avondale
18.74
0.2305
4.32
W4
East Branch below Avondale
14.33
0.1703
2.44
W5
Mill Creek
12.95
0.1698
2.20
W6
Pike Creek
6.65
0.2408
1.60
W7
Middle Run
3.89
0.2389
0.93
W8
Main Stem above Newark
10.13
0.1698
1.72
W9
Main Stem above Delaware Park
9.05
0.2399
2.17
W10
Main Stem at Churchmans Marsh
5.51
0.2196
1.21
C/ay Creek Watershed:
R1
West Branch
17.48
0.2122
3.71
R2
East Branch
9.91
0.1403
1.39
R3
Burroughs Run
7.10
0.1197
0.85
R4
Main Stem above Wooddale
12.46
0.1099
1.37
R5
Main Stem below Wooddale
7.11
0.5092
3.62
7 - Loads to the System
-------
Table 7-2. Estimated nonpoint source (background) concentrations for HSPF watersheds in Christina River Basin.
Watershed Description
CYA
DIA
GRN
RPC
LPC
DOC
RPP
LPP
DOP
P4T
RPN
LPN
DON
NH4
N03
SUU
SAA
COD
DOO
TAM
FCB
TSS
CLO
WSID
(ug/L)
(ug/L)
(ug/L)
(mg/L)
(mg/L)
(mg/L)
(mg/L)
(mg/L)
(mg/L)
(mg/L)
(mg/L)
(mg/L)
(mg/L)
(mg/L)
(mg/L)
(mg/L)
(mg/L)
(mg/L)
(mg/L)
(mg/L)
E
o
o
(mg/L)
(mg/L)
Brandywine Creek Watershed:
B1
Upper West Br. at Honeybrook
0.0
0.0
1.0
0.00
0.00
2.10
0.003
0.003
0.004
0.010
0.045
0.045
0.090
0.020
1.80
0.00
0.00
6.00
7.34
0.0
20.0
2.0
11.0
B2
Upper West Br. at Hibernia
0.0
0.0
1.0
0.00
0.00
2.10
0.003
0.003
0.004
0.010
0.045
0.045
0.090
0.020
1.80
0.00
0.00
6.00
7.34
0.0
20.0
2.0
11.0
B3
Upper West Br. at Coatesville
0.0
0.0
1.0
0.00
0.00
2.10
0.003
0.003
0.004
0.010
0.045
0.045
0.090
0.020
1.80
0.00
0.00
6.00
7.34
0.0
20.0
2.0
11.0
B4
Upper West Br. at Embreeville
0.0
0.0
1.0
0.00
0.00
2.10
0.003
0.003
0.004
0.010
0.045
0.045
0.090
0.020
1.80
0.00
0.00
6.00
7.34
0.0
20.0
2.0
11.0
B5
Buck Run
0.0
0.0
1.0
0.00
0.00
2.10
0.003
0.003
0.004
0.010
0.045
0.045
0.090
0.020
1.80
0.00
0.00
6.00
7.34
0.0
20.0
2.0
11.0
B6
Doe Run
0.0
0.0
1.0
0.00
0.00
2.10
0.003
0.003
0.004
0.010
0.045
0.045
0.090
0.020
1.80
0.00
0.00
6.00
7.34
0.0
20.0
2.0
11.0
B7
Broad Creek
0.0
0.0
1.0
0.00
0.00
2.10
0.003
0.003
0.004
0.010
0.045
0.045
0.090
0.020
1.80
0.00
0.00
6.00
7.34
0.0
20.0
2.0
11.0
B8
Upper East Br. at Struble Lake
0.0
0.0
1.0
0.05
0.05
2.40
0.003
0.003
0.005
0.007
0.045
0.045
0.090
0.020
1.16
0.00
0.00
6.00
7.34
0.0
20.0
2.0
11.0
B9
Upper East Br. at Shamona Creek
0.0
0.0
1.0
0.05
0.05
2.40
0.003
0.003
0.005
0.007
0.045
0.045
0.090
0.020
1.16
0.00
0.00
6.00
7.34
0.0
20.0
2.0
11.0
B10
Lower East Branch
0.0
0.0
1.0
0.05
0.05
2.40
0.003
0.003
0.005
0.007
0.045
0.045
0.090
0.020
1.16
0.00
0.00
6.00
7.34
0.0
20.0
2.0
11.0
B11
Marsh Creek
0.0
0.0
1.0
0.05
0.05
2.40
0.003
0.003
0.005
0.007
0.045
0.045
0.090
0.020
1.16
0.00
0.00
6.00
7.34
0.0
20.0
2.0
11.0
B12
Beaver Creek
0.0
0.0
1.0
0.05
0.05
2.40
0.003
0.003
0.005
0.007
0.045
0.045
0.090
0.020
1.16
0.00
0.00
6.00
7.34
0.0
20.0
2.0
11.0
B13
Valley Creek
0.0
0.0
1.0
0.05
0.05
2.40
0.003
0.003
0.005
0.007
0.045
0.045
0.090
0.020
1.16
0.00
0.00
6.00
7.34
0.0
20.0
2.0
11.0
B14
Main Stem above Chadds Ford
0.0
0.0
1.0
0.05
0.05
2.40
0.003
0.003
0.005
0.007
0.045
0.045
0.090
0.020
1.16
0.00
0.00
6.00
7.34
0.0
20.0
2.0
11.0
B15
Pocopson Creek
0.0
0.0
1.0
0.05
0.05
2.40
0.003
0.003
0.005
0.007
0.045
0.045
0.090
0.020
1.16
0.00
0.00
6.00
7.34
0.0
20.0
2.0
11.0
B16
Main Stem below Chadds Ford
0.0
0.0
1.0
0.05
0.05
2.40
0.003
0.003
0.005
0.007
0.045
0.045
0.090
0.020
1.16
0.00
0.00
6.00
7.34
0.0
20.0
2.0
11.0
B17
Main Stem through Wlmington
0.0
0.0
1.0
0.05
0.05
2.40
0.003
0.003
0.005
0.007
0.045
0.045
0.090
0.020
1.16
0.00
0.00
6.00
7.34
0.0
20.0
2.0
11.0
Christina River Watershed:
C1
Christina R. above Cooches Bridqe
0.0
0.0
1.0
1.00
1.00
2.00
0.004
0.004
0.002
0.010
0.045
0.045
0.090
0.020
0.80
0.00
0.00
6.00
7.34
0.0
20.0
2.0
11.0
C1wb
Christina R. West Branch
0.0
0.0
1.0
1.00
1.00
2.00
0.004
0.004
0.002
0.010
0.045
0.045
0.090
0.020
0.80
0.00
0.00
6.00
7.34
0.0
20.0
2.0
11.0
C2
Muddy Run
0.0
0.0
1.0
1.00
1.00
2.00
0.004
0.004
0.002
0.010
0.045
0.045
0.090
0.020
0.80
0.00
0.00
6.00
7.34
0.0
20.0
2.0
11.0
C3
Belltown Run
0.0
0.0
1.0
1.00
1.00
2.00
0.004
0.004
0.002
0.010
0.045
0.045
0.090
0.020
0.80
0.00
0.00
6.00
7.34
0.0
20.0
2.0
11.0
C4
Little Mill Creek
0.0
0.0
1.0
1.00
1.00
2.00
0.004
0.004
0.002
0.010
0.045
0.045
0.090
0.020
0.80
0.00
0.00
6.00
7.34
0.0
20.0
2.0
11.0
C5
Main Stem above Smalley's Pond
0.0
0.0
1.0
1.00
1.00
2.00
0.004
0.004
0.002
0.010
0.045
0.045
0.090
0.020
0.80
0.00
0.00
6.00
7.34
0.0
20.0
2.0
11.0
C6
Main Stem Lower Tidal
0.0
0.0
1.0
1.00
1.00
2.00
0.004
0.004
0.002
0.010
0.045
0.045
0.090
0.020
0.80
0.00
0.00
6.00
7.34
0.0
20.0
2.0
11.0
White Clay Creek Watershed
W1
West Branch White Clay Creek
0.0
0.0
1.0
0.15
0.15
1.50
0.007
0.007
0.002
0.004
0.045
0.045
0.090
0.020
1.59
0.00
0.00
6.00
7.34
0.0
20.0
2.0
11.0
W2
Middle Branch White Clay Creek
0.0
0.0
1.0
0.15
0.15
1.50
0.007
0.007
0.002
0.004
0.045
0.045
0.090
0.020
1.59
0.00
0.00
6.00
7.34
0.0
20.0
2.0
11.0
W3
East Branch above Avondale
0.0
0.0
1.0
0.15
0.15
1.50
0.007
0.007
0.002
0.004
0.045
0.045
0.090
0.020
1.59
0.00
0.00
6.00
7.34
0.0
20.0
2.0
11.0
W4
East Branch below Avondale
0.0
0.0
1.0
0.15
0.15
1.50
0.007
0.007
0.002
0.004
0.045
0.045
0.090
0.020
1.59
0.00
0.00
6.00
7.34
0.0
20.0
2.0
11.0
W5
Mill Creek
0.0
0.0
1.0
0.15
0.15
1.50
0.007
0.007
0.002
0.004
0.045
0.045
0.090
0.020
1.59
0.00
0.00
6.00
7.34
0.0
20.0
2.0
11.0
W6
Pike Creek
0.0
0.0
1.0
0.15
0.15
1.50
0.007
0.007
0.002
0.004
0.045
0.045
0.090
0.020
1.59
0.00
0.00
6.00
7.34
0.0
20.0
2.0
11.0
W7
Muddy Run
0.0
0.0
1.0
0.15
0.15
1.50
0.007
0.007
0.002
0.004
0.045
0.045
0.090
0.020
1.59
0.00
0.00
6.00
7.34
0.0
20.0
2.0
11.0
W8
Main Stem above Newark
0.0
0.0
1.0
0.15
0.15
1.50
0.007
0.007
0.002
0.004
0.045
0.045
0.090
0.020
1.59
0.00
0.00
6.00
7.34
0.0
20.0
2.0
11.0
W9
Main Stem above Delaware Park
0.0
0.0
1.0
0.15
0.15
1.50
0.007
0.007
0.002
0.004
0.045
0.045
0.090
0.020
1.59
0.00
0.00
6.00
7.34
0.0
20.0
2.0
11.0
W10
Main Stem at Churchmans Marsh
0.0
0.0
1.0
0.15
0.15
1.50
0.007
0.007
0.002
0.004
0.045
0.045
0.090
0.020
1.59
0.00
0.00
6.00
7.34
0.0
20.0
2.0
11.0
Red Clay Creek Watershed:
R1
West Branch Red Clay Creek
0.0
0.0
1.0
0.00
0.00
2.80
0.014
0.014
0.006
0.011
0.045
0.045
0.090
0.020
1.78
0.00
0.00
6.00
7.34
0.0
20.0
2.0
11.0
R2
East Branch Red Clay Creek
0.0
0.0
1.0
0.00
0.00
2.80
0.014
0.014
0.006
0.011
0.045
0.045
0.090
0.020
1.78
0.00
0.00
6.00
7.34
0.0
20.0
2.0
11.0
R3
Burrouqhs Run
0.0
0.0
1.0
0.00
0.00
2.80
0.014
0.014
0.006
0.011
0.045
0.045
0.090
0.020
1.78
0.00
0.00
6.00
7.34
0.0
20.0
2.0
11.0
R4
Main Stem above Wooddale
0.0
0.0
1.0
0.00
0.00
2.80
0.014
0.014
0.006
0.011
0.045
0.045
0.090
0.020
1.78
0.00
0.00
6.00
7.34
0.0
20.0
2.0
11.0
R5
Main Stem below Wooddale
0.0
0.0
1.0
0.00
0.00
2.80
0.014
0.014
0.006
0.011
0.045
0.045
0.090
0.020
1.78
0.00
0.00
6.00
7.34
0.0
20.0
2.0
11.0
CYA
= cyanobacteria
RPP
= refractory particulate phosphorus
RPN
= refractory particulate nitrogen
SUU
= unavailable silica
DIA
= diatom algae
LPP
= labile particulate phosphorus
LPN
= labile particulate nitrogen
SAA
= available biogenic silica
GRN
= green algae
DOP
= dissolved organic phosphorus
DON
= dissolved organic nitrogen
COD
= chemical oxygen demand
RPC
= refractory particulate carbon
P4T
= total orthophosphate
NH4
= ammonia nitrogen
DOO
= dissolved oxygen
LPC
= labile particulate carbon
N03
= nitrite+nitrate nitrogen
TAM
= total active metal
DOC
= dissolved organic carbon
FCB
TSS
CLO
= fecal coliform bacteria
= total suspended solids
= chlorides
-------
Table 7-3. Wastewater treatment plant monitoring, August 1997.
PA0024058
Kennett Sq
PA0012815
Sunoco
PA0012815
Sunoco
PA0026531
DARA
PA0026531
DARA
PA0043982
Broad Run
PA0043982
Broad Run
PA0036897
S. Coatesville
PA0026859
Coatesville
PA0025488
Avondale
08/07/97
08/13/97
08/14/97
08/13/97
08/14/97
08/13/97
08/14/97
08/20/97
08/20/97
08/27/97
Parameter
(mg/L)
(mg/L)
(mg/L)
(mg/L)
(mg/L)
(mg/L)
(mg/L)
(mg/L)
(mg/L)
(mg/L;
BOD20 inh.
52.000
136.000
5.200
6.900
3.200
2.700
2.400
4.600
6.500
BOD20
121.000
143.000
9.700
16.000
11.200
12.000
8.600
11.000
24.000
BOD5 inh
21.000
29.000
51.000
1.500
2.800
1.500
0.600
0.600
2.500
2.800
BOD5
20.000
30.000
52.000
2.800
4.600
1.700
0.800
0.500
2.700
7.100
Total N
21.800
2.860
4.280
25.200
23.300
36.800
42.100
27.100
20.700
3.860
NH3-N
9.000
0.120
0.460
0.050
0.230
0.050
0.090
0.030
0.060
0.070
N02-N
0.270
0.010
0.020
0.020
0.030
0.020
0.030
0.010
0.020
0.010
N03-N
7.580
0.040
0.040
14.800
14.400
27.900
28.070
19.850
16.790
2.110
Total P
5.980
0.060
0.300
0.790
1.460
0.720
0.080
1.200
1.480
8.050
Diss OrthoP
4.850
0.012
0.016
0.770
0.510
0.550
0.870
1.350
6.620
Total OrthoP
5.940
0.020
0.780
0.600
1.390
TSS (residue total
nonvolatile)
62.000
36.000
40.000
10.000
20.000
8.000
10.000
16.000
2.000
Chloride
122.000
44.000
49.000
83.000
79.000
109.000
114.000
55.000
70.000
106.000
Ratios
Secondary
Treatment
Average
Advanced
Treatment
Average
BOD20 : BOD5
6.0500
2.7500
3.4643
3.4783
6.5882
15.0000
17.2000
4.0741
3.3803
4.4000
7.5979
'BOD20: BOD5 )inh
2.4762
2.6667
3.4667
2.4643
2.1333
4.5000
4.0000
1.8400
2.3200
2.5714
2.9608
IN : NH3
2.4222
23.8333
9.3043
504.0000
101.3043
736.0000
467.7778
903.3333
345.0000
55.1429
11.8533
444.6512
N02 : NH3
0.0300
0.0833
0.0435
0.4000
0.1304
0.4000
0.3333
0.3333
0.3333
0.1429
0.0523
0.2962
N03 : NH3
0.8422
0.3333
0.0870
296.0000
62.6087
558.0000
311.8889
661.6667
279.8333
30.1429
0.4208
314.3058
Diss OrthoP : TP
0.8110
0.2000
0.0533
0.9747
0.7083
6.8750
0.7250
0.9122
0.8224
0.3548
1.8363
Total OrthoP : TP
0.9933
0.3333
0.9873
0.8333
0.9392
0.6633
0.9200
TP : NH3
0.6644
0.5000
0.6522
15.8000
6.3478
14.4000
0.8889
40.0000
24.6667
115.0000
0.6055
31.0148
Notes: Secondary treatment: Kennett Square WWTP and Sunoco
Advanced treatment (nitrification): DARA, Broad Run, South Coatesville, and Coatesville
-------
Table 7-4. Methodology for developing EFDC point source loads from DMR data.
Water Quality Parameter
EFDC Code
Calculation
CBOD-5-day
CBOD5 = BOD5 * (CBOD5:BOD5 ratio)
CBOD-ultimate
CBODu = CBOD5 * (CBODu:CBOD5 ratio)
Total organic carbon
TOC
TOC = CBODu * (TOC:CBODu ratio)
Dissolved organic carbon
DOC
DOC = TOC * (DOC:TOC ratio)
Refractory particulate organic carbon
RPOC
0.5 * (TOC - DOC)
Labile particulate organic carbon
LPOC
0.5 * (TOC - DOC)
Total phosphorus
Total organic phosphorus
If TP not reported on DMR, use default TP from Table 6-4
TOP = TP - (TP * (0P04:TP ratio))
Refractory particulate organic phosphorus
RPOP
0.25 TOP
Labile particulate organic phosphorus
LPOP
0.25 TOP
Dissolved organic phosphorus
DOP
0.50 TOP
Total orthophosphate
P04T
TP * (OP04:TP ratio)
Total nitrogen
Nitrite nitrogen
Total organic nitrogen
TN = NH3-N * (TN:NH3 ratio)
N02-N = NH3-n * (N02:N03 ratio)
TON = TN - N02-N - N03-N - NH3-N
Refractory particulate organic nitrogen
RPON
0.25 TON
Labile particulate organic nitrogen
LPON
0.25 TON
Dissolved organic nitrogen
DON
0.50 TON
Ammonia nitrogen
NH3
reported on DMR (or use default NH3-N from Table 6-4)
Nitrate nitrogen
N03
N03-N = NH3 * (N03:NH3 ratio)
Unavailable biogenic silica
SUU
0.10 mg/L (default value)
Dissolved available silica
SAA
1.00 mg/L (default value)
Chemical oxygen demand
COD
9.6 * CBOD5
Dissolved oxygen
DOO
reported on DMR (or use default value from Table 6-4)
Total active metal
TAM
0.0 (not simulated)
Fecal coliform bacteria
FCB
reported on DMR (or use default value from Table 6-4)
7-6
7 - Loads to the System
-------
Table 7-5. EFDC water quality parameter concentrations for WWTPs monitored in August 1997 study.
PA0024058
PA0012815
PA0012815
PA0026531
PA0026531
PA0043982
PA0043982
PA0036897
PA0026859
PA0025488
Kennett Sq
Sunoco
Sunoco
DARA
DARA
Broad Run
Broad Run
S.Coatesville
Coatesville
Avondale
08/07/97
08/13/97
08/14/97
08/13/97
08/14/97
08/13/97
08/14/97
08/20/97
08/20/97
08/27/97
Parameter
(mg/L)
(mg/L)
(mg/L)
(mg/L)
(mg/L)
(mg/L)
(mg/L)
(mg/L)
(mg/L)
(mg/L;
BOD20 ink
52.000
136.000
5.200
6.900
3.200
2.700
2.400
4.600
6.500
BOD20
121.000
143.000
9.700
16.000
11.200
12.000
8.600
11.000
24.000
BOD5 inh
21.000
29.000
51.000
1.500
2.800
1.500
0.600
0.600
2.500
2.800
BOD5
20.000
30.000
52.000
2.800
4.600
1.700
0.800
0.500
2.700
7.100
Total N
21.800
2.860
4.280
25.200
23.300
36.800
42.100
27.100
20.700
3.860
NH3-N
9.000
0.120
0.460
0.050
0.230
0.050
0.090
0.030
0.060
0.070
N02-N
0.270
0.010
0.020
0.020
0.030
0.020
0.030
0.010
0.020
0.010
N03-N
7.580
0.040
0.040
14.800
14.400
27.900
28.070
19.850
16.790
2.110
Total P
5.980
0.060
0.300
0.790
1.460
0.720
0.800
1.200
1.480
8.050
Diss OrthoP
4.850
0.012
0.016
0.770
0.510
0.550
0.870
1.350
6.620
Total OrthoP
5.940
0.020
0.780
0.600
0.600
0.870
1.390
6.620
TSS (res tot nonvolatile)
62.000
36.000
40.000
10.000
20.000
8.000
10.000
16.000
2.000
Chloride
122.000
44.000
49.000
83.000
79.000
109.000
114.000
55.000
70.000
106.000
EFDC Parameters
TOC
38.951
97.378
6.605
10.895
7.627
8.172
5.856
7.491
16.343
RPOC
9.738
24.345
1.651
2.724
1.907
2.043
1.464
1.873
4.086
LPOC
9.738
24.345
1.651
2.724
1.907
2.043
1.464
1.873
4.086
DOC
19.476
48.689
3.303
5.448
3.813
4.086
2.928
3.745
8.172
RPOP
0.010
0.010
0.003
0.030
0.050
0.083
0.023
0.358
LPOP
0.010
0.010
0.003
0.030
0.050
0.083
0.023
0.358
DOP
0.020
0.020
0.005
0.060
0.100
0.165
0.045
0.715
P4T
5.940
0.020
0.780
0.600
0.600
0.870
1.390
6.620
RPON
1.238
0.673
2.583
2.208
3.477
1.802
0.958
0.418
LPON
1.238
0.673
2.583
2.208
3.477
1.802
0.958
0.418
DON
2.475
1.345
5.165
4.415
6.955
3.605
1.915
0.835
NH4
9.000
0.120
0.050
0.050
0.090
0.030
0.060
0.070
N03
7.850
0.050
14.820
27.920
28.100
19.860
16.810
2.120
-------
Table 7-6. Locations of NPDES point source discharges included in the model.
RIVER
CELL
NPDES
FLOWLIM
MILE
I
J
NUMBER
MGD
CODE
OWNER
STREAM
TYPE
DESCRIPTION
Brandywine Creek
(main stem)
76.610
54
15
DE0050 962
0.0000
SWR
AMTRAK
TB-Brandywine Creek
Industrial
Stormwater
83 . 554
54
27
DE002176 8
0 . 0250
STP
Winterthur Museum
Clenney Run
Municipal
Small STP
88.644
54
37
PA0053 082
0 . 0206
STP
Mendenhall Inn
TB Brandywine Creek
Commercial
Small STP
89.917
54
38
PA0052 663
0.0900
STP
Knight's Bridge Co/Villages at Painters
Harvey Run
Commercial
Small STP
89.917
54
38
PA005547 6
0.0400
STP
Birmingham TSA/Ridings at Chadds Ford
TB Harvey Creek
Municipal
Small STP
89.917
54
38
PA0055085
0.0005
SRD
Winslow Nancy Ms.
TB Brandywine Creek
Municipal
Single Residence STP
89.917
54
38
PA0055484
0.0005
SRD
Keating Herbert & Elizabeth
TB Brandywine Creek
Municipal
Single Residence STP
89.917
54
38
PA0047252
0.0700
STP
Pantos Corp/Painters Crossing
Harvey Run
90.553
54
39
PA0030848
0.0063
STP
Unionville - Chadds Ford Elem. School
Ring Run
Municipal
Small STP
93 . 098
54
42
PA0056120
0.0005
SRD
Schindler
Pocopson Creek
Municipal
Single Residence STP
92.462
54
43
PA003109 7
0.0170
STP
Radley Run C.C.
Radley Run
Municipal
Small STP
92.462
54
43
PA0053449
0.1500
STP
Birmingham Twp. STP
Radley Run
Municipal
Small STP
93.735
54
43
PA0057 011
0.0773
STP
Thornbury Twp./Bridlewood Farms STP
Radley Run
92.462
54
44
PA003 62 0 0
0.0320
STP
Radley Run Mews
Plum Run
Municipal
Small STP
94.371
54
44
PA0056171
0.0005
SRD
McGlaughlin Jeffrey
Plum Run
Municipal
Single Residence STP
94.371
54
44
PA0050 005
0.1400
GWC
Sun Company
TB Brandywine Creek
GWCleanup
New permit 03/27/98
94.371
54
44
PA005149 7
0.0300
NCW
Lenape Forge
Brandywine Creek
Industrial
Cooling Water
Brandywine Creek
East Branch
98.647
54
52
PA002 6 018
1. 8000
MUN
West Chester Borough MUA/Taylor Run
Taylor Run
Municipal
Large STP
98.647
54
52
PA0054747
0.0000
SWR
Trans-Materials, Inc.
Taylor Run
Industrial
Stormwater
98.647
54
52
PA00572 82
0.0005
SRD
Jonathan & Susan Pope
TB Valley Creek
Municipal
Single Residence STP
99.276
54
53
PA00513 65
0.3690
WFP
West Chester Area Mun. Auth.
EB Brandywine Creek
Municipal
Ingram's Mill-Filter Backwash
100.535
54
55
PA0053 93 7
0.0005
SRD
Johnson Ralph & Gayla
Broad Creek
Municipal
Single Residence STP
100.535
54
55
PA005632 4
0.0440
GWC
Mobil SS#16-GPB
TB-WB Valley Run
Commercial
DP
100.535
54
55
PA0056 618
0.0005
SRD
O'Cornwell David & Jeanette
Broad Run
Municipal
Single Residence STP
100.535
54
55
PA0054305
0.0000
IND
Sun Co, Inc. (R&M)
TB Valley Creek
Industrial
100.535
54
55
PA0053561
0.0360
GWC
Johnson Matthey
Valley Creek
GWCleanup
Permitted 03/12/96
101.794
54
57
PA0043 982
0 . 4000
ATP2
Broad Run Sew Co.
EB Brandywine Creek
Municipal
Large STP
103.682
54
61
PA0012815
1.0280
IND
Sunoco Products
EB Brandywine Creek
Industrial
Paper Company - Mill Raceway
103.682
54
60
PA0026531
7.1340
ATP2
Downingtown Area Regional Authority
EB Brandywine Creek
Municipal
Large STP
104.312
54
61
PA0051918
0.1440
NCW
Pepperidge Farms
Parke Run Creek
Industrial
Cooling Water
103.682
54
61
PA0055531
0 . 0007
STP
Khalife Paul
TB Valley Run
Commercial
Small STP
104.312
54
61
PA0057126
0.0000
IND
Hess Oil - SS #38291
Valley Run
Commercial
DP
104.312
54
61
PA003 022 8
0.0225
STP
Downingtown I&A School
Beaver Creek
Municipal
No flow since Feb 1994
104.312
54
61
PA0053678
0.0000
IND
Lambert Earl R.
EB Brandywine Creek
Industrial
DP
104.312
54
61
PA0053660
0.0000
IND
Mobil Oil Company #016
EB Brandywine Creek
Commercial
Air stripper at Service Sta
106.830
54
65
PA0054917
0.4750
STP
Uwchlan Twp. Municipal Authority
Shamona Creek
Municipal
Eagleview CC STP
107.459
54
66
PA0057 045
0 . 0000
SWR
Shyrock Brothers, Inc.
EB Brandywine Creek
Commercial
Stormwater
108.088
54
67
PA002 7 98 7
0.0500
STP
Pennsylvania Tpk./Caruiel Service Plaza
Marsh Creek
Commercial
Small STP
108.088
54
67
PA003 63 74
0.0150
STP
Eaglepoint Dev. Assoc.
TB Marsh Creek
Municipal
Small STP
108.088
54
67
PA0052949
0.0000
IND
Phila. Suburban Water Co.
Marsh Creek
Industrial
Uwchlan DP
108.088
54
67
PA00572 74
0.0005
SRD
Michael & Antionette Hughes
TB Marsh Creek
Municipal
Single Residence STP
109.977
54
70
PA0050458
0.0531
STP
Little Washington Drainage Co.
Culbertson Run
Municipal
Small STP
112.495
54
74
PA005022 9
0.0005
SRD
unknown
Indian Run
Municipal
Single Residence STP
112.495
54
74
PA0050547
0.0375
STP
Indian Run Village MHP
Indian Run
Municipal
Small STP
112.495
54
74
PA0055492
0.0005
SRD
Topp John & Jane
Indian Run
Municipal
Single Residence STP
113.753
54
76
PA0054691
0.0005
SRD
Stoltzfus Ben Z.
TB Brandywine Creek
Municipal
Single Residence STP
-------
Table 7-6. Locations of NPDES point source discharges included in the model (continued).
RIVER
CELL
NPDES
FLOWLIM
MILE
I, J
NUMBER
MGD
CODE
Brandywine Creek
West Branch
97.976
46, 79
PA0056561
0.0000
SWR
101.708
40, 79
PA002 9 912
0 . 1000
STP
102.330
39, 79
PA0053996
0.0005
SRD
107.306
29, 79
PA0053228
0.0005
SRD
107.306
29, 79
PA0053236
0.0005
SRD
110.416
24, 79
PA003 6 89 7
0.3900
ATP1
111.038
23, 79
PA0026859
3.8500
ATP1
111.038
23, 79
PA001156 8-
001
0 . 5000
IND
111.038
23, 79
PA001156 8-
016
0.5000
IND
111.038
23, 79
PA0053 821
0.0000
SWR
112 .282
20, 79
PA0012416
0.1400
WFP
112.282
20, 79
PA0052990
0.0005
SRD
112 .282
20, 79
PA0056 073
0.0005
SRD
113.526
18, 79
PA0052 72 8
0.0004
STP
114.770
16, 79
PA0055697
0.0490
STP
120.368
06, 79
PA003 6412
0.0550
STP
120.368
06, 79
PA0044776
0 . 6000
STP
120.368
06, 79
PA005733 9
0.0005
SRD
Buck Run
117.041
33, 61
PA0024473
0.7000
STP
117.041
33, 61
PA003 6161
0.0360
STP
117.041
33, 61
PA0057231
0.0005
SRD
DESCRIPTION
Christina River
82.274 45,13
83.561 43,09
(tidal)
DE 0000400-001
DE0051004
Christina River West Branch
White Clay Creek
93.090 32,18
102.824 15,18
108.696 06,18
0.0000 NCW
0.0000 SWR
99.587
16, 09
MD0065145
0 . 0500
STP
100.209
14, 09
MD0022641
0 . 4500
STP
Red Clay
Creek
89.828
43, 26
DE 0000221-001
0.0060
NCW
89.828
43, 26
DE 0000221-003
0.0040
NCW
91.746
43, 29
DE 0000230-001
0.3500
NCW
95.583
43, 35
DE 0021709-001
0.0150
STP
96.861
43, 37
PA0055425
0.0005
SRD
98.780
43, 40
DE0050 06 7
0.0015
STP
98.780
43, 40
DE0000451-002
2.1700
NCW
101.337
43, 44
PA0055107
0.1500
STP
103.255
43, 47
PA0054755
0.0000
SWR
Red Clay
Creek West Branch
103.313
32, 43
PA0053554
0.0000
SWR
103.950
30, 43
PA0024058
1.1000
STP
104.268
29, 43
PA0050 67 9
0.2500
NCW
104.579
28, 43
PA0057 72 0-001
0.0500
STP
104.579
28, 43
PA0057 72 0-003
0.0900
NCW
DE 0000191-001
PA0053 783
PA0024 06 6
0.03 00 NCW
0.0200 STP
0.2500 STP
Richard M. Armstrong Co.
Embreeville Hospital
Redmond Michael
Gramm Jeffery
Woodward Raymond Sr. STP
South Coatesville Borough
Coatesville City Authority
Lukens Steel Co.
Lukens Steel Co.
Chester County Aviation Inc.
Coatesville Water Plant
Mitchell Rodney
Vreeland Russell Dr.
Farmland Industries Inc./Turkey Hill
Spring Run Estates
Tel Hai Retirement Community
NW Chester Co. Municipal Authority
Brian & Cheryl Davidson
Parkersburg Borough Authority WWTP
Lincoln Crest MHP STP
Archie & Cloria Shearer
Ciba-Geigy Corp.
Boeing
Highlands WWTP
Meadowview Utilities, Inc.
HAVEG/AMTEK (eliminated July 1996)
HAVEG/AMTEK (eliminated July 1996)
Hercules Inc.
Greenville Country Club
D'Ambro Anthony Jr.-Lot #2 2
Center for Creative Arts
NVF Yorklyn
East Marlborough Township STP
Trans-Materials Inc.
Earthgro Inc.
Kennett Square Boro. WWTP
National Vulcanized Fiber (NVF)
Sunny Dell Foods, Inc.
Sunny Dell Foods, Inc.
FMC Corp.
Avon Grove School Dist
West Grove Borough Authority STP
Broad Run
WB Brandywine Creek
TB-WB Brandywine Creek
WB Brandywine Creek
WB Brandywine Creek
WB Brandywine Creek
WB Brandywine Creek
Sucker Run
Sucker Run
Sucker Run
Rock Run
Rock Run
TB Rock Run
WB Brandywine Creek
WB Brandywine Creek
TB-WB Brandywine Creek
WB Brandywine Creek
TB-WB Brandywine Creek
TB-Buck Run
Buck Run
TB-Buck Run
Christina River
Nonesuch Creek
WB Christina River
WB Christina River
Red Clay Creek
Red Clay Creek
Red Clay Creek
TB-Red Clay Creek
TB-EB Red Clay Creek
TB-Red Clay Creek
Red Clay Creek
TB-EB Red Clay Creek
EB Red Clay Creek
WB Red Clay Creek
WB Red Clay Creek
TB-WB Red Clay Creek
WB-Red Clay Creek
WB-Red Clay Creek
Cool Run
TB-WB White Clay Creek
MB White Clay Creek
Commercial Stormwater
Municipal Large STP
Municipal Single Residence STP
Municipal Single Residence STP
Municipal Single Residence STP
Municipal Large STP
Municipal Large STP
Industrial Large STP
Industrial Large STP
Commercial Stormwater
Industrial Water Filtration Plant-Backwash
Municipal Single Residence STP
Municipal Single Residence STP
Industrial Small STP
Commercial Small STP
Municipal Small STP
Municipal Large STP
Municipal Single Residence STP
Municipal Small STP-discontinued 06/10/97
Municipal Small STP
Municipal Single Residence STP
Industrial Cooling Water
Industrial Stormwater
Municipal
Municipal
Small STP
Small STP
Industrial Cooling Water
Industrial Cooling Water
Industrial Cooling Water
Municipal Small STP
Municipal Single Residence STP
Municipal Small STP
Industrial Stormwater/Cooling Water
Municipal Large STP
Industrial Stormwater
Industrial Stormwater
Municipal Large STP
Industrial Cooling Water
Industrial Mushroom Canning/Process Water
Industrial Mushroom Canning/Cooling Water
Industrial Stormwater/Cooling Water
Commercial Small STP
Municipal Large STP
-------
Table 7-6. Locations of NPDES point source discharges included in the model (continued).
RIVER CELL
MILE I, J
NPDES
NUMBER
FLOWLIM
MGD CODE OWNER
DESCRIPTION
White Clay Creek East Branch
102.750 19,24 PA0052451
104.020 19,26
106.560 19,30
106.560 19,30
106.560 19,30
107.195 19,31
107.830 19,32
107.830 19,32
107.830 19,32
Little Mill Creek
82.441 41,55 DE00 0052 3- 001
83.373 38,55
Delaware River
PA0057 02 9
PA002548 8
PA0052019
PA0056 89 8
PA0056952
PA002 9343
PA004043 6
PA0040 665
DE00 0056 6
0.0012 STP Frances L. Hamilton Oates STP
0.1440 GWC Hewlett Packard Co.
0.3000 ATP2 Avondale Borough Sewer Authority
0.0075 STP Avon Grove Trailer Court
0.0650 IND To-Jo Mushrooms Inc.
0.0029 IND Sun Company Inc.
0.02 70 STP Chatham Acres
0.0090 STP Chadds Ford Investment Co./Red Fox GC
0.0100 STP Stone Barn Restuarantand Apt. Cplx
0.0000 SWR General Motors Assembly
0.0000 SWR DuPont Chestnut Run
EB White Clay Creek
Egypt Run
Indian Run
EB White Clay Creek
Trout Run
EB White Clay Creek
TB-EB White Clay Creek
TB-EB White Clay Creek
EB White Clay Creek
Little Mill Creek
Little Mill Creek
Municipal Small STP
GWCleanup Groundwater Cleanup
Municipal Large STP
Municipal Small STP
Industrial Small STP-online Jan 98
GWCleanup Groundwater Cleanup
Municipal Small STP
Municipal Small STP
Commercial Small STP
Industrial Stormwater
Industrial Stormwater/Cooling Water
63.839
57, 04
DE0021555-001
0.5500
MUN
Delaware City STP
Delaware
River
Municipal
65.272
57, 05
DE 0000256-601
13.0000
IND
Star Enterprises
Delaware
River
Industrial
65.272
57, 05
DE 0000612-001
0 . 8000
IND
Formosa Plastics Corp.
Delaware
River
Industrial
65.272
57, 05
DE 0020001-001
0 . 6800
MUN
Standard Chlorine
Delaware
River
Municipal
65.272
57, 05
DE 0050911-001
0.3000
MUN
Occidental Chemical Corp.
Delaware
River
Municipal
75.237
57, 15
DE 0020320-001
90.0000
MUN
City of Wilmington
Delaware
River
Municipal
77.162
57, 17
DE00 00 051-001
5.2000
IND
Dupont-Edgemoor
Delaware
River
Industrial
77.162
57, 17
DE00 00 051-002
3.0000
IND
Dupont-Edgemoor
Delaware
River
Industrial
77.162
57, 17
DE00 00 051-003
6.0000
IND
Dupont-Edgemoor
Delaware
River
Industrial
81.307
57,20
DE00 00 655-001
33.3000
IND
General Chemical Corporation
Delaware
River
Industrial
83 . 907
57, 22
PA0012 63 7-002
52.3500
IND
Bayway Manufacturing
Delaware
River
Industrial
SEE
NOTE
1
83 . 907
57, 22
PA0012 637-101
69.8000
IND
Bayway Manufacturing
Delaware
River
Industrial
SEE
NOTE
1
83 . 907
57, 22
PA0012 63 7-201
3.3400
IND
Bayway Manufacturing
Delaware
River
Industrial
SEE
NOTE
1
85.199
57, 23
PA002 7103- 001
44.0000
MUN
Delcora
Delaware
River
Municipal
82 . 639
58,21
NJ00 05 045-001
0.5000
IND
Monsanto
Delaware
River
Industrial
SEE
NOTE
2
63 . 839
59, 04
NJ0024 85 6-001
1.4450
MUN
City of Salem
Delaware
River
Municipal
SEE
NOTE
1
69.534
59, 09
NJ 0021598-001
2.4650
MUN
Pennsville Sewage Authority
Delaware
River
Municipal
SEE
NOTE
1
73 . 339
59, 12
NJ00 0510 0-661
22 . 9000
IND
Dupont-Chambers Works
Delaware
River
Industrial
SEE
NOTE
1
75.237
59, 15
NJ 0021601-001
1.7290
MUN
Carneys Pt. Sewage Authority
Delaware
River
Municipal
SEE
NOTE
1
76.045
59, 16
NJ0024 02 3- 001
0.9500
MUN
Penns Grove Sewage Authority
Delaware
River
Municipal
SEE
NOTE
1
77.162
59, 17
NJ0024 63 5-001
0 . 0366
MUN
Fort Dix/Pedricktown Facility
Delaware
River
Municipal
SEE
NOTE
1
79.919
59, 19
NJ0004286-001
2.1000
IND
Geon
Delaware
River
Industrial
82 . 639
59,21
NJ002 7545-001
0.9860
MUN
Logan Township MUA
Delaware
River
Municipal
SEE
NOTE
1
[1] No flow limit available in PCS data base; flow limit shown is maximum reported flow during 01/01/95 to 12/31/98
[2] No flow limit or reported flow available in PCS data base; flow limit shown is an estimate
-------
Table 7-7. Locations of consumptive use water withdrawals included in the model.
RIVER CELL
MILE I, J
WITHDRAWAL
NUMBER
FLOWLIM
MGD
CODE
OWNER
STREAM
LOCATION
Brandywine Creek
79.100 54,20
DE-01
16 . 0000
CUW
City of Wilmington
Brandywine Creek
Brandywine WTP
79.100 54,20
DE-02
20.0000
CUW
City of Wilmington
Brandywine Creek
Porter WTP
79.100 54,20
DE-08
0.7500
CUW
Wilmington Finishing
Brandywine Creek
Wilmington, DE
Brandywine Creek
East Branch
99.906 54,54
PA-4
4 .5000
CUW
West Chester MUA
EB Brandywine
Creek
West Chester, PA
100.535 54,55
6996-006
0.4650
CUW
General Crushed Stone
Valley Creek
100.535 54,55
6996-013
0 . 0030
CUW
General Crushed Stone
Valley Creek
103.682 54,60
6987-004
0 . 9900
CUW
Sonoco Products Co.
EB Brandywine
Creek
Downingtown, PA
104.312 54,61
250156-022
0 . 0098
CUW
Ingleside Golf Course
EB Brandywine
Creek
104.312 54,61
6990-004
0.0180
CUW
Brandywine Paperboard
EB Brandywine
Creek
Downingtown, PA
104.941 54,62
PA-3
2 .5000
CUW
Downingtown MUA
EB Brandywine
Creek
Downingtown, PA
106.830 54,65
7266-004
0 . 0075
CUW
Shyrock Brothers Inc.
EB Brandywine
Creek
Uwchlan, PA
Brandywine Creek
West Branch
101.708 40,79
PA-5
0.1500
CUW
Embreeville State Hospital
WB Brandywine
Creek
Embreeville, PA
109.172 26,79
7045-004
0 . 0660
CUW
Sealed Air Corp.
Dennis Run
111.660 21,79
6971-004
3 .4407
CUW
Lukens Steel Co.
WB Brandywine
Creek
Coatesville, PA
112.904 19,79
PA-2
2.2500
CUW
City of Coatesville Authority
Rock Run
Rock Run Reservoir
113.526 18,79
450090-002
0 . 0660
CUW
Byran L. Hawthowne
WB Brandywine
Creek
West Brandywine
113.526 18,79
450090-003
0.0113
CUW
Byron L. Hawthorne
WB Brandywine
Creek
West Brandywine
116.636 12,79
PA-1
0.7500
CUW
City of Coatesville Authority
Birch Run
Hibernia Reservoir
Christina River
(nontidal)
89.900 32,13
DE-5
0.1500
CUW
Marvin Hershberger
Christina River
Smalley's Pond Headwaters
89.900 32,13
DE-05
4 . 0000
CUW
United Water Delaware
Christina River
Smalley's Pond WTP
Red Clay Creek
92.386 43,30
DE-2
0.6750
CUW
Hercules Research Center
Red Clay Creek
Woodale, DE
93.665 43,32
DE-4
0 . 0225
CUW
Samuel Beard
Red Clay Creek
Wilmington, DE
98.780 43,40
DE-1
2.2500
CUW
National Vulcanized Fiber
Red Clay Creek
Yorklyn, PA
Red Clay Creek West Branch
103.631 31,43
450088-002
0 . 0004
CUW
J.H. Thompson, Inc.
WB Red Clay Creek
New Garden, PA
White Clay Creek
88.557 40,18
DE-04
12 . 0000
CUW
United Water Delaware
White Clay Creek
Stanton WTP
95.032 28,18
DE-15
1. 0000
CUW
Curtis Paper
White Clay Creek
Newark, DE
95.680 27,18
DE-03
0 . 0001
CUW
City of Newark
White Clay Creek
Papermill WTP
White Clay Creek
East Branch
104.020 19,26
7227-005
0 . 0239
CUW
Laurel Valley Farms
EB White Clay
Creek
New Garden, PA
-------
Table 7-8. Atmospheric dry deposition rates used in Christina River Basin EFDC model.
Parameter
Deposition Rate
(g/m2/day)
Parameter
Deposition Rate
(g/m2/day)
Refractory Part. Organic Carbon
0.000387
Refractory Part. Organic Nitrogen
0.000530
Labile Part. Organic Carbon
0.000387
Labile Part. Organic Nitrogen
0.000530
Dissolved Organic Carbon
0.000773
Dissolved Organic Nitrogen
0.000771
Dissolved Organic Phosphorus
0.000054
Ammonia Nitrogen
0.000214
Orthophosphate
0.000019
Nitrate+Nitrite Nitrogen
0.000393
Available Silica
0.000247
Table 7-9. Atmospheric wet deposition concentrations used in Christina River
Basin EFDC mode
Parameter
Concentration
(mg/L)
Parameter
Concentration
(mg/L)
Refractory Part. Organic Carbon
0.325
Refractory Part. Organic Nitrogen
0.0
Labile Part. Organic Carbon
0.325
Labile Part. Organic Nitrogen
0.0
Dissolved Organic Carbon
0.650
Dissolved Organic Nitrogen
0.140
Dissolved Organic Phosphorus
0.045
Ammonia Nitrogen
0.222
Orthophosphate
0.016
Nitrate+Nitrite Nitrogen
0.332
Available Silica
0.0
7-12
7 - Loads to the System
-------
B8
B1
Run
?od,ey^
B6
R)1
West
PA
MD
C6
C5
Figure 7-1. Watershed delineation for HSPF model of Christina Basin.
7 - Loads to the System
7-13
-------
This page intentionally left blank
7-14 7 - Loads to the System
-------
8 - DELAWARE RIVER BOUNDARY CONDITIONS
Tides were specified at the north and south boundaries in the Delaware River based on the
astronomical harmonic constants for the NOAA subordinate tide stations at Reedy Point, Delaware
(south boundary) and Chester, Pennsylvania (north boundary). The predicted tides from the harmonic
constants will not include any low-frequency influences due to storms or regional low pressure
conditions.
The specification of boundary conditions was required at the model north and south interface
with the Delaware River. The EFDC water quality model accommodates 21 boundary variables, each
specified in an individual time-series data file of concentrations (Table 8-1). Advective boundary
conditions in the Peconic Estuary model were of the "upwind" type. Evaluation of the boundary
concentration depended on the direction of flow at the boundary. When flow was out of the model, the
boundary concentration was assigned the concentration in the model cell immediately upstream of the
boundary. When the tidal flow was into the model, the boundary concentration was assigned a specified,
time-varying value representative of conditions outside the model domain. To estimate recirculation at
the boundary near the time of flow reversal from outgoing to incoming tide, the last outgoing
concentration at the boundary is used as the incoming concentration for a certain amount of time
specified by the user. This concentration linearly approaches the specified outside boundary
concentration over that time period. For the Christina River model, the recirculation time interval was
specified as 60 minutes based on experience gained from previous water quality model applications of
the EFDC model.
Delaware River boundary conditions for salinity, temperature, total suspended sediment, algae,
organic carbon, dissolved oxygen, nitrogen, phosphorus, silica, and fecal coliform bacteria were specified
based on available STORET data at stations in the Delaware River. The boundary time-series were
created using observations that were averaged by month over the simulation period. If data for a
parameter were not available for any given month, then the long-term average (over the period 1988-
1998) for that month was used instead. The boundary conditions for two parameters, unavailable silica
and COD, were set to constant values because no information was available to produce a time-varying
boundary. The boundary condition for unavailable silica was set to 0.10 mg Si/L based on the value used
for the Long Island Sound Study model (HydroQual 1991). The boundary condition for COD was set to
a nominal value of 1.0 mg/L. Total active metal was not included in the simulation. The time-series
boundary conditions for each parameter are shown in Figures 8-1 to 8-7 for the calibration period.
8 - Delaware River Boundary Conditions
8-1
-------
Table 8-1. Specified boundary condition parameters in EFDC water quality model.
1) cyanobacteria (CYA)
2) diatom algae (DIA)
3) green algae (GRN)
4) refractory particulate organic carbon (RPC)
5) labile particulate organic carbon (LPC)
6) dissolved organic carbon (DOC)
7) refractory particulate organic phosphorus (RPP)
8) labile particulate organic phosphorus (LPP)
9) dissolved organic phosphorus (DOP)
10) total orthophosphate (P4T)
11) refractory particulate organic nitrogen (RPN)
(12) labile particulate organic nitrogen (LPN)
(13) dissolved organic nitrogen (DON)
(14) ammonia nitrogen (NH4)
(15) nitrate nitrogen (N03)
(16) unavailable biogenic silica (SUU)
(17) available dissolved silica (SAA)
(18) chemical oxygen demand (COD)
(19) dissolved oxygen (DOO)
(20) total active metal (not simulated) (TAM)
(21) fecal coliform bacteria (FCB)
8-2
8 - Delaware River Boundary Conditions
-------
10
Boundary Conditions for EFDC Christina River Model
0 30 60 90 120 150 180 210 240 270 300 330 360
Julian Day (1997)
Boundary Conditions for EFDC Christina River Model
10
8
6
4
2
0
0
30
60
90
120 150 180 210 240 270 300 330 360
Julian Day (1997)
Boundary Conditions for EFDC Christina River Model
10
8
6
4
2
0
0
30
60
90
120 150 180 210 240 270 300 330 360
Julian Day (1997)
Figure 8-1. Boundary concentrations for CYA, DIA, and GRN algae.
8 - Delaware River Boundary Conditions
8-3
-------
Boundary Conditions for EFDC Christina River Model
Julian Day (1997)
Boundary Conditions for EFDC Christina River Model
1° i 1 1 1 1 1 1 1 1 1 1 1 r
Julian Day (1997)
Boundary Conditions for EFDC Christina River Model
90 120 150 180 210 240 270 300 330 360
Julian Day (1997)
Figure 8-2. Boundary concentrations for RPC, LPC, and DOC.
8-4
8 - Delaware River Boundary Conditions
-------
Boundary Conditions for EFDC Christina River Model
0.8
0.2
30 60 90 120 150 180 210 240 270 300 330 360
Julian Day (1997)
Boundary Conditions for EFDC Christina River Model
0.8
0.2
30 60 90 120 150 180 210 240 270 300 330 360
Julian Day (1997)
Boundary Conditions for EFDC Christina River Model
0.8
0.2
30 60 90 120 150 180 210 240 270 300 330 360
Julian Day (1997)
Figure 8-3. Boundary concentrations for RPP, LPP, and POC.
8 - Delaware River Boundary Conditions
8-5
-------
Boundary Conditions for EFDC Christina River Model
~i 1 1 1 1 1 1 1 1 1 1 r
0.8 -
0.2 -
0 i I I I I I I I I I I I i_
0 30 60 90 120 150 180 210 240 270 300 330 360
Julian Day (1997)
Boundary Conditions for EFDC Christina River Model
~i 1 1 1 1 1 1 1 1 1 1 r
4 -
0 i I I I I I I I I I I I i_
0 30 60 90 120 150 180 210 240 270 300 330 360
Julian Day (1997)
Boundary Conditions for EFDC Christina River Model
~i 1 1 1 1 1 1 1 1 1 1 r
4 -
0 i I I I I I I I I I I I i_
0 30 60 90 120 150 180 210 240 270 300 330 360
Julian Day (1997)
Figure 8-4. Boundary concentrations for P4T, RPN, and LPN.
8-6
8 - Delaware River Boundary Conditions
-------
Boundary Conditions for EFDC Christina River Model
~i 1 1 1 1 1 1 1 1 1 1 r
Julian Day (1997)
Boundary Conditions for EFDC Christina River Model
~i 1 1 1 1 1 1 1 1 1 1 r
0 i i i i i i i i i i i i i_
0 30 60 90 120 150 180 210 240 270 300 330 360
Julian Day (1997)
Boundary Conditions for EFDC Christina River Model
90 120 150 180 210 240 270 300 330 360
Julian Day (1997)
Figure 8-5. Boundary concentrations for DON, NH4, and N03.
8 - Delaware River Boundary Conditions
8-7
-------
-I 1.0-
S 0.
Boundary Conditions for EFDC Christina River Model
1 1 1 1 1 1 1 1 1 1 1 r
0 i i i i i i i i i i i i i_
0 30 60 90 120 150 180 210 240 270 300 330 360
Julian Day (1997)
Boundary Conditions for EFDC Christina River Model
~i 1 1 1 1 1 1 1 1 1 1 r
0 i i i i i i i i i i i i i_
0 30 60 90 120 150 180 210 240 270 300 330 360
Julian Day (1997)
Boundary Conditions for EFDC Christina River Model
1° i 1 1 1 1 1 1 1 1 1 1 1 r
90 120 150 180 210 240 270 300 330 360
Julian Day (1997)
Figure 8-6. Boundary concentrations for SUU, SAA, and COD.
8-8
8 - Delaware River Boundary Conditions
-------
Boundary Conditions for EFDC Christina River Model
Julian Day (1997)
0.8
0.2
Boundary Conditions for EFDC Christina River Model
30 60 90 120 150 180 210 240 270 300 330 360
Julian Day (1997)
Boundary Conditions for EFDC Christina River Model
Julian Day (1997)
Figure 8-7. Boundary concentrations for DOO, TAM, and FCB.
8 - Delaware River Boundary Conditions
8-9
-------
This page intentionally left blank
8-10 8- Delaware River Boundary Conditions
-------
9 - MODEL CALIBRATION
Model calibration involves the adjustment of certain model input quantities in an attempt to
achieve a specified level of model performance. An extensive set of field data were gathered, processed,
and displayed for modeling hydrodynamics and water quality transport in the Christina River Basin. The
data set included database files containing more than 40,000 records at about 200 stations scattered
throughout the interior of the basin as well as in the Delaware River itself. This section presents the
results of the calibration of the EFDC hydrodynamic and water quality model. Parameters considered for
calibration include flow rate, tidal surface elevation, chlorides, and a suite of water quality parameters.
9.1 Computational Grid
The basic equations in EFDC were solved using the finite-difference method. The grid was
designed to resolve velocity shears both axially and laterally, and at the same time allow a time step
suitable for efficient computation. Solutions to the hydrodynamics were obtained using a 60-second time
step. The spatial domain of the study area was divided into a grid of discrete cells. To achieve close
conformance of the grid to the estuary geometry, the cells in the Delaware River were represented using
curvilinear horizontal grid cells constructed using an orthogonal mapping procedure (Ryskin and Leal
1983) to form a 2-D grid domain. The cells in the narrow tidal and nontidal streams were represented in
a 1-D Cartesian coordinate system (see Figure 9-1). To obtain adequate resolution in the streams,
longitudinal cells were configured to lengths ranging from 500 to approximately 1,000 meters. Cell
widths were adjusted according to estimated stream channel widths. Velocities were computed on the
boundaries between cells, and temperature, salinity, and density were computed at the center of each cell.
The numerical grid consisted of 406 cells in the horizontal plane and a single vertical layer. A single
layer was chosen because the estuary and streams are well mixed, thereby implying that stratification
would not be an issue. In addition, field data available from STORET and from Davis (1998) did not
distinguish vertical sample depths.
9.2 Model Configuration
The general procedure for application of the EFDC model to the Christina River Basin followed
a sequence of steps beginning with model set up or configuration and continued through model execution
of the calibration time period. Model configuration involved the construction of the horizontal grid for
the waterbodies in the basin, interpolation of bathymetric data to the grid, construction of EFDC input
files, and compilation of the FORTRAN source code with appropriate parameter specification of array
dimensions. The model included 120 point source discharges (see Figure 9-1) and 28 consumptive use
water withdrawals (see Figure 9-2).
9 - Model Calibration
9-1
-------
The numerical model domain includes the tidal Delaware River from Reedy Point on the south to
Chester on the north. Both the tidal and nontidal Christina River are included in the model. The lower
Christina River is directly connected to the Delaware River. The nontidal Christina River is connected to
the tidal portion by a dam control structure at Smalley's Pond. The tidal Brandywine Creek is connected
to the Christina River by means of an inlet control structure. The tidal White Clay Creek is also
connected to the Christina River via an inlet control structure. There are 27 control structures in the
model, 4 inlet structure pairs, 11 confluence connections, and 12 control structures. These control
structures represent low head dams and abrupt bottom elevation changes at bridge crossings as well as
larger dams. The locations of the control structures are shown in Figure 9-1 and are described below in
Table 9-1.
Table 9-1. Hydraulic control structures in Christina River Basin EFDC model.
Structure
ID
Description
Structure
ID
Description
1
Dam at Smalley's Pond, Christina River
18
Fall line, lower Red Clay Creek
2,3
Tidal inlet, mouth of Brandywine Creek
19
Connection, E. Br. Red Clay to Red Clay Creek
4,5
Tidal inlet, Nonesuch Cr.-Christina River
20
Connection, W. Br. Red Clay to Red Clay Creek
6,7
Tidal inlet, Nonesuch Cr.-Christina River
21
Fall line, lower White Clay Creek
8,9
Tidal inlet, mouth of White Clay Creek
22
Bridge culvert, Harmony Rd., White Clay Creek
10
Low dam, lower Brandywine Creek
23
Low dam, White Clay Creek
11
Submerged weir, Lenape, Brandywine Creek
24
Bridge culvert, Hopkins Road, White Clay Creek
12
Connection, Brandywine Cr to East Branch
25
Connection, E. Br. White Clay to White Clay Creek
13
Submerged weir, Embreeville, E.Br. Brandywine
26
Connection, W. Br. Christina to Christina River
14
Submerged weir, Mortonville, E.Br. Brandywine
27
Connection, Little Mill Creek to Christina River
15
Submerged weir, South Coatesville, E.Branch
28
Connection, Burroughs Run to Red Clay Creek
16
Dam, Icedale Lake, E.Br. Brandywine Creek
29
Connection, Mill Creek to White Clay Creek
17
Connection, Buck Run to E.Br. Brandywine Cr.
30
Connection, Pike Creek to White Clay Creek
31
Connection, Muddy Run to White Clay Creek
9.3 Calibration Period
The time period for model calibration, May 1 to September 21, 1997, was selected because it
included the detailed field survey period in which water quality data were collected by Davis (1998) as
well as other monitoring data from DNREC, PADEP, USGS, and others. During the August and early
September 1997 time period, stream flow throughout the basin was near the 7Q10 flow rate. Data for
comparison to the model water quality results for stream reaches, other than those sampled by Davis
9-2
9 - Model Calibration
-------
(1998) and three USGS locations, were generally monitored on a bimonthly basis during the calibration
period.
9.4 Hydrodynamic and Hydraulic Calibration
Calibration of the hydrodynamic model involved adjustment of the open boundary water surface
elevation forcing, the bottom boundary roughness, and local bathymetry. The open boundary tidal
elevation, specified as a linear variation of the tidal constituent amplitudes and phases, was adjusted until
predicted amplitudes and phases agreed with those obtained from an analysis of the USGS tide gage
records at the Port of Wilmington and Newport. The model was executed for a period of 143 days from
May 1 to September 21, 1997. The model results were then compared against available observations at
interior monitoring stations. Comparisons were made for tide height and phase, flow rate, and chloride
concentration at various locations in the model.
9.4.1 Calibration of Tide Elevation
Calibration of the model with respect to water surface elevation was accomplished by analysis of
observed and model predicted time-series data at two interior tide stations. For tidal waters, least squares
harmonic analysis is the most commonly utilized procedure (Oey, Mellor, and Hires 1985; Cheng et al.
1993; Shen et al. 1999). Tide elevation data were obtained from the USGS tide stations on the Christina
River at the Port of Wilmington near the mouth and at Newport about 7.0 miles upstream of the mouth.
These data were compared with surface elevations computed by the model at cell 56,13 (Port of
Wilmington) and 45,13 (Newport). The time-series of tide elevations for the month of August 1997 for
both the field data and model results were subjected to a harmonic analysis. The five most important
astronomical harmonic constituents (M2, S2, N2, Kl, and 01) were computed for both the field data and
model simulation results. The harmonic analysis results, shown in Table 9-2, indicate the model is in
good agreement with the measured tide data for both amplitude and phase. The model-data amplitudes
for the M2 harmonic constituent agree within 5 cm (6%) and the phases agree to within 4 degrees (3%).
Time-series graphs (Figure 9-3) of the observed and model tide elevations at both the Port of Wilmington
and Newport covering a 15-day period (August 1 - 15, 1997) provide a visual means of assessing the skill
of the model in simulating tidal elevations. The model tides are forced at the north and south boundaries
in the Delaware River based on the NOAA predictions at the Reedy Point and Chester subordinate
stations (NOAA 1998). These predictions do not consider low-frequency phenomenon caused by
regional low pressure systems or storms that will be found in the signal of the tide data collected at the
two USGS tide stations on the Christina River.
9 - Model Calibration
9-3
-------
Table 9-2. Harmonic analysis of tides at Port of Wilmington and Newport.
Harmonic Constant
Port of Wilmington
Newport
Amplitude (m)
Phase (degrees)
Amplitude (m)
Phase (degrees)
M2 - observed
0.7594
130.382
0.6901
153.634
M2 - model
0.7135
134.180
0.6768
155.560
Difference
0.0459
-3.798
0.0133
-1.926
S2 - observed
0.0894
20.621
0.0900
36.374
S2 - model
0.1001
30.806
0.0890
59.180
Difference
-0.0107
-10.185
0.0010
-22.806
N2 - observed
0.1271
323.153
0.1275
345.054
N2 - model
0.1383
336.181
0.1240
3.603
Difference
-0.0112
-13.028
0.0035
-18.549
K1 - observed
0.0802
174.059
0.0615
184.740
K1 - model
0.0633
178.335
0.0606
190.948
Difference
0.0169
-4.276
0.0009
-6.208
Ol - observed
0.0626
316.879
0.0546
332.386
Ol - model
0.0546
326.765
0.0514
337.937
Difference
0.0080
-9.886
0.0032
-5.551
9.4.2 Hydraulic Flow Balance
Calibration of the hydraulic flow balance in the model system was determined by comparing the
model and observed hydrograph at 12 USGS stream gage locations in the Christina River Basin for the
calibration period. Estimates of unit discharge rates (cfs per square mile) were determined for each of
the 39 subwatersheds for each day in the calibration period. The daily flow rates for each subwatershed
were then distributed uniformly among each of the grid cells within a subwatershed, except for
headwater grid cells, which were assigned a flow rate in accordance with the contributing area to that
cell. The model was then run for the 143-day period, and the flow rate at the appropriate gage location
was compared with the model results. The model flow rates compared reasonably well with the daily
average flows at the stream gages (see Figures 9-4 to 9-7). The purpose of the hydraulic flow balance
comparison to the USGS gage data was to determine whether the runoff rates from the contributing
subwatersheds were properly apportioned to the model grid cells. Normally, a watershed runoff model
would be used to provide flows to the receiving water model. However, the calibrated HSPF watershed
model will not be available for a few more years. The model hydrograph agrees well with the stream
gage data during periods between storm runoff events. At certain locations, the model tends to
underpredict the peak flow rates of the storm events. The use of a watershed runoff model in the future
will likely improve the peak flow calibration because the timing of the peak runoff from each
subwatershed will be taken into account (a procedure that was not possible in the present model
application).
9-4
9 - Model Calibration
-------
9.4.3 Water Depth and Stream Velocity
Measurements of flow, water depth, and stream velocity were made at eight locations during the
August 1997 field survey (Davis 1998). The field measurements were made on the following dates: East
Branch Brandywine Creek (08/12 - 08/14/97), West Branch Brandywine Creek (08/19 - 08/20/97), West
Branch Red Clay Creek (08/05 - 08/07/97 and 08/12 - 08/14/97), and East Branch White Clay Creek
(08/26 - 08/28/97). A comparison of these measurements with the model results at the appropriate grid
cell location is given in Table 9-3.
Table 9-3. Model-data comparison of velocity, flow, and geometry (August 1997 data).
Stream Reach
EFDC
Cell
Velocity (fps)
Depth (ft)
Flow (cfs)
Rect. Channel Width (ft)
Field
EFDC
Field
EFDC
Field
EFDC
Field
EFDC
East Branch Brandywine Creek
54,61
0.33
0.48
0.82
0.87
14.5
25.6
53.6
52.5
East Branch Brandywine Creek
54,56
0.85
0.56
1.02
1.11
34.3
34.5
39.6
52.5
West Branch Brandywine Creek
19,79
0.40
0.41
1.09
0.94
9.5
14.9
45.0
42.6
West Branch Brandywine Creek
26,79
0.41
0.36
0.70
0.82
32.0
32.9
111.5
111.5
East Branch White Clay Creek
19,31
0.44
0.40
0.93
0.96
5.30
5.33
13.0
12.8
East Branch White Clay Creek
19,29
0.42
0.41
0.85
0.86
7.35
7.34
20.6
20.3
West Branch Red Clay Creek
29,43
0.35
0.44
0.75
0.78
3.55
3.35
13.5
13.5
West Branch Red Clay Creek
33,43
0.49
0.52
0.90
0.94
5.45
4.92
12.4
12.4
9.4.4 Chloride Concentrations
The ability of a numerical hydrodynamic model to predict the transient distribution of chlorides
or salinity is viewed as the most important measure calibration skill if the ultimate use of the model is
prediction of the transport and fate of dissolved contaminants. Since chloride distribution is a direct
consequence of physical transport by advection and turbulent diffusion, chloride calibration substantiates
advective and diffusive transport on a global flux scale rather than the point scale addressed by water
surface elevation calibration. An acceptable chloride calibration supports the accuracy of global scale
transport even under conditions of marginal verification of the model's ability to predict velocity and
water surface elevation at specific observation points. The model-data comparisons of chloride
concentration for the longitudinal transects representing the stream reaches included in the model are
presented in figures in Appendix A. The "box-and-whisker" data shown on these graphs were obtained
from STORET over the period June 1 to September 30, 1997. The box-and-whisker data points represent
the average, median, 25th percentile, 75th percentile, minimum, and maximum. The sample data collected
during the August 1997 field survey (Davis 1998) are also shown on these graphs as the mean and
9 - Model Calibration
9-5
-------
standard deviation. Chloride concentrations computed by the EFDC model compare well with the
observed data for all stream reaches.
9.5 Water Quality Calibration Results
Each observation in STORET was collected at an instant in time and at a single point in space.
Time scales realistically represented in the EFDC model were determined by time scales of primary
forcing functions: 60-second tidal hydrodynamics, hourly meteorological updates, monthly ocean
boundary conditions, constant nonpoint source concentration estimates, daily nonpoint source flows,
monthly point source loads, and hourly atmospheric wet deposition. The minimum model spatial scales
were determined by the size of the grid cells, ranging from 500 to about 1,000 meters in the longitudinal
direction along the streams. Data for longitudinal transect comparisons were averaged over the period
June 1 to September 30, 1997. The disparity in the temporal and spatial scales between the model and
prototype, especially for the nonpoint and point source loads, meant that individual observations may not
be directly comparable with model prediction at a specific time in a given model grid cell.
Model-data comparisons will be made by means of longitudinal transect plots as well as time-
series plots for the 11 major stream reaches in the study area: Brandywine Creek, East Branch
Brandywine Creek, West Branch Brandywine Creek, Buck Run, Christina River (tidal), Christina River
(nontidal), Red Clay Creek, West Branch Red Clay Creek, White Clay Creek, East Branch White Clay
Creek, and the Delaware River. The transects are delineated in river miles referenced to River Mile 74.9
located at the mouth of the Christina River based on EPA REACH FILE 1 (Table 9-4). Longitudinal
transect plots for each water column parameter are presented in Appendix A arranged by stream reach.
There are 18 transect plots for each reach, representing 18 different water quality parameters. The model
results for the transect plots were averaged over the period August 5 to August 20, 1997. The horizontal
axis of each plot represents the river mile from the mouth of the Christina River measured along the
stream network. The vertical axis represents the water column parameter concentration. The observed
data are shown as "box-and-whisker" symbols indicating the maximum, minimum, 25th percentile, 75th
percentile, mean, and median statistics. The model output results are represented by three lines, the solid
line indicates the mean over the averaging period at a given model grid cell and the two dashed lines
represent the minimum and maximum values simulated over the averaging period.
The time-series plots are provided in Appendix B and cover the entire 143-day calibration period
beginning on May 1 (day 121) and continuing to September 21, 1997 (day 264). The time-series model-
data comparisons were made at 16 monitoring locations on the various stream reaches in the study area
(see Figure 9-8). The concentrations of the nonpoint source loads were considered to be constant
throughout the calibration period, and the loads vary in accordance with the changes in nonpoint source
9-6
9 - Model Calibration
-------
flow rate. The nonpoint source concentrations were based on the summer low-flow monitoring data and
are representative of background water quality conditions. In reality, the concentrations of the various
water quality parameters will vary in relation to storm events and changes in watershed runoff.
Determination of the time-varying nonpoint source concentrations was outside the scope of the present
study, but will be addressed in the future following completion of the HSPF watershed model of the
Christina River Basin.
Table 9-4. Stream reaches included in EFDC Christina River Basin water quality model.
Stream Reach
River Mile at
Mouth
River Mile at
Upstream Extent
Christina River (tidal)
74.9
89.6
Christina River (nontidal)
89.6
103.0
Christina River West Branch
98.5
100.4
Brandywine Creek (main stem)
76.3
95.8
Brandywine Creek East Branch
95.8
113.7
Brandywine Creek West Branch
95.8
120.7
Buck Run
106.6
117.3
Red Clay Creek and East Branch
87.6
104.9
Red Clay Creek West Branch
100.3
104.9
White Clay Creek and Middle Branch
85.6
109.7
White Clay Creek East Branch
99.9
107.1
Delaware River
62.6
86.5
Little Mill Creek
79.8
85.4
Mill Creek
87.9
94.7
Burroughs Run
97.1
100.2
Pike Creek
90.6
95.9
Muddy Run
93.2
96.5
9 - Model Calibration
9-7
-------
9.5.1 Brandywine Creek Main Stem Water Quality Calibration Results
The transect plots for all water quality parameters for the main stem of Brandywine Creek are
shown in Figures A01 - A03, and the time-series plots at stations 104021 and WQN0105 are shown in
Figures B01 - B04. The conservative constituents (chlorides and TSS) match the observed data very well
in both the transect and time-series plots. The time-series of TSS shows little variation because a
constant nonpoint source concentration was used for the entire time period, whereas in reality the TSS
concentrations would increase during storm runoff events. The grab samples of dissolved oxygen all lie
within the minimum and maximum range computed by the model for both the transect and time-series
views. The observed organic carbon indicates an increasing trend in the downstream direction that is
stronger than computed by the model. This is likely due to missing sources of organic carbon. Also, the
total phosphorus and dissolved orthophosphate indicate an increasing concentration in the downstream
direction that is not simulated in the model. The nitrogen species (total nitrogen, ammonia nitrogen, and
nitrate nitrogen) all match the observations along the transect quite well.
9.5.2 Brandywine Creek East Branch Water Quality Calibration Results
The transect plots for all water quality parameters for the East Branch Brandywine Creek are
shown in Figures A04 - A06, and the time-series plots at station 01480870 are shown in Figures B05 and
B06. The Downingtown WWTP (PA0026531) discharges at river mile 103.7, which accounts for the
spike in concentrations of various water quality parameters at that location. Along the transect, all water
quality parameters are in agreement with observations. The time-series plots also indicate the model is in
agreement with observations with the exception of fecal coliform bacteria, which is underpredicted in the
model. This is most likely due to nonpoint sources that are not accounted for in the model.
9.5.3 Brandywine Creek West Branch Water Quality Calibration Results
The transect plots for all water quality parameters for the West Branch Brandywine Creek are
shown in Figures A07 - A09, and the time-series plots at station 01480617 are shown in Figures B07 and
B08. The South Coatesville WWTP (PA0036987) and Coatesville City WWTP (PA0026859) discharge
at river mile 110.4 and 111.0, respectively. The spike in the concentrations of various water quality
parameters is due to these two discharges. The model agrees with the observed data very well for
chlorides, TSS, dissolved oxygen, chlorophyll-a, organic carbon, total nitrogen, ammonia nitrogen, and
nitrate nitrogen. The model somewhat underpredicts the phosphorus species downstream of the two
aforementioned WWTPs. The reason for this is not clear because phosphorus was a measured parameter
reported on the discharge monitoring records at these two WWTPs and was not based on default
estimates. Fecal coliform bacteria (see Figure B08) simulated by the model are about an order of
magnitude less than the observations. This is likely due to nonpoint sources that are not accounted for in
the model.
9-8
9 - Model Calibration
-------
9.5.4 Buck Run Water Quality Calibration Results
The transect plots for all water quality parameters for Buck Run are shown in Figures A10 - A12.
No time-series plots are presented for Buck Run. No observed data were available for Buck Run during
the calibration period, so calibration cannot be assessed in this stream reach.
9.5.5 Christina River (Tidal) Water Quality Calibration Results
The transect plots for all water quality parameters for the tidal Christina River are shown in
Figures A13 to A15, and the time-series plots at stations 106291 and 106021 are shown in Figures B23 to
B26. The lower portion of the Christina River is strongly influenced by the Delaware River because of
tidal excursion. The transect plots indicate the model agrees well with the data for all water quality
parameters with the exception of total nitrogen, which is slightly high in the model. The time-series plots
also indicate similar good model-data agreement. The two observations of chlorophyll-a at station
106021 reach levels of about 108 ug/L in late May and 67 ug/L in mid-July whereas the model computes
a maximum concentration of about 40 ug/L in late May. One possibility for this discrepancy may be
influences from nearby Churchman's Marsh, which is not included in the model.
9.5.6 Christina River (Nontidal) Water Quality Calibration Results
The transect plots for all water quality parameters for the nontidal Christina River are shown in
Figures A16 - A18, and the time-series plots at station 106031 are shown in Figures B27 and B28. Along
the transect, all water quality parameters agree with the observations quite well. The spike in chlorides
concentration at river mile 98.1 is due to the West Branch Christina River, which carries loads from the
two Maryland WWTPs (MD0022641 and MD0065145).
9.5.7 Red Clay Creek and East Branch Water Quality Calibration Results
The transect plots for all water quality parameters for Red Clay Creek and East Branch Red Clay
Creek are shown in Figures A19 to A21, and the time-series plots at stations 103031, WQN0150, and
RCEB04 are shown in Figures B09 to B14. The West Branch Red Clay Creek enters at river mile 100.3
and accounts for the spikes in concentration that are evident in a number of the transect plots. Overall,
the model does a reasonable job of simulating the observations for both the transect and time-series
views. At stations 103031 and WQN0150, the total phosphorus and dissolved orthophosphate data
indicate an increasing trend from May to September (see Figures B10 and B12). This tends to support
the hypothesis that the primary sources for phosphorus may be from a relatively steady-state source (i.e.,
point source or groundwater source) because as the stream flow decreases in the summer months, the
concentration of phosphorus is increasing.
9 - Model Calibration
9-9
-------
9.5.8 Red Clay Creek West Branch Water Quality Calibration Results
The transect plots for all water quality parameters for West Branch Red Clay Creek are shown in
Figures A22 - A24, and the time-series plots at station RCWR2 are shown in Figures B15 and B16. The
Kennett Square WWTP (PA0024058) discharges at river mile 103.9, accounting for the spike in
concentrations at that location. The simulated concentrations of all water quality parameters agree with
the observed data along the transect. Station RCWR2 is Reach 2 of the August 1997 study (Davis 1998).
Dissolved oxygen is controlled by reaeration, sediment oxygen demand, nitrification, denitrification,
decay of organic substances, photosynthesis of algae, and respiration of algae. The model represents
both the mean and the range of dissolved oxygen very well. The observations indicate a strong oxygen
sag occurring about 1.0 mile downstream of the Kennett Square WWTP where the daily minimum
dissolved oxygen decreases from about 8.0 mg/L above the WWTP to about 1.9 mg/L at the maximum
sag location. In the model, the minimum dissolved oxygen decreases from 7.5 mg/L above the WWTP to
a value of 1.7 mg/L below the WWTP discharge. At river mile 103.1, the observed data indicate
chlorophyll-a levels as high as 42 ug/L, whereas the model indicates concentrations of about 7 ug/L. It is
possible that the measured chlorophyll-a may have contained periphyton cells that detached from the
stream bottom.
9.5.9 White Clay Creek and Middle Branch Water Quality Calibration Results
The transect plots for all water quality parameters for White Clay Creek and Middle Branch
White Clay Creek are shown in Figures A25 - A28, and the time-series plots at stations 105151 and
WQN0149 are shown in Figures B17 - B20. The transect plots indicate good model-data agreement for
all parameters except phosphorus. Downstream of river mile 103, the monitoring data indicate total
phosphorus concentrations in the 0.1 to 0.6 mg/L range, whereas the model computes concentrations of
about 0.1 mg/L. The reason for this discrepancy is not clear but may be due to inadequate nonpoint
source loadings since the only two NPDES point sources downstream of mile 103 are small (Avon Grove
School District and FMC Corp.).
9.5.10 White Clay Creek East Branch Water Quality Calibration Results
The transect plots for all water quality parameters for White Clay Creek East Branch are shown
in Figures A28 - A30, and the time-series plots at stations WCER2 are shown in Figure B21. Station
WCER2 is Reach 2 from the August 1997 study (Davis 1998). The Avondale Borough WWTP
(PA0025488) is the largest point source on this stream and discharges at river mile 106.6. The model
results are in reasonable agreement for all parameters along the transect and in the time-series views.
9-10
9 - Model Calibration
-------
9.5.11 Delaware River Water Quality Calibration Results
The transect plots for all water quality parameters for the Delaware River are shown in Figures
A31 - A33. The model indicates reasonable agreement for all water quality parameters. One surprising
result was the simulated dissolved oxygen sag at river mile 82 that reaches a minimum value of about 2.1
mg/L. The model results can not be validated at that location since no observed data were available.
9.5.12 Muddy Run and Pike Creek Water Quality Calibration Results
Time-series plots for all water quality parameters for Muddy Run (station 105131) and Pike
Creek (station 105101) are shown in Figures B29 to B32. The model results agree well with the
observations for all parameters with the exception of an apparent algae bloom in mid-July (day 195) at
the Muddy Run station. The data indicate a chlorophyll-a concentration of about 11 ug/L, whereas the
model computes about 2.5 ug/L. However, in late May and mid-September the model agrees very well
with the chlorophyll-a measurements.
9.6 Diel Dissolved Oxygen Calibration Results
An important feature of the Christina River Basin water quality model is the ability to compute
the daily dissolved oxygen range as well as the daily average value. Water quality standards for
dissolved oxygen in the Christina River Basin must meet two criteria, one for the daily average and one
for the daily minimum concentration. The data available for model calibration included diel dissolved
oxygen at a number of locations. The August 1997 survey (Davis 1998), used automatic monitors to
record dissolved oxygen concentrations at 15-minute intervals over 2-day periods at locations on the East
Branch Brandywine Creek, West Branch Brandywine Creek, West Branch Red Clay Creek, and East
Branch White Clay Creek. In addition, the USGS collected diel dissolved oxygen data at three gages for
the entire May-September 1997 calibration period: (1) 01480870 at Downingtown on the East Branch
Brandywine Creek, (2) 01480617 at Modena on the West Branch Brandywine Creek, and (3) 01481000
at Chadds Ford on the main stem Brandywine Creek.
Achieving the proper range in daily dissolved oxygen is primarily a function of the community
periphyton biomass available at a given model grid cell. The periphyton growth and basal metabolism
rates as well as the growth rate density limitation parameters were adjusted to simulate the periphyton
biomass needed to achieve reasonable daily DO ranges at the monitoring sites during August 1997, the
critical low-flow period. The monitored daily minimum, maximum, and average dissolved oxygen
concentrations at the three USGS gage locations are shown in Figure 9-9 along with the model results. It
is evident that the model does a reasonable job of simulating the daily DO range at these three locations
during the month of August. The daily minimum, maximum, and average water temperatures at the three
USGS gage locations and the simulated model temperatures are shown in Figure 9-10. Again, the model
9 - Model Calibration
9-11
-------
is in good agreement with the data for the entire 5-month simulation period. The time-series of
periphyton biomass at the three gage locations is shown in Figure 9-11. The periphyton biomass has
been displayed in units of ug/L of chlorophyll-a for comparison with the floating chlorophyll-a
concentrations. Thus, the periphyton biomass (50 - 1,000 ug/L) is as much as two orders of magnitude
greater than the floating chlorophyll-a biomass (3-10 ug/L) in these stream reaches. This means that an
off-the-shelf model, such as WASP or QUAL2E, which does not include a periphyton state variable,
would not be able to simulate the diel DO range by use of floating chlorophyll-a alone.
No measurements of periphyton biomass were available for the 1997 calibration period.
However, in 1985 a study was conducted on the East Branch Brandywine Creek and periphyton biomass
was measured at six locations during the period July 15 to August 7, 1985 (Knorr and Fairchild 1987).
The conditions in the stream may have been different between 1985 and 1997 because of changes in
wastewater treatment and the magnitude of nutrient loads reaching the stream. Nonetheless, these
periphyton measurements represent the only data available for assessing the validity of the model
periphyton calculations. A comparison of the periphyton biomass measured in 1985 with the biomass
computed by the model is given in Table 9-5. The model biomass for August 4, 1999 (day 216) is
reported in the table and is in reasonable agreement with the 1985 biomass measurements. The model
periphyton biomass at sites 1 and 3 upstream of the Downingtown WWTP (river mile 103.7) is somewhat
less than reported in 1985. The periphyton biomass computed by the model at sites 4 and 5 downstream
of the Downingtown WWTP is slightly higher than the 1985 measurements, and at site 6 the model
periphyton biomass is within the range reported in 1985.
Table 9-5. Comparison of model periphyton with 1985 measurements (Knorr and Fairchild 1987).
Site
ID
River
Mile
1985 Periphyton Biomass
(ug chlorophyll-a / cm2)
EFDC
Grid Cell
Model Periphyton
(ug chlorophyll-a / L)
Water
Depth (m)
Model Periphyton Biomass
(ug chlorophyll-a / cm2)
1
109.3
6.2-10.2
54,69
70
0.30
2.1
2
NA
8.0-16.5
NA
NA
NA
NA
3
106.2
8.5 - 13.0
54,64
160
0.33
5.3
4
102.4
9.0-17.0
54,58
550
0.36
19.8
5
101.2
11.5-21.0
54,56
700
0.37
25.9
6
96.1
8.0-14.3
54,48
240
0.35
8.4
As stated in Section 4.13, periphyton growth is limited by a number of factors including the
availability of nitrogen, phosphorus, and solar radiation, as well as by temperature, stream velocity, and
biomass density limitations. Time-series plots of each of these limitation factors are presented in Figures
9-12 to 9-16 for five locations. All five locations indicate that there is an abundance of nitrogen available
9-12
9 - Model Calibration
-------
and that parameter is not limiting periphyton growth. Phosphorus is the more limiting of the two
nutrients according to the model calculations.
The model-data diel dissolved oxygen comparisons for the automatic monitors deployed in the
August 1997 survey (Davis 1998) are presented in Figure 9-17 (East Branch Brandywine Creek);
Figure 9-18 (West Branch Brandywine Creek); Figures 9-19 and 9-20 (West Branch Red Clay Creek);
and Figures 9-21 and 9-22 (East Branch White Clay Creek). Considering that the model simulation
began 3 months prior to the August time period, the fact that the diel dissolved oxygen agrees so well
with the monitor data is noteworthy. At most monitor locations, the model agrees with both the
minimum recorded dissolved oxygen and the dissolved oxygen range. The magnitude of the diel
dissolved oxygen range is a very localized phenomenon related to sunlight and periphyton biomass. The
resolution of the model grid and the temporal resolution of the various nutrient sources as well as the lack
of canopy shading information are all possible contributors to deviations in the model versus monitored
diel dissolved oxygen. Nonetheless, even with these sources of uncertainty in model resolution, the diel
dissolved oxygen computed by the model agrees very favorably with the observations measured by the
automatic monitors.
9.7 Sediment Oxygen Demand and Benthic Nutrient Flux Rates
The need for a predictive benthic sediment submodel for water quality modeling projects has
been apparent for some time. When using a water quality model for management scenario analysis, one
of the biggest sources of uncertainty involves what to use for the future sediment flux rates after a
proposed management control has been implemented. The predictive sediment submodel in EFDC helps
address this uncertainty with two fundamental capabilities: (1) the ability to predict effects of
management alternatives on sediment-water exchange processes and (2) the ability to predict the time
scale for alterations in the sediment-water exchange processes. To meet these requirements, a predictive
sediment process model was incorporated into the EFDC model framework and was based on DiToro and
Fitzpatrick (1993). The sediment submodel is driven by net settling of organic matter from the water
column to the sediments. In the benthos, the sediment submodel simulates the decay (diagenesis) of
organic matter, which produces oxygen demand and inorganic nutrients. Oxygen demand takes three
paths out of the sediments: (1) export to the water column as chemical oxygen demand, (2) oxidation at
the sediment-water interface as sediment oxygen demand, or (3) burial to a deep, inactive sediment layer.
The inorganic nutrients produced by diagenesis can take two pathways out of the bottom sediment: (1)
release back to the overlying water column or (2) burial to the deep, inactive sediment layer.
In the predictive sediment submodel, benthic sediments are represented as two layers with a total
depth of 10 cm. The upper benthic layer is in contact with the water column and may be oxic or anoxic
9 - Model Calibration
9-13
-------
depending on the dissolved oxygen concentration in the water. The lower benthic layer is permanently
anoxic. The thickness of the upper benthic layer is determined by the penetration of oxygen into the
sediments, and at its maximum thickness, the oxic layer depth is a small fraction of the total thickness.
The sediment submodel consists of three basic processes:
• Particulate organic matter settles from the water column to the sediments. Because of the
negligible thickness of the upper benthic layer, deposition proceeds from the water column
directly to the lower anoxic layer.
• Within the lower layer, organic matter is subject to decay (diagenesis).
• The flux of substances produced by diagenesis moves to the upper benthic layer, to the water
column, and to the deep, inactive benthic layer (burial). The flux portion of the sediment
submodel is the most complex. The computation of flux requires consideration of
(1) reactions in both benthic layers, (2) sedimentation from the upper to lower benthic layer
as well as from the lower benthic layer to the deep inactive sediments, (3) particle mixing
between layers, (4) diffusion between layers, and (5) mass transfer between the upper layer
and the water column.
No field data were available during the calibration period to verify the flux rates computed by the
predictive sediment submodel. However, SOD rates were measured in July and August 1996, at three
locations in the tidal Christina River and Brandywine Creek. An SOD rate of 0.5 g/m2/day was used in
the tidal Delaware River in another model study conducted by HydroQual for DRBC and was also
adopted for this study. The simulated SOD rates were converted to rates at 20 °C and are compared with
the measured data in Table 9-6. The relative errors were less than 13% at all locations. Time-series plots
of sediment oxygen demand, benthic ammonia flux, benthic nitrate flux, benthic phosphate flux, benthic
COD flux, benthic silica flux, and sediment temperature at the same 16 monitoring stations used for the
water quality calibration are presented in Appendix C. Transect plots of these sediment flux parameters
for the 11 major stream reaches in the model are presented in Appendix D.
Table 9-6. Model-data comparison of sediment oxygen demand rates (g/m2/day).
Location
Sampling Date
Monitored
SOD at 20 °C
1997 Calibration
Model
SOD at 20 °C
Relative
Error
Christina River at 1-495 bridge
Aug 12, 1996
0.81
0.91
12.9%
Christina River at Newport, Rt. 141 bridge
Jul 10, 1996
1.67
1.56
6.5%
Brandywine Creek, 0.6 mi. from mouth
Aug 12, 1996
1.23
1.19
3.4%
Delaware River (from HydroQual study)
-
0.50
0.46
8.8%
9-14
9 - Model Calibration
-------
West Br. Brandywine Creek
33,'
54,76
33,61
Little Mill Creek
Burroughs Run
43,49
W.Br. Red Clay
54,46
54,'
Mill Creek
Pike Creek
Muddy Run
19,32
43,23
CD
White Clay
Creek
54,20
9,20
54,18
Mid.Br. White Clay
43,17
(2,3)
43,15
54,15
Christina River
West Br.
Inlet control structure
Flow control structure
PA52+51 NPDES Point Source Discharge
Figure 9 — 1. Schematic of EFDC model of Christina River Basin.
9 - Mode/ Calibration
9-15
-------
West Br. Brandywine Creek
33,77
54,76
33,61
Little Mill Creek
Burroughs Run
43,49
W.Br. Red Clay
54,46
54,44
Mill Creek
Pike Creek
Muddy Run
19,32
O
o
CD
43,23
CD
White Clay
Creek
54,20
19,20
CO
" Mid.Br. White Clay
54,18
43,17
(8,9)
43,1 5
54,15
(6,7)
(4,5)
Christina River
West Br.
19,20 EFDC I,J cell number
(6J) Inlet control structure
I27] Flow control structure
—=-pa-i Consumptive Use Water Withdrawa
Figure 9—2. Locations of consumptive use water withdrawals.
9-16
9 - Mode/ Calibration
-------
CHRISTINA RIVER BASIN MODEL - MAY-SEP 1997 CALIBRATION
-------
CHRISTINA RIVER BASIN MODEL
MAY-SEP 1997 CALIBRATION
0°
120
150
180 210
MAY 1 TO SEPTEMBER 21, 1997
240
270
BRANDYWINE CREEK 01481500 (Cell 54,23) (QYY)
o OBSERVED
MODEL
CHRISTINA RIVER BASIN MODEL
MAY-SEP 1997 CALIBRATION
0°
120
150
180 210
MAY 1 TO SEPTEMBER 21, 1997
240
270
BRANDYWINE CREEK 01481000 (Cell 54,37) (QYY)
o OBSERVED
MODEL
CHRISTINA RIVER BASIN MODEL
MAY-SEP 1997 CALIBRATION
0°
120
150
180
MAY 1 TO SEPTEMBER 21, 1997
210
240
270
E.BR. BRANDYWINE 01480870 (Cell 54,56) (QYY)
o OBSERVED
MODEL
9-18
Figure 9-4. Model-data hydrographs, Brandywine Creek and E. Br. Brandywine Creek.
9 - Model Calibration
-------
CHRISTINA RIVER BASIN MODEL
MAY-SEP 1997 CALIBRATION
0°
120
150
180 210
MAY 1 TO SEPTEMBER 21, 1997
240
270
E.BR. BRANDYWINE 01480700 (Cell 54,64) (QYY)
o OBSERVED
MODEL
CHRISTINA RIVER BASIN MODEL
MAY-SEP 1997 CALIBRATION
0°
120
150
180 210
MAY 1 TO SEPTEMBER 21, 1997
240
270
W.BR. BRANDYWINE 01480617 (Cell 26,79) (QXX)
o OBSERVED
MODEL
CHRISTINA RIVER BASIN MODEL
MAY-SEP 1997 CALIBRATION
0°
120
150
180
MAY 1 TO SEPTEMBER 21, 1997
210
240
270
W.BR. BRANDYWINE 01480500 (Cell 20,79) (QXX)
o OBSERVED
MODEL
Figure 9-5. Model-data hydrographs, E. Branch and W. Branch Brandywine Creek.
9 - Mode/ Calibration 9-19
-------
CHRISTINA RIVER BASIN MODEL
MAY-SEP 1997 CALIBRATION
0°
120
150
180 210
MAY 1 TO SEPTEMBER 21, 1997
240
270
CHRISTINA RIVER 01478000 (Cell 22,13) (QXX)
o OBSERVED
MODEL
CHRISTINA RIVER BASIN MODEL
MAY-SEP 1997 CALIBRATION
0°
120
150
180 210
MAY 1 TO SEPTEMBER 21, 1997
240
270
WHITE CLAY CREEK 01479000 (Cell 38,18) (QXX)
o OBSERVED
MODEL
CHRISTINA RIVER BASIN MODEL
MAY-SEP 1997 CALIBRATION
0°
120
150
180
MAY 1 TO SEPTEMBER 21, 1997
210
240
270
WHITE CLAY CREEK 01478650 (Cell 28,18) (QXX)
o OBSERVED
MODEL
9-20
Figure 9-6. Model-data hydrographs, Christina River and White Clay Creek.
9 - Model Calibration
-------
CHRISTINA RIVER BASIN MODEL
MAY-SEP 1997 CALIBRATION
0°
120
150
180 210
MAY 1 TO SEPTEMBER 21, 1997
240
270
RED CLAY CREEK 01478015 (Cell 43,24) (QYY)
o OBSERVED
MODEL
CHRISTINA RIVER BASIN MODEL
MAY-SEP 1997 CALIBRATION
0°
120
150
180 210
MAY 1 TO SEPTEMBER 21, 1997
240
270
RED CLAY CREEK 01480000 (Cell 43,31) (QYY)
o OBSERVED
MODEL
CHRISTINA RIVER BASIN MODEL
MAY-SEP 1997 CALIBRATION
0°
120
150
180
MAY 1 TO SEPTEMBER 21, 1997
210
240
270
RED CLAY CREEK 01479820 (Cell 43,40) (QYY)
o OBSERVED
MODEL
9
Figure 9-7. Model-data hydrographs, Red Clay Creek.
- Model Calibration
9-2 7
-------
1480617
'N01Q:
rcwr:
ICEBQ?
PA
Figure 9-8. Monitoring stations used for model-data time-series comparisons.
9-22 9 - Mode/ Calibration
-------
CHRISTINA RIVER BASIN MODEL: MIN/MAX DISSOLVED OXYGEN AT USGS STATIONS
180 210
JULIAN DAY (MAY 1 TO SEP 21, 1997)
CHADDS FORD GAGE (Cell 54,38)
-DAILY AVERAGE
•DAILY MAXIMUM
•• DAILY MINIMUM
CHRISTINA RIVER BASIN MODEL: MIN/MAX DISSOLVED OXYGEN AT USGS STATIONS
V
180 210
JULIAN DAY (MAY 1 TO SEP 21, 1997)
DOWNINGTOWN GAGE (Cell 54,55)
-DAILY AVERAGE
•DAILY MAXIMUM
•• DAILY MINIMUM
CHRISTINA RIVER BASIN MODEL: MIN/MAX DISSOLVED OXYGEN AT USGS STATIONS
itj
$
-DAILY AVERAGE
•DAILY MAXIMUM
180 210
JULIAN DAY (MAY 1 TO SEP 21, 1997)
MODENA GAGE (Cell 27,79)
DAILY MINIMUM
Figure 9-9. Diel dissolved oxygen at USGS monitoring stations.
9 - Mode/ Calibration
9-23
-------
CHRISTINA RIVER BASIN MODEL: MIN/MAX DISSOLVED OXYGEN AT USGS STATIONS
-
-
1®
& ,
W"\ '¦ ¦" %
jpgft -
s
-
-
180 210
JULIAN DAY (MAY 1 TO SEP 21, 1997)
CHADDS FORD GAGE (Cell 54,38)
| OBSERVED DAILY AVERAGE
•DAILY MAXIMUM
•• DAILY MINIMUM
CHRISTINA RIVER BASIN MODEL: MIN/MAX DISSOLVED OXYGEN AT USGS STATIONS
I
180 210
JULIAN DAY (MAY 1 TO SEP 21, 1997)
DOWNINGTOWN GAGE (Cell 54,55)
| OBSERVED DAILY AVERAGE
•DAILY MAXIMUM
•• DAILY MINIMUM
CHRISTINA RIVER BASIN MODEL: MIN/MAX DISSOLVED OXYGEN AT USGS STATIONS
180 210
JULIAN DAY (MAY 1 TO SEP 21, 1997)
MODENA GAGE (Cell 27,79)
| OBSERVED DAILY AVERAGE
•DAILY MAXIMUM
•• DAILY MINIMUM
Figure 9-10. Water temperature at USGS monitoring stations.
9-24
9 - Mode/ Calibration
-------
CHRISTINA RIVER BASIN MODEL: MIN/MAX DISSOLVED OXYGEN AT USGS STATIONS
-DAILY AVERAGE
•DAILY MAXIMUM
180 210
JULIAN DAY (MAY 1 TO SEP 21, 1997)
CHADDS FORD GAGE (Cell 54,38)
DAILY MINIMUM
CHRISTINA RIVER BASIN MODEL: MIN/MAX DISSOLVED OXYGEN AT USGS STATIONS
180 210
JULIAN DAY (MAY 1 TO SEP 21, 1997)
DOWNINGTOWN GAGE (Cell 54,55)
-DAILY AVERAGE DAILY MAXIMUM DAILY MINIMUM
CHRISTINA RIVER BASIN MODEL: MIN/MAX DISSOLVED OXYGEN AT USGS STATIONS
180 210
JULIAN DAY (MAY 1 TO SEP 21, 1997)
MODENA GAGE (Cell 27,79)
-DAILY AVERAGE DAILY MAXIMUM DAILY MINIMUM
Figure 9-11. Periphyton biomass at USGS monitoring stations.
9 - Mode/ Calibration
9-25
-------
CHRISTINA RIVER BASIN MODEL: MAY-SEP 1997 CALIBRATION
150 180 210 240
MAY 01 TO SEP 21, 1997
MODENA GAGE (Cell 27,79) (NLM)
(NO OBSERVED) MODEL
CHRISTINA RIVER BASIN MODEL: MAY-SEP 1997 CALIBRATION
180 210 240
MAY 01 TO SEP 21, 1997
MODENA GAGE (Cell 27,79) (PLM)
(NO OBSERVED) MODEL
CHRISTINA RIVER BASIN MODEL: MAY-SEP 1997 CALIBRATION
150 180 210 240
MAY 01 TO SEP 21, 1997
MODENA GAGE (Cell 27,79) (LLM)
(NO OBSERVED) MODEL
CHRISTINA RIVER BASIN MODEL: MAY-SEP 1997 CALIBRATION
180 210
MAY 01 TO SEP 21, 1997
MODENA GAGE (Cell 27,79) (TLM)
(NO OBSERVED) MODEL
CHRISTINA RIVER BASIN MODEL: MAY-SEP 1997 CALIBRATION
CHRISTINA RIVER BASIN MODEL: MAY-SEP 1997 CALIBRATION
150 180 210 240
MAY 01 TO SEP 21, 1997
MODENA GAGE (Cell 27,79) (VLM)
(NO OBSERVED) MODEL
150 180 210 240
MAY 01 TO SEP 21, 1997
MODENA GAGE (Cell 27,79) (DLM)
(NO OBSERVED) MODEL
9-26
Figure 9-12. Periphyton limitation factors (Modena gage, W. Br. Brandywine Cr.).
9 - Mode/ Calibration
-------
CHRISTINA RIVER BASIN MODEL: MAY-SEP 1997 CALIBRATION
CHRISTINA RIVER BASIN MODEL: MAY-SEP 1997 CALIBRATION
150 180 210 240
MAY 01 TO SEP 21, 1997
DOWNINGTOWN GAGE (Cell 54,55) (NLM)
(NO OBSERVED) MODEL
180 210 240
MAY 01 TO SEP 21, 1997
DOWNINGTOWN GAGE (Cell 54,55) (PLM)
(NO OBSERVED) MODEL
CHRISTINA RIVER BASIN MODEL: MAY-SEP 1997 CALIBRATION
150 180 210 240
MAY 01 TO SEP 21, 1997
DOWNINGTOWN GAGE (Cell 54,55) (LLM)
(NO OBSERVED) MODEL
CHRISTINA RIVER BASIN MODEL: MAY-SEP 1997 CALIBRATION
180 210 2
MAY 01 TO SEP 21, 1997
DOWNINGTOWN GAGE (Cell 54,55) (TLM)
(NO OBSERVED) MODEL
CHRISTINA RIVER BASIN MODEL: MAY-SEP 1997 CALIBRATION
CHRISTINA RIVER BASIN MODEL: MAY-SEP 1997 CALIBRATION
IT
150 180 210 240
MAY 01 TO SEP 21, 1997
DOWNINGTOWN GAGE (Cell 54,55) (VLM)
(NO OBSERVED) MODEL
150 180 210 240
MAY 01 TO SEP 21, 1997
DOWNINGTOWN GAGE (Cell 54,55) (DLM)
(NO OBSERVED) MODEL
Figure 9-13. Periphyton limitation factors (Downingtown gage, E. Br. Brandywine Cr.).
9 - Mode/ Calibration 9-27
-------
CHRISTINA RIVER BASIN MODEL: MAY-SEP 1997 CALIBRATION
150 180 210 240
MAY 01 TO SEP 21, 1997
CHADDS FORD GAGE (Cell 54,38) (NLM)
(NO OBSERVED) MODEL
CHRISTINA RIVER BASIN MODEL: MAY-SEP 1997 CALIBRATION
180 210 240
MAY 01 TO SEP 21, 1997
CHADDS FORD GAGE (Cell 54,38) (PLM)
(NO OBSERVED) MODEL
CHRISTINA RIVER BASIN MODEL: MAY-SEP 1997 CALIBRATION
CHRISTINA RIVER BASIN MODEL: MAY-SEP 1997 CALIBRATION
150 180 210 240
MAY 01 TO SEP 21, 1997
CHADDS FORD GAGE (Cell 54,38) (LLM)
(NO OBSERVED) MODEL
180 210 2
MAY 01 TO SEP 21, 1997
CHADDS FORD GAGE (Cell 54,38) (TLM)
(NO OBSERVED) MODEL
CHRISTINA RIVER BASIN MODEL: MAY-SEP 1997 CALIBRATION
CHRISTINA RIVER BASIN MODEL: MAY-SEP 1997 CALIBRATION
150 180 210 240
MAY 01 TO SEP 21, 1997
CHADDS FORD GAGE (Cell 54,38) (VLM)
150 180 210 240
MAY 01 TO SEP 21, 1997
CHADDS FORD GAGE (Cell 54,38) (DLM)
(NO OBSERVED) MODEL
(NO OBSERVED) MODEL
9-28
Figure 9-14. Periphyton limitation factors (Chadds Ford gage, Brandywine Cr.).
9 - Mode/ Calibration
-------
CHRISTINA RIVER BASIN MODEL: MAY-SEP 1997 CALIBRATION
CHRISTINA RIVER BASIN MODEL: MAY-SEP 1997 CALIBRATION
150 180 210 240
MAY 01 TO SEP 21, 1997
SMALLEYS POND (Cell 32,13) (NLM)
(NO OBSERVED) MODEL
180 210 240
MAY 01 TO SEP 21, 1997
SMALLEYS POND (Cell 32,13) (PLM)
(NO OBSERVED) MODEL
CHRISTINA RIVER BASIN MODEL: MAY-SEP 1997 CALIBRATION
150 180 210 240
MAY 01 TO SEP 21, 1997
SMALLEYS POND (Cell 32,13) (LLM)
(NO OBSERVED) MODEL
CHRISTINA RIVER BASIN MODEL: MAY-SEP 1997 CALIBRATION
180 210
MAY 01 TO SEP 21, 1997
SMALLEYS POND (Cell 32,13) (TLM)
(NO OBSERVED) MODEL
CHRISTINA RIVER BASIN MODEL: MAY-SEP 1997 CALIBRATION
150 180 210 240
MAY 01 TO SEP 21, 1997
SMALLEYS POND (Cell 32,13) (VLM)
(NO OBSERVED) MODEL
CHRISTINA RIVER BASIN MODEL: MAY-SEP 1997 CALIBRATION
150 180 210 240
MAY 01 TO SEP 21, 1997
SMALLEYS POND (Cell 32,13) (DLM)
(NO OBSERVED) MODEL
Figure 9-15. Periphyton limitation factors (Smalleys Pond, Christina River).
9 - Mode/ Calibration
9-29
-------
CHRISTINA RIVER BASIN MODEL: MAY-SEP 1997 CALIBRATION
150 180 210 240
MAY 01 TO SEP 21, 1997
RED CLAY CREEK WBR (Cell 38,43) (NLM)
(NO OBSERVED) MODEL
CHRISTINA RIVER BASIN MODEL: MAY-SEP 1997 CALIBRATION
180 210 24
MAY 01 TO SEP 21, 1997
RED CLAY CREEK WBR (Cell 38,43) (PLM)
(NO OBSERVED) MODEL
CHRISTINA RIVER BASIN MODEL: MAY-SEP 1997 CALIBRATION
(NO OBSERVED)
180 210 24
MAY 01 TO SEP 21, 1997
RED CLAY CREEK WBR (Cell 38,43) (LLM)
CHRISTINA RIVER BASIN MODEL: MAY-SEP 1997 CALIBRATION
(NO OBSERVED)
180 210 24
MAY 01 TO SEP 21, 1997
RED CLAY CREEK WBR (Cell 38,43) (TLM)
CHRISTINA RIVER BASIN MODEL: MAY-SEP 1997 CALIBRATION
CHRISTINA RIVER BASIN MODEL: MAY-SEP 1997 CALIBRATION
150 180 210 240
MAY 01 TO SEP 21, 1997
RED CLAY CREEK WBR (Cell 38,43) (VLM)
(NO OBSERVED) MODEL
150 180 210 240
MAY 01 TO SEP 21, 1997
RED CLAY CREEK WBR (Cell 38,43) (DLM)
(NO OBSERVED) MODEL
9-30
Figure 9-16. Periphyton limitation factors (W. Br. Red Clay Creek).
9 - Mode/ Calibration
-------
BRANDYWINE CREEK EAST BRANCH: DIEL DISSOLVED OXYGEN AT MONITOR #1
225 226
JULIAN DAY (AUGUST 12-14, 1997)
EBR BRANDYWINE CR MONITOR 1 (Cell 54,62)
o OBSERVED
-HOUR AVERAGE
•HOUR MAXIMUM
• HOUR MINIMUM
BRANDYWINE CREEK EAST BRANCH: DIEL DISSOLVED OXYGEN AT MONITOR #2
225 226
JULIAN DAY (AUGUST 12-14, 1997)
EBR BRANDYWINE CR MONITOR 2 (Cell 54,56)
o OBSERVED
-HOUR AVERAGE
•HOUR MAXIMUM
• HOUR MINIMUM
BRANDYWINE CREEK EAST BRANCH: DIEL DISSOLVED OXYGEN AT MONITOR #3
225 226
JULIAN DAY (AUGUST 12-14, 1997)
EBR BRANDYWINE CR MONITOR 3 (Cell 54,55)
o OBSERVED
-HOUR AVERAGE
•HOUR MAXIMUM
• HOUR MINIMUM
Figure 9-17. Model-data diel D.O. comparison, Brandywine Creek East Branch.
9 - Mode/ Calibration
9-31
-------
BRANDYWINE CREEK WEST BRANCH: DIEL DISSOLVED OXYGEN AT MONITOR #1
16 | 1 1 1
2 -
0 1
231
232 233
JULIAN DAY (AUGUST 19-21, 1997)
WBR BRANDYWINE CR MONITOR 1 (Cell 18,79)
o OBSERVED
-HOUR AVERAGE
•HOUR MAXIMUM
• HOUR MINIMUM
BRANDYWINE CREEK WEST BRANCH: DIEL DISSOLVED OXYGEN AT MONITOR #3
< 12 -
O)
E
2 -
0 1
231
232 233
JULIAN DAY (AUGUST 19-21, 1997)
WBR BRANDYWINE CR MONITOR 3 (Cell 25,79)
o OBSERVED
-HOUR AVERAGE
•HOUR MAXIMUM
• HOUR MINIMUM
BRANDYWINE CREEK WEST BRANCH: DIEL DISSOLVED OXYGEN AT MONITOR #4
2 -
0 1
231
o OBSERVED
232 233
JULIAN DAY (AUGUST 19-21, 1997)
WBR BRANDYWINE CR MONITOR 4 (Cell 26,89)
-HOUR AVERAGE
•HOUR MAXIMUM
• HOUR MINIMUM
9-32
Figure 9-18. Model-data diel D.O. comparison, Brandywine Creek West Branch.
9 - Model Calibration
-------
RED CLAY CREEK WEST BRANCH: DIEL DISSOLVED OXYGEN AT MONITOR #1
1
1
1
217
218
219
220
JULIAN DAY (AUGUST 05-07, 1997)
RED CLAY CREEK WBR (Cell 29,43)
o OBSERVED HOUR AVERAGE HOUR MAXIMUM HOUR MINIMUM
RED CLAY CREEK WEST BRANCH: DIEL DISSOLVED OXYGEN AT MONITOR #2
218 219
JULIAN DAY (AUGUST 05-07, 1997)
RED CLAY CREEK WBR (Cell 30,43)
o OBSERVED HOUR AVERAGE HOUR MAXIMUM HOUR MINIMUM
RED CLAY CREEK WEST BRANCH: DIEL DISSOLVED OXYGEN AT MONITOR #3
1
1
1
1
217
218
219
220
JULIAN DAY (AUGUST 05-07, 1997)
RED CLAY CREEK WBR (Cell 31,43)
o OBSERVED HOUR AVERAGE HOUR MAXIMUM HOUR MINIMUM
Figure 9-19. Model-data diel D.O. comparison, Red Clay Creek West Branch
9 - Mode/ Calibration
9-33
-------
RED CLAY CREEK WEST BRANCH: DIEL DISSOLVED OXYGEN AT MONITOR #4
225 226
JULIAN DAY (AUGUST 12-14, 1997)
RED CLAY CREEK WBR (Cell 32,43)
o OBSERVED HOUR AVERAGE HOUR MAXIMUM HOUR MINIMUM
RED CLAY CREEK WEST BRANCH: DIEL DISSOLVED OXYGEN AT MONITOR #5
1
1
1
1
224
225
226
227
JULIAN DAY (AUGUST 12-14, 1997)
RED CLAY CREEK WBR (Cell 34,43)
o OBSERVED HOUR AVERAGE HOUR MAXIMUM HOUR MINIMUM
Figure 9-20. Model-data diel D.O. comparison, Red Clay Creek West Branch
9-34 9 - Mode/ Calibration
-------
WHITE CLAY CREEK EAST BRANCH: DIEL DISSOLVED OXYGEN AT MONITOR #1
16 | 1 1 1
2 -
0 1
238
239 240
JULIAN DAY (AUGUST 26-28, 1997)
EBR WHITE CLAY CR MONITOR 1 (Cell 19,32)
o OBSERVED
-HOUR AVERAGE
•HOUR MAXIMUM
• HOUR MINIMUM
WHITE CLAY CREEK EAST BRANCH: DIEL DISSOLVED OXYGEN AT MONITOR #2
2 -
0 1
238
239 240
JULIAN DAY (AUGUST 26-28, 1997)
EBR WHITE CLAY CR MONITOR 2 (Cell 19,31)
o OBSERVED
-HOUR AVERAGE
•HOUR MAXIMUM
• HOUR MINIMUM
WHITE CLAY CREEK EAST BRANCH: DIEL DISSOLVED OXYGEN AT MONITOR #3
2 -
0 1
238
o OBSERVED
239 240
JULIAN DAY (AUGUST 26-28, 1997)
EBR WHITE CLAY CR MONITOR 3 (Cell 19,30)
-HOUR AVERAGE
•HOUR MAXIMUM
• HOUR MINIMUM
Figure 9-21. Model-data diel D.O. comparison, White Clay Creek East Branch
9 - Mode/ Calibration
9-35
-------
WHITE CLAY CREEK EAST BRANCH: DIEL DISSOLVED OXYGEN AT MONITOR #4
16 | 1 1 1
2 -
0 1
238
o OBSERVED
239 240
JULIAN DAY (AUGUST 26-28, 1997)
EBR WHITE CLAY CR MONITOR 4 (Cell 19,29)
-HOUR AVERAGE
•HOUR MAXIMUM
• HOUR MINIMUM
Figure 9-22. Model-data diel D.O. comparison, White Clay Creek East Branch
9-36 9 - Mode/ Calibration
-------
10 - MODEL VALIDATION
Model validation involved the application of the calibrated model using a different time period,
namely, May 1 to September 21, 1995. This period was characterized by extremely low stream flows
from late August to the middle of September. An extensive field monitoring program was conducted by
the states of Delaware and Pennsylvania during the summer of 1995 in which grab samples were
collected at a number of locations throughout the basin at a frequency of at least once a month. These
data were assembled into an electronic database and were used to assess the model validation simulation.
This section presents the results of the validation of the EFDC hydrodynamic and water quality model.
Parameters considered for validation include flow rate, tidal surface elevation, chlorides, and a suite of
water quality parameters including dissolved oxygen, nitrogen species, phosphorus species, organic
carbon, chlorophyll-a, total suspended solids, and fecal coliform bacteria.
10.1 Validation Period
The time period for model validation, May 1 to September 21, 1995, was selected because it
included an ambitious field monitoring program conducted by DNREC, PADEP, USGS, and others.
During the late-August to mid-September 1995 time period, stream flow throughout the basin was below
historical 7Q10 flow rates. Data for comparison to the model water quality results for stream reaches
were monitored generally on a monthly basis during the validation period. The USGS maintained
continuous monitors at three locations (Chadds Ford, Downingtown, and Modena) to record daily
minimum and maximum values of dissolved oxygen, temperature, and pH.
10.2 Hydrodynamic and Hydraulic Validation
Assessment of the validation of the hydrodynamic model was accomplished by comparing model
results to field observations of flow and tidal elevation. Tidal constituent amplitudes and phases were
compared with those measured at two USGS tide gages on the Christina River at the Port of Wilmington
and at Newport. The model was executed for a period of 143 days from May 1 to September 21, 1995.
The simulated model stream flow rates were compared with available observations at 12 USGS stream
gages
10.2.1 Validation of Tide Elevation
Validation of the model with respect to tidal water surface elevation was accomplished by
analysis of observed and model predicted time-series data at interior tide stations. For tidal waters, least
squares harmonic analysis is the most commonly utilized procedure (Oey, Mellor and Hires 1985; Cheng
et al. 1993; Shen et al. 1999). Tide elevation data were obtained from the USGS tide stations on the
Christina River at the Port of Wilmington near the mouth. No tidal data were available for the Newport
10- Model Validation
10-1
-------
station during the validation period. These data were compared with surface elevations computed by the
model at cell 56,13 (Port of Wilmington). The time-series of tide elevations for the month of August
1995 for both the field data and the model results were subjected to a harmonic analysis. The five most
important astronomical harmonic constituents (M2, S2, N2, Kl, and 01) were computed for both the
field data and model simulation results. The harmonic analysis results, shown in Table 10-1, indicate the
model is in reasonable agreement with the measured tide data for both amplitude and phase. The model-
data amplitudes for the M2 harmonic constituent agree within 6 cm (8%) and the phases agree to within
7 degrees. Time-series graphs (Figure 10-1) of the observed and model tide elevations at the Port of
Wilmington covering a 31-day period (August 1-31, 1995) provide a visual means of assessing the skill
of the model in simulating tidal elevations. The model tides are forced at the north and south boundaries
in the Delaware River based on the NOAA predictions at the Reedy Point and Chester subordinate
stations (NOAA 1998). These predictions do not consider the low-frequency phenomenon caused by
regional low pressure systems or storms that will be found in the signal of the tide data collected at the
two USGS tide stations on the Christina River.
Table 10-1. Harmonic analysis of tides at Port of Wilmington and Newport.
Harmonic Constant
Port of Wilmington (USGS #01481062)
Newport (USGS #01480065)
Amplitude (m)
Phase (degrees)
Amplitude (m)
Phase (degrees)
M2 - observed
0.7755
303.381
_
_
M2 - model
0.7148
310.070
0.7287
316.763
Difference
0.0607
-6.698
-
-
S2 - observed
0.0962
26.056
_
_
S2 - model
0.1017
31.289
0.1031
41.384
Difference
-0.0055
-5.233
-
-
N2 - observed
0.1251
294.181
_
_
N2 - model
0.1348
326.473
0.1347
335.573
Difference
-0.0097
-32.292
-
-
Kl - observed
0.1119
186.070
_
_
Kl - model
0.0646
172.191
0.0646
175.690
Difference
0.0473
13.879
-
-
Ol - observed
0.0696
147.329
_
_
Ol - model
0.0595
151.881
0.0598
154.997
Difference
0.0101
-4.552
-
-
10.2.2 Hydraulic Flow Balance
Validation of the hydraulic flow balance in the model system was determined by comparing the
model and observed hydrograph at 12 USGS stream gage locations in the Christina River Basin for the
validation period. Estimates of unit discharge rates (cfs per square mile) were estimated for each of the
39 subwatersheds for each day in the validation period. The daily flow rates for each subwatershed were
10-2
10- Model Validation
-------
then distributed uniformly among each of the grid cells within a subwatershed, except for headwater grid
cells which were assigned a flow rate in accordance with the contributing area to that cell. The model
was then run for the 143-day simulation period, and the flow rate at the appropriate gage location was
compared with the model results. The model flow rates compared reasonably well with the daily average
flows at the stream gages (see Figures 10-2 to 10-5). The purpose of validating simulated hydraulic flow
balance to the USGS gage data was to determine whether the runoff rates from the contributing
subwatersheds were properly apportioned to the model grid cells. Normally, a watershed runoff model
would be used to provide flows to the receiving water model. However, the calibrated HSPF watershed
model will not be available for a few more years. The model hydrograph agrees well with the stream
gage data during periods between storm runoff events. At certain locations, the model tends to under
predict the peak flow rates of the storm events. The use of a watershed runoff model in the future will
likely improve the peak flow validation because the timing of the peak runoff from each subwatershed
will be taken into account (a procedure that was not possible in the present model application).
10.3 Water Quality Validation Results
Each observation in STORET was collected at an instant in time and at a single point in space.
Time scales realistically represented in the EFDC model were determined by time scales of primary
forcing functions: 60-second tidal hydrodynamics, hourly meteorological updates, monthly ocean
boundary conditions, constant nonpoint source concentration estimates, daily nonpoint source flows,
monthly point source loads, and hourly atmospheric wet deposition. The minimum model spatial scales
were determined by the size of the grid cells, ranging from 500 to about 1000 meters in the longitudinal
direction along the streams. Data for longitudinal transect comparisons were averaged over the period
August 1 to September 30, 1995. The disparity in the temporal and spatial scales between the model and
prototype, especially for the nonpoint and point source loads, meant that individual observations may not
be directly comparable with model prediction at a specific time in a given model grid cell.
Model-data comparisons will be made by means of longitudinal transect plots as well as time-
series plots for the 11 major stream reaches in the study area: Brandywine Creek, East Branch
Brandywine Creek, West Branch Brandywine Creek, Buck Run, Christina River (tidal), Christina River
(nontidal), Red Clay Creek, West Branch Red Clay Creek, White Clay Creek, East Branch White Clay
Creek, and the Delaware River. The transects are delineated in river miles referenced to River Mile 74.9
located at the mouth of the Christina River based on EPA REACH FILE 1 (see Table 9-4). Longitudinal
transect plots for each water column parameter are presented in Appendix E arranged by stream reach.
There are 18 transect plots for each reach, representing 18 different water quality parameters. The model
results for the transect plots were averaged over the period August 25 to September 10, 1995. The
horizontal axis of each plot represents the river mile from the mouth of the Christina River measured
10- Model Validation
10-3
-------
along the stream network. The vertical axis represents the water column parameter concentration. The
observed data are shown as "box-and-whisker" symbols indicating the maximum, minimum, 25th
percentile, 75th percentile, mean, and median statistics. The model results are represented by three lines:
the solid line is the mean over the averaging period at a given model grid cell and the two dashed lines
are the minimum and maximum values simulated over the averaging period.
The time-series plots are presented in Appendix F and cover the entire 143-day validation period
beginning on May 1 (day 121) and continuing to September 21, 1995 (day 264). The time-series model-
data comparisons were made at 16 monitoring locations on the various stream reaches in the study area
(Figure 9-8). The concentrations of the nonpoint source loads were considered to be constant throughout
the validation period, and the loads vary in accordance with the changes in nonpoint source flow rate.
The nonpoint source concentrations were based on the summer low-flow monitoring data and are
representative of background water quality conditions. In reality, the concentrations of the various water
quality parameters will vary in relation to storm events and changes in watershed runoff. Determination
of the time-varying nonpoint source concentrations was outside the scope of the present study, but will be
addressed in the future following completion of the HSPF watershed model of the Christina River Basin.
10.3.1 Brandywine Creek Main Stem Water Quality Validation Results
The transect plots for all water quality parameters for the main stem of Brandywine Creek are
shown in Figures E01 to E03, and the time-series plots at stations 104021 and WQN0105 are shown in
Figures F01 to F04. The conservative constituents (chlorides and TSS) match the observed data
reasonably well in both the transect and time-series plots. The time-series of TSS shows little variation
because a constant nonpoint source concentration was used for the entire time period, whereas in reality
the TSS concentrations would increase during storm runoff events. The grab samples of dissolved
oxygen generally lie within the minimum and maximum range computed by the model for both the
transect and time-series views. The observed organic carbon indicates an increasing trend in the
downstream direction that is not reflected by the model. This is likely due to missing sources of organic
carbon. The total phosphorus and dissolved orthophosphate indicate a decreasing concentration in the
downstream direction that is simulated in the model. The nitrogen species (total nitrogen, ammonia
nitrogen, and nitrate nitrogen) all match the observations along the transect reasonably well. Total
nitrogen simulated by the model is slightly higher than the observations and may be due to excess
dissolved organic nitrogen since the other species agree well with the data.
10.3.2 Brandywine Creek East Branch Water Quality Validation Results
The transect plots for all water quality parameters for the East Branch Brandywine Creek are
shown in Figures E04 to E06, and the time-series plots at station 01480870 are shown in Figures F05 and
10-4
10- Model Validation
-------
F06. The Downingtown WWTP (PA0026531) discharges at mile 103.7, which accounts for the abrupt
change concentrations of some of the water quality parameters at that location. Along the transect, all
water quality parameters are in reasonable agreement with observations. The time-series plots also
indicate the model is in agreement with observations. The exception is fecal coliform bacteria, which
does not agree well with observations due to nonpoint sources that are not accounted for in the model.
10.3.3 Brandywine Creek West Branch Water Quality Validation Results
The transect plots for all water quality parameters for the West Branch Brandywine Creek are
shown in Figures E07 to E09, and the time-series plots at station 01480617 are shown in Figures F07 and
F08. The South Coatesville (PA0036987) and Coatesville City (PA0026859) WWTPs discharge at river
mile 110.4 and 111.0, respectively. The spike in the concentrations of various water quality parameters
is due to these two discharges. The model agrees with the observed data very well for chlorides, TSS,
dissolved oxygen, chlorophyll-a, organic carbon, total nitrogen, ammonia nitrogen, and nitrate nitrogen.
Fecal coliform bacteria (see Figure F08) simulated by the model are about an order of magnitude less
than the observations. This is likely due to nonpoint sources that are not accounted for in the model.
10.3.4 Buck Run Water Quality Validation Results
The transect plots for all water quality parameters for Buck Run are shown in Figures E10 to
E12, and the time-series plots are shown in Figures F09 and F10. No observed data were available for
Buck Run during the validation period, so validation cannot be assessed in this stream reach.
10.3.5 Christina River (Tidal) Water Quality Validation Results
The water-quality transect plots for the tidal Christina River are shown in Figures E13 to E15,
and the time-series plots at stations 106291 and 106021 are shown in Figures F23 to F26. The lower
portion of the Christina River is strongly influenced by the Delaware River due to tidal excursion. The
transect plots indicate the model agrees well with the data for all parameters with the exception of total
nitrogen, which is slightly high in the model; ammonia nitrogen, which is low in the model; and total
organic carbon, which is about 2.5 mg/L low in the model. The time-series plots also indicate similar
reasonable model-data agreement with the exception of the above-mentioned three parameters.
10.3.6 Christina River (Nontidal) Water Quality Validation Results
The transect plots for all water quality parameters for the nontidal Christina River are shown in
Figures E16 to El8, and the time-series plots at station 106031 are shown in Figures F27 and F28. Along
the transect, all water quality parameters agree with the observations quite well. The spike in chlorides
concentration at river mile 98.1 is due to the West Branch Christina River, which carries loads from the
two Maryland WWTPs (MD0022641 and MD0065145).
10- Model Validation
10-5
-------
10.3.7 Red Clay Creek and East Branch Water Quality Validation Results
The transect plots for all water quality parameters for Red Clay Creek and East Branch Red Clay
Creek are shown in Figures E19 to E21, and the time-series plots at stations 103031, WQN0150, and
RCEB04 are shown in Figures F09 to F14. The West Branch Red Clay Creek enters at river mile 100.3
and accounts for the spikes in concentration that are evident in a number of the transect plots. Overall,
the model does a reasonable job of simulating the observations for both the transect and time-series
views. At station WQN0150 the total phosphorus and dissolved orthophosphate data indicate an
increasing trend from May to September (see Figure F12). This tends to support the hypothesis that the
primary sources for phosphorus may be from a relatively steady-state source (i.e., point source or
groundwater source) because as the stream flow decreases in the summer months, the concentration of
phosphorus is increasing.
10.3.8 Red Clay Creek West Branch Water Quality Validation Results
The transect plots for all water quality parameters for West Branch Red Clay Creek are shown in
Figures E22 to E24, and the time-series plots at station RCWR2 are shown in Figures F15 and F16. The
Kennett Square WWTP (PA0024058) discharges at river mile 103.9, accounting for the spike in
concentrations at that location. The observed data shown on the transect plots were not measured in the
West Branch Red Clay Creek, but rather were measured in two tributaries (the NVF tributary and the
Toughkenamon tributary). Thus, a direct model-data comparison should not be assumed for any of the
transect plots for the West Branch Red Clay Creek.
10.3.9 White Clay Creek and Middle Branch Water Quality Validation Results
The transect plots for all water quality parameters for White Clay Creek and Middle Branch
White Clay Creek are shown in Figures E25 to E28, and the time-series plots at stations 105151 and
WQN0149 are shown in Figures F17 to F20. The transect plots indicate reasonable model-data
agreement for all parameters except dissolved orthophosphate phosphorus. Downstream of river mile
103 the monitoring data indicate dissolved orthophosphate concentrations in the 0.1 to 0.2 mg/L range,
whereas the model computes concentrations of about one-half of the observed values.
10.3.10 White Clay Creek East Branch Water Quality Validation Results
The transect plots for all water quality parameters for White Clay Creek East Branch are shown
in Figures E28 to E30, and the time-series plots at station WCER2 are shown in Figures F21 and F22.
Station WCER2 is Reach 2 from the August 1997 study (Davis 1998), and no observed data were
available for the 1995 validation period. The Avondale Borough WWTP (PA0025488) is the largest
point source on this stream and discharges at river mile 106.6. No data were available downstream of the
Avondale WWTP to compare to the model results.
10-6
10- Model Validation
-------
10.3.11 Delaware River Water Quality Validation Results
The transect plots for all water quality parameters for the Delaware River are shown in Figures
E31 to E33. The model indicates reasonable agreement for most water quality parameters. The largest
discrepancies occur for chlorophyll-a (model is higher than the observations) and total organic carbon
(model is lower than the observations). The simulated dissolved oxygen sag at river mile 82 reaches a
minimum value of about 2.4 mg/L. The model results cannot be validated at that location since no
observed data were available.
10.3.12 Muddy Run and Pike Creek Water Quality Validation Results
Time-series plots for all water quality parameters for Muddy Run (station 105131) and Pike
Creek (station 105101) are shown in Figures F29 to F32. The model results agree well with the
observations for all parameters with the exception of an apparent algae bloom in mid-July (day 198) at
the Muddy Run station. The data indicate a chlorophyll-a concentration of about 16 ug/L, whereas the
model computes about 3.0 ug/L. However, in May to June and August to September the model agrees
well with the chlorophyll-a measurements.
10.4 Diel Dissolved Oxygen Validation Results
An important feature of the Christina River Basin water quality model is the ability to compute
the daily dissolved oxygen range as well as the daily average value. Water quality standards for
dissolved oxygen in the Christina River Basin must meet two criteria, one for the daily average and one
for the daily minimum concentration. The data available for model validation included diel dissolved
oxygen at several locations. The USGS collected diel dissolved oxygen data at three gages for the entire
May to September 1995 validation period: (1) 01480870 at Downingtown on the East Branch
Brandywine Creek, (2) 01480617 at Modena on the West Branch Brandywine Creek, and (3) 01481000
at Chadds Ford on the main stem Brandywine Creek.
Achieving the proper range in daily dissolved oxygen is primarily a function of the community
periphyton biomass available at a given model grid cell. The periphyton growth and basal metabolism
rates as well as the growth rate density limitation parameters were adjusted to simulate the periphyton
biomass needed to achieve reasonable daily DO ranges at the monitoring sites during mid-August to mid-
September 1995, the critical low-flow period. The monitored daily minimum, maximum, and average
dissolved oxygen concentrations at the three USGS gage locations are shown in Figure 10-6, along with
the model results. It is evident that the model does a reasonable job of simulating the daily DO range at
these three locations during the critical low-flow period (August 15 to September 15, 1995; day 228 to
259). At the Downingtown station, the dissolved oxygen does not agree with the observed data from
10- Model Validation
10-7
-------
May 1 to about July 31, 1995. The reason for this may be due to the Broad Run WWTP (PA0043982),
which is discharging ammonia nitrogen at a concentration of 12 to 15 mg/L during the May to July
period. In August and September the effluent ammonia concentration drops to between 0.8 and 5.3 mg/L
at this WWTP. The daily minimum, maximum, and average water temperatures at the three USGS gage
locations and the simulated model temperatures are shown in Figure 10-7. The model is in good
agreement with the measured temperature data for the entire 5-month simulation period. The time-series
of periphyton biomass at the three gage locations is shown in Figure 10-8. The periphyton biomass has
been displayed in units of ug/L of chlorophyll-a for comparison with the floating chlorophyll-a
concentrations. The periphyton biomass (500 to 1,000 ug/L) is as much as two orders of magnitude
greater than the floating chlorophyll-a biomass (3 to 10 ug/L) at these three locations.
10.5 Sediment Oxygen Demand Rates
No field data were available during the 1995 validation period to verify the flux rates computed
by the predictive sediment submodel. However, SOD rates were measured in July and August 1996, at
three locations in the tidal Christina River and Brandywine Creek. An SOD rate of 0.5 g/m2/day was
used in the tidal Delaware River in another model study conducted by HydroQual for DRBC and was
also adopted for this study. The simulated SOD rates were converted to rates at 20 °C and are compared
with the measured data in Table 10-2. The relative errors vary from 2.2% on the Christina River at
Newport to 22.5% at the mouth of Christina River.
Table 10-2. Model-data comparison of sediment oxygen demand rates (g/m2/day).
Location
Sampling Date
Monitored
SOD at 20 °C
1995 Validation
Model
SOD at 20 °C
Relative
Error
Christina River at 1-495 bridge
Aug 12, 1996
0.81
0.99
22.5%
Christina River at Newport, Rt. 141 bridge
Jul 10, 1996
1.67
1.63
2.2%
Brandywine Creek, 0.6 mi. from mouth
Aug 12, 1996
1.23
1.48
20.0%
Delaware River (from HydroQual study)
-
0.50
0.52
4.8%
10-8
10- Model Validation
-------
CHRISTINA RIVER BASIN MODEL - MAY-SEP 1995 VALIDATION
0
1
i
I
o'
5
218 219 220 221 222 223
JULIAN DAY (AUGUST 01 TO AUGUST 15, 1995)
PORT OF WILMINGTON 01481602 (Cell 56,13) (SEL)
o OBSERVED
MODEL
CHRISTINA RIVER BASIN MODEL - MAY-SEP 1995 VALIDATION
-2
-3
-4
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
JULIAN DAY (AUGUST 16 TO AUGUST 30, 1995)
PORT OF WILMINGTON 01481602 (Cell 56,13) (SEL)
o
-------
CHRISTINA RIVER BASIN MODEL
MAY-SEP 1995 VALIDATION
0°
120
150
180 210
MAY 1 TO SEPTEMBER 21, 1995
240
270
BRANDYWINE CREEK 01481500 (Cell 54,23) (QYY)
o OBSERVED
MODEL
CHRISTINA RIVER BASIN MODEL
MAY-SEP 1995 VALIDATION
0°
120
150
180
MAY 1 TO SEPTEMBER 21, 1995
210
240
270
BRANDYWINE CREEK 01481000 (Cell 54,37) (QYY)
o OBSERVED
MODEL
CHRISTINA RIVER BASIN MODEL
MAY-SEP 1995 VALIDATION
0°
120
150
180
MAY 1 TO SEPTEMBER 21, 1995
210
240
270
E.BR. BRANDYWINE 01480870 (Cell 54,56) (QYY)
o OBSERVED
MODEL
70-70
Figure 10-2. Model-data hydrographs, Brandywine Creek and E. Br. Brandywine Cr.
70 - Model Validation
-------
CHRISTINA RIVER BASIN MODEL
MAY-SEP 1995 VALIDATION
0°
120
150
180 210
MAY 1 TO SEPTEMBER 21, 1995
240
270
E.BR. BRANDYWINE 01480700 (Cell 54,64) (QYY)
o OBSERVED
MODEL
CHRISTINA RIVER BASIN MODEL
MAY-SEP 1995 VALIDATION
0°
120
150
180 210
MAY 1 TO SEPTEMBER 21, 1995
240
270
W.BR. BRANDYWINE 01480617 (Cell 26,79) (QXX)
o OBSERVED
MODEL
CHRISTINA RIVER BASIN MODEL
MAY-SEP 1995 VALIDATION
0°
120
150
180
MAY 1 TO SEPTEMBER 21, 1995
210
240
270
W.BR. BRANDYWINE 01480500 (Cell 20,79) (QXX)
o OBSERVED
MODEL
Figure 10-3. Model-data hydrographs, E. Br. Brandywine Cr. and W. Br. Brandywine Cr.
10 - Mode/ Validation 10-11
-------
CHRISTINA RIVER BASIN MODEL
MAY-SEP 1995 VALIDATION
0°
120
150
180 210
MAY 1 TO SEPTEMBER 21, 1995
240
270
CHRISTINA RIVER 01478000 (Cell 22,13) (QXX)
o OBSERVED
MODEL
CHRISTINA RIVER BASIN MODEL
MAY-SEP 1995 VALIDATION
0°
120
150
180 210
MAY 1 TO SEPTEMBER 21, 1995
240
270
WHITE CLAY CREEK 01479000 (Cell 38,18) (QXX)
o OBSERVED
MODEL
CHRISTINA RIVER BASIN MODEL
MAY-SEP 1995 VALIDATION
0°
120
150
180
MAY 1 TO SEPTEMBER 21, 1995
210
240
270
WHITE CLAY CREEK 01478650 (Cell 28,18) (QXX)
o OBSERVED
MODEL
10-12
Figure 10-4. Model-data hydrographs, Christina River and White Clay Creek.
10 - Model Validation
-------
CHRISTINA RIVER BASIN MODEL
MAY-SEP 1995 VALIDATION
0°
120
150
180 210
MAY 1 TO SEPTEMBER 21, 1995
240
270
RED CLAY CREEK 01478015 (Cell 43,24) (QYY)
o OBSERVED
MODEL
CHRISTINA RIVER BASIN MODEL
MAY-SEP 1995 VALIDATION
0°
120
150
180 210
MAY 1 TO SEPTEMBER 21, 1995
240
270
RED CLAY CREEK 01480000 (Cell 43,31) (QYY)
o OBSERVED
MODEL
CHRISTINA RIVER BASIN MODEL
MAY-SEP 1995 VALIDATION
1
1
1
1
120
150
180
MAY 1 TO SEPTEMBER 21, 1995
210
240
270
RED CLAY CREEK 01479820 (Cell 43,42) (QYY)
o OBSERVED
MODEL
10
Figure 10-5. Model-data hydrographs, Red Clay Creek.
- Model Validation
10-13
-------
CHRISTINA RIVER BASIN MODEL: MIN/MAX DISSOLVED OXYGEN AT USGS STATIONS
16
14
12
10
8
6
4
2
0 1—
120
JULIAN DAY (MAY 1 TO SEP 21, 1995)
CHADDS FORD GAGE (Cell 54,38)
|| OBSERVED DAILY AVERAGE DAILY MAXIMUM DAILY MINIMUM
CHRISTINA RIVER BASIN MODEL: MIN/MAX DISSOLVED OXYGEN AT USGS STATIONS
-DAILY AVERAGE
•DAILY MAXIMUM
180 210
JULIAN DAY (MAY 1 TO SEP 21, 1995)
DOWNINGTOWN GAGE (Cell 54,55)
DAILY MINIMUM
CHRISTINA RIVER BASIN MODEL: MIN/MAX DISSOLVED OXYGEN AT USGS STATIONS
16
14
12
10
8
6
4
2
0 1—
120
JULIAN DAY (MAY 1 TO SEP 21, 1995)
MODENA GAGE (Cell 26,79)
|| OBSERVED DAILY AVERAGE DAILY MAXIMUM DAILY MINIMUM
Figure 10-6. Diel dissolved oxygen at USGS monitoring stations.
74
70 - Mode/ Validation
-------
CHRISTINA RIVER BASIN MODEL: MIN/MAX DISSOLVED OXYGEN AT USGS STATIONS
-DAILY AVERAGE
•DAILY MAXIMUM
180 210
JULIAN DAY (MAY 1 TO SEP 21, 1995)
CHADDS FORD GAGE (Cell 54,38)
DAILY MINIMUM
CHRISTINA RIVER BASIN MODEL: MIN/MAX DISSOLVED OXYGEN AT USGS STATIONS
-DAILY AVERAGE
•DAILY MAXIMUM
180 210
JULIAN DAY (MAY 1 TO SEP 21, 1995)
DOWNINGTOWN GAGE (Cell 54,55)
DAILY MINIMUM
CHRISTINA RIVER BASIN MODEL: MIN/MAX DISSOLVED OXYGEN AT USGS STATIONS
-DAILY AVERAGE
•DAILY MAXIMUM
180 210
JULIAN DAY (MAY 1 TO SEP 21, 1995)
MODENA GAGE (Cell 26,79)
DAILY MINIMUM
Figure 10-7. Water temperature at USGS monitoring stations.
10 - Mode/ Validation
10-15
-------
CHRISTINA RIVER BASIN MODEL: MIN/MAX DISSOLVED OXYGEN AT USGS STATIONS
180 210
JULIAN DAY (MAY 1 TO SEP 21, 1995)
CHADDS FORD GAGE (Cell 54,38)
-DAILY AVERAGE
•DAILY MAXIMUM
•• DAILY MINIMUM
CHRISTINA RIVER BASIN MODEL: MIN/MAX DISSOLVED OXYGEN AT USGS STATIONS
180 210
JULIAN DAY (MAY 1 TO SEP 21, 1995)
DOWNINGTOWN GAGE (Cell 54,55)
-DAILY AVERAGE
•DAILY MAXIMUM
•• DAILY MINIMUM
CHRISTINA RIVER BASIN MODEL: MIN/MAX DISSOLVED OXYGEN AT USGS STATIONS
180 210
JULIAN DAY (MAY 1 TO SEP 21, 1995)
MODENA GAGE (Cell 26,79)
-DAILY AVERAGE
•DAILY MAXIMUM
•• DAILY MINIMUM
Figure 10-8. Periphyton biomass at USGS monitoring stations.
10-16
10 - Mode/ Validation
-------
11 - STATISTICAL SUMMARY OF CALIBRATION AND VALIDATION
The model-data comparisons in Appendices A and B (1997 calibration) and in Appendices E and
F (1995 validation) provide a qualitative evaluation of model performance. A seasoned modeler can
examine the plots and form an experience-based judgment on the status of model calibration and
verification. In this section, model-data comparisons are presented as quantitative statistical summaries.
This presentation provides a different perspective on model-data comparison that numerically quantifies
the state of model calibration/verification (sometimes referred to as model "skill assessment").
Although numerous methods exist for analyzing and summarizing model performance, there is
no consensus in the modeling community on a standard analytical suite. A set of basic statistical
methods were used to compare model predictions and sampling observations which included the mean
error statistic, the absolute mean error, the root-mean-square error, and the relative error. The
observations and model predictions were analyzed over the period May 1 to September 21 for both the
1997 calibration data set and the 1995 validation data set at 20 monitoring locations throughout the
Christina River Basin.
11.1 Mean Error Statistic
The mean error between model predictions and observations is defined in Eq. 11-1. A mean
error of zero is ideal. A non-zero value is an indication that the model may be biased toward either over-
or underprediction. A positive mean error indicates that on average the model predictions are less than
the observations. A negative mean error indicates that on average the model predictions are greater than
the observed data. The mean error statistic may give a false ideal value of zero (or near zero) if the
average of the positive deviations between predictions and observations is about equal to the average of
the negative deviations in a data set. Because of that possibility, it is never a good idea to rely solely on
this statistic as a measure of performance. Instead, it should be used in tandem with the other statistical
measures that are described in this section.
77 2 (O - P)
E = —^ >- (11-1)
n
where:
E = mean error
O = observation, aggregated by month and over the water column
P = model prediction, aggregated by month and over vertical layers
n = number of observed-predicted pairs
11 - Statistical Summary of Calibration and Validation
11-1
-------
11.2 Absolute Mean Error Statistic
The absolute mean error between model predictions and observations is defined in Eq. 11-2. An
absolute mean error of zero is ideal. The magnitude of the absolute mean error indicates the average
deviation between model predictions and observed data. Unlike the mean error, the absolute mean error
cannot give a false zero.
IT 2 \0 - P\
abs = ~L 1 (11-2)
where:
E,,,,v = absolute mean error.
11.3 Root-Mean-Square Error Statistic
The root-mean-square error (E..„,v) is defined in Eq. 11-3. A root-mean-square error of zero is
ideal. The root-mean-square error is an indicator of the deviation between model predictions and
observations. The Ems statistic is an alternative to (and is usually larger than) the absolute mean error.
S {() ^2. (11-3)
E
rms
n
where:
E^, = root-mean-square error
11.4 Relative Error Statistic
The relative error between model predictions and observations is defined in Eq. 11-4. A relative
error of zero is ideal. The relative error is the ratio of the absolute mean error to the mean of the
observations and is expressed as a percent.
2 \0 ~ P\
rel = s o (11"4)
where:
EreI = relative error.
11-2
11 - Statistical Summary of Calibration and Validation
-------
11.5 Evaluation of Results
Summary statistics have been developed for each of the individual monitoring locations as well
as for the entire Christina River Basin (i.e., all monitoring stations taken together as a whole) for both the
1997 calibration period and the 1995 validation period. A summary of the error statistics for all water
quality parameters for the 1997 model calibration simulation is given in Table 11-1, and the 1995
validation summary is given in Table 11-2. The relative error statistic permits comparisons between the
various water quality substances. Temperature and dissolved oxygen were the parameters with the
smallest relative error. The results for temperature indicate a relative error of 5.0% or less, and the
relative error for dissolved oxygen was less than 7.1% for both the calibration and validation runs. The
relative error for total nitrogen was less than 18%, total phosphorus was less than 35%, and total organic
carbon was less than 36% for both the calibration and validation runs. The relative error for chlorophyll-
a was about 19% in the calibration run and 37% in the validation run. The variability of the chlorophyll-
a parameter reflects the nonconservative behavior of algal dynamics and the approximate nature of
mathematical models of biological processes. A rule of thumb for chlorophyll-a monitoring is that at any
given station and any given time, the sampled concentrations can vary by a factor of one-half to double.
The highly dynamic, short-term variations of the chlorophyll-a parameter are extremely difficult to
model. Eutrophication models are better suited to simulating the long-term (daily to monthly time scale)
chlorophyll-a levels rather than the short-term (hourly) concentrations.
The relative errors for total nitrogen and total phosphorus for the combined 1995 and 1997
simulation periods for the individual stream reaches are presented in Table 11-3. The relative errors in
total nitrogen for the primary stream reaches are as follows: West Branch Brandywine Creek (5.2%),
East Branch Brandywine Creek (12.6%), Brandywine Creek main stem (14.4%), West Branch Red Clay
Creek downstream of Kennett Square (9.4%), and East Branch White Clay Creek (18.7%). The relative
errors in total phosphorus for these same stream reaches are West Branch Brandywine Creek (27.8%),
East Branch Brandywine Creek (23.6%), Brandywine Creek main stem (32.8%), West Branch Red Clay
Creek downstream of Kennett Square (16.5%), and White Clay Creek (35.4%).
11 - Statistical Summary of Calibration and Validation
11-3
-------
Table 11-1. Statistical summary of Christina River model 1997 calibration results.
Parameter
Mean Error
Absolute Mean Error
RMS Error
Relative Error
No. Samples
Chlorides (mg/L)
0.2684
2.8507
4.6268
11.40%
55
Chlorophyll-a (ug/L)
0.3741
0.8374
1.3293
18.67%
34
Dissolved Oxygen (mg/L)
0.2598
0.6224
0.9063
7.10%
68
Total Organic Carbon (mg/L)
-0.6543
1.7280
2.4286
35.92%
37
Diss. Organic Carbon (mg/L)
-0.1237
1.4143
2.1255
34.00%
37
Total Nitrogen (mg/L)
0.0617
0.4686
0.6689
13.61%
51
Ammonia Nitrogen (mg/L)
0.0283
0.0322
0.0820
44.58%
55
Nitrate Nitrogen (mg/L)
-0.0106
0.3933
0.5334
14.69%
58
Total Phosphorus (mg/L)
-0.0025
0.0618
0.1143
34.17%
53
Diss. Orthophosphate P (mg/L)
-0.0147
0.0324
0.0536
31.37%
52
Temperature (degC)
-0.6974
0.9147
1.2818
4.72%
70
Table 11-2. Statistical summary of Christina River model 1995 validation results.
Parameter
Mean Error
Absolute Mean Error
RMS Error
Relative Error
No. Samples
Chlorides (mg/L)
2.6778
4.5443
6.0969
18.43%
63
Chlorophyll-a (ug/L)
-0.2194
1.1667
2.2680
53.42%
38
Dissolved Oxygen (mg/L)
0.1747
0.5204
0.7058
6.30%
89
Total Organic Carbon (mg/L)
-0.0552
2.0567
2.9104
33.27%
63
Diss. Organic Carbon (mg/L)
0.2037
1.8365
2.4630
36.43%
63
Total Nitrogen (mg/L)
-0.4560
0.5863
1.0084
21.57%
63
Ammonia Nitrogen (mg/L)
0.0301
0.0347
0.0617
59.06%
63
Nitrate Nitrogen (mg/L)
-0.2554
0.4711
0.7607
22.35%
63
Total Phosphorus (mg/L)
-0.0077
0.0407
0.0593
27.99%
63
Diss. Orthophosphate P (mg/L)
0.0059
0.0246
0.0333
30.37%
15
Temperature (degC)
-0.8597
1.0222
1.3666
5.04%
90
Table 11-3. Relative error of total nitrogen and total phosphorus for 1995 and 1997 simulation periods.
Location
Total Nitrogen
Total Phosphorus
Relative Error
No. Samples
Relative Error
No. Samples
Christina River (Smalleys Pond)
22.6%
8
35.6%
8
Brandywine Creek main stem
14.4%
21
32.8%
23
Brandywine Creek East Branch
12.6%
12
23.6%
14
Brandywine Creek West Branch
5.2%
11
27.8%
13
Red Clay Creek
23.0%
12
31.9%
14
Red Clay Creek West Branch below Kennett Square
1.5%
3
16.5%
2
White Clay Creek
18.7%
22
35.4%
21
White Clay Creek East Branch
9.9%
3
48.1%
3
Muddy Run
28.5%
8
20.4%
7
Pike Creek
25.8%
8
28.6%
6
11-4
11 - Statistical Summary of Calibration and Validation
-------
11.6 Comparison with Other Model Studies
The combined 1997 calibration and 1995 validation results of the Christina River EFDC model
were compared with results from a number of other water quality model studies. Some of these results
were presented in the Chesapeake Bay model report (Cerco and Cole 1994) and were incorporated into
this study. The model studies that will be considered in the comparison include Peconic Estuary (Tetra
Tech 1999), Long Island Sound (HydroQual 1991), Massachusetts Bay (HydroQual 1995), Chesapeake
Bay (Cerco and Cole 1994), Delaware Inland Bays (Cerco et al. 1993), Tolo Harbour (Chau and Jin
1998), New York Bight (Hall and Dortch 1993), Chesapeake Steady-State model (HydroQual 1987),
Potomac Estuary model (Thomann and Fitzpatrick 1982), Gunston Cove (Cerco 1985), Eau Galle
Reservoir (Wlosinski and Collins 1985), and Lake Ontario (Thomann et al. 1979).
Comparing statistics from different model studies is not straightforward. In contrast to classical
statistics, no standard methodology has existed for determining model performance statistics. Various
treatments of predictions and observations among different model studies (for example, aggregation of
data temporally and spatially) affect the statistical results and complicate comparisons between studies.
Since these conflicts cannot be avoided without reworking past modeling results, a review of model
application and statistical computation methods used in the past studies is warranted to provide a better
understanding of the interstudy comparisons. The reviews in Sections 11.6.5 to 11.6.12 were adopted
from the Chesapeake Bay model report (Cerco and Cole 1994). A summary of the characteristics of the
various model applications is given in Table 11-4. It is noteworthy to compare and contrast the duration
of the simulations among the various model applications in the comparison group. With the exception of
the Lake Ontario model, the Peconic Estuary model was run for the longest duration (8 continuous years
of simulation) of any model in the group. Some models were run for as little as a summer season (3
months) whereas others were steady-state simulations. This is important to keep in mind since user-
specified initial conditions can impact model results for several months or even up to as much as a year
beyond model startup. The longer a simulation is run, the less effect initial conditions have on water
quality predictions.
11.6.1 Peconic Estuary Study
The Peconic Estuary model (Tetra Tech 1999) was a three-dimensional application of EFDC, a
fully coupled hydrodynamic and water quality model. The water quality model simulated 22 state
variables. The Peconic model was calibrated and validated over an 8-year continuous time period from
October 1988 to October 1996. Model-data comparisons were computed using the same statistics
defined earlier in this section of the report.
11 - Statistical Summary of Calibration and Validation
11-5
-------
Table 11-4. Summary of various models in comparison group.
Model Application (code)
Spatial
Dimensions
Simulation
Time
State
Variables
Dynamic
or
Steady-State
Sediment
Fluxes
Christina River Basin (CRB)
2D
5 months
05/95-09/95
05/97-09/97
22
Dynamic
Predicted
Peconic Estuary (PE)
3D
8 years
10/88-10/96
22
Dynamic
Predicted
Long Island Sound (LIS)
3D
18 months
4/88 - 10/89
25
Dynamic
Predicted
Massachusetts Bay (BEM)
3D
18 mo. / 1 yr.
10/89-4/91
1/92 - 12/92
24
Dynamic
Predicted
Chesapeake Bay (CB)
3D
3 years
1/84 - 12/87
21
Dynamic
Predicted
Tolo Harbour (TH)
2D-horizontal
2-layer vertical
2 years
1/88 - 12/89
9
Dynamic
Specified
Delaware Inland Bays (DIB)
3D
3 years
21
Dynamic
Specified
New York Bight (NYB)
3D
one summer
21
Dynamic
Specified
Chesapeake Bay Steady-State (CBS)
3D
3 individual
summers (1964,
1984,1985)
NA
Steady-state
Specified
Potomac River Estuary (PR)
ID
6 individual
summers (1968,
69, 70, 77, 78, 79)
9
Dynamic
Specified
Gunston Cove (GC)
2D
one summer
NA
Dynamic
Specified
Eau Galle Reservoir (EGR)
ID
two 6 mo. periods
(Apr - Nov)
NA
NA
Specified
Lake Ontario (LO)
2D
10 years
NA
Dynamic
Specified
NA - not available.
11.6.2 Long Island Sound Study
The Long Island Sound Study model (HydroQual 1991) was a three-dimensional application of a
hydrodynamic model linked to a 25-state variable water quality model. The LISS model was calibrated
over a period of 18 months (April 1988 to September 1989). Model-data comparison statistics were not
presented in the model report. Instead, the various time-series graphs in the report showing model
predictions and observed data were digitized and reverse engineered to create data files that were then
used to compute statistics according to Eq. 11-1 to 11-4. It is possible that only the "best fit" model-data
comparison graphs were presented in the LISS model report; therefore, the statistics computed using
these graphs may overstate model performance to a certain degree.
11-6
11 - Statistical Summary of Calibration and Validation
-------
11.6.3 Massachusetts Bay
The Massachusetts Bay Eutrophication model (HydroQual 1995), called BEM, was a 3-D
application of a hydrodynamic model (ECOM) linked to a 24-state variable water quality model (BEM).
The model was calibrated using data for two time periods, October 1989 through April 1991, and January
through December 1992. Since model-data statistics were not presented in the model report, the various
time-series graphs showing observed data and model predictions were digitized and reverse engineered
into data files for computing statistics according to Eq. 11-1 to 11-4. It is possible that only the "best fit"
model-data comparison graphs were presented in the BEM model report. Therefore, any statistics
computed using these graphs may overstate model performance to a certain degree.
11.6.4 Chesapeake Bay
The Chesapeake Bay model (Cerco and Cole 1994) was a three-dimensional application of the
hydrodynamic model CH3D-WES and eutrophication model CE-QUAL-IC. The sediment processes
model (DiToro and Fitzpatrick 1993) was activated. The model was run continuously for a period of 3
years. Statistics were computed according to the formulas presented earlier in this section. Observations
and predictions were aggregated by season, by spatial zone, and by vertical level. The water quality
model was essentially identical to the EFDC model used in this study.
11.6.5 Tolo Harbour
A two-layer, two-dimensional hydrodynamic model was integrated with a two-layer, two-
dimensional eutrophication model and applied to Tolo Harbour, Hong Kong (Chau and Jin 1998). The
water quality model simulated the transport and transformation of nine water quality constituents
associated with eutrophication. Sediment oxygen demand and benthic nutrient fluxes were specified
based on field monitoring data. The model was run for a period of 2 years (January 1985 through
December 1986). Model-data statistics were not presented in the journal article; however, the various
time-series graphs showing model-data comparisons were digitized and used to compute statistics
according to Eq. 11-1 to 11-4. The model-data comparison graphs presented in the journal article
represented four of seven monitoring stations in Tolo Harbour, so the computed statistics may not be a
true indicator of overall model performance.
11.6.6 Delaware Inland Bays
The Delaware Inland Bays model (Cerco et al. 1993) was a two-dimensional application of the
hydrodynamic (CH3D-WES) and eutrophication (CE-QUAL-IC) components of the Chesapeake Bay
model. The sediment processes model was not activated. The model was run continuously for 3 years.
Statistics were computed according to the formulas presented earlier in this section. Reported results
were for spatial and temporal aggregations comparable to the Chesapeake Bay model study.
11 - Statistical Summary of Calibration and Validation
11-7
-------
11.6.7 New York Bight
The New York Bight model (Hall and Dortch 1993) was a three-dimensional application of the
hydrodynamic (CH3D-WES) and eutrophication (CE-QUAL-IC) components of the Chesapeake Bay
model. The sediment processes model was not activated. The model was run for one summer. Statistics
were computed according to formulas presented earlier in this section. Reported results were for one-to-
one comparisons of predictions and observations (i.e., no aggregration).
11.6.8 Chesapeake Bay Steady-State Model
The steady-state model study of Chesapeake Bay (HydroQual 1987) was a three-dimensional
eutrophication model applied to summer-average conditions. Statistics were reported individually for 3
years (1964, 1984, and 1985). Computation of statistics differed from those shown earlier in this section.
The median absolute error and the median of individual relative errors were selected for comparison with
absolute mean error and relative error in the Peconic Estuary study. The steady-state nature of the model
implied temporal aggregation of model and observations.
11.6.9 Potomac River Estuary Model
The Potomac River Estuary model (Thomann and Fitzpatrick 1982) included a one-dimensional
eutrophication model coupled to a rudimentary sediment model. The median of individual relative errors
was reported for 6 different years (1968, 1969, 1970, 1977, 1978, and 1979). Observations from June
through September in the upper 83 km of the Potomac Estuary were compared to model results.
Unfortunately, the Potomac River Estuary model was compared to various data sets by readjusting the
model parameters for each of the six calibration years. This is not an accepted practice since the purpose
of model calibration and verification is not to "force fit" the model to the data.
11.6.10 Gunston Cove
The Gunston Cove model (Cerco 1985) was a two-dimensional eutrophication model applied for
one summer period to an embayment of the tidal Potomac River. For statistical evaluation, observations
and model predictions were aggregated spatially but not temporally. The root-mean-square error was
used in the computation of the relative error.
11.6.11 Eau Galle Reservoir
The Eau Galle Reservoir model (Wlosinski and Collins 1985) was a one-dimensional
eutrophication model. The model was executed for the period April through November for 2 years. The
relative error was computed as shown in Eq. 11-4 except that "O" was the mean of the observations and
"P" was the mean of the model predictions.
11-8
11 - Statistical Summary of Calibration and Validation
-------
11.6.12 Lake Ontario
The Lake Ontario model (Thomann et al. 1979) was a two-layer eutrophication model. The
median relative error was reported for an analysis of 10 years of data.
11.6.13 Comparison of Absolute Mean Errors
Absolute mean errors for salinity, chlorophyll-a, total nitrogen, total phosphorus, and dissolved
oxygen for the various water quality model studies are shown in Figure 11-1. The absolute mean error of
salinity was near zero for the Christina River model since this was a freshwater system. The other model
studies are for estuary environments having larger salinity values. The absolute mean error of dissolved
oxygen (0.53 mg/L) for the Christina River model for the combined 1995 and 1997 validation periods
was among the better predicting models included in the study. The absolute mean error of chlorophyll-a
(0.80 ug/L) in the Christina River model was less than most of the other models. The absolute mean
error of total phosphorus (0.05 mg/L) and total nitrogen (0.47 mg/L) in the Christina River model was
larger than most of the other models. This was expected since the magnitude of total nitrogen and total
phosphorus in the Christina River system was higher than in the other estuarine models. For example, the
highest total nitrogen concentrations in the Peconic Estuary range from about 0.3 to 1.0 mg/L, whereas
the typical total nitrogen concentrations in the Christina River system were in the 2 to 8 mg/L range.
11.6.14 Comparison of Mean Errors for Dissolved Oxygen
Based on the mean error statistic, 6 of the 11 models for which data were available indicate that
predicted dissolved oxygen concentrations are higher than observations (see Figure 11-2). Only the
Gunston Cove model under predicted dissolved oxygen by a significant amount (about 1.9 mg/L). The
Christina River model under predicted dissolved oxygen during the 1995 and 1997 periods by about
0.20 mg/L. Based on the mean error statistic, the Christina River model is ranked as the fourth best
predictor of dissolved oxygen behind only the Peconic Estuary model, the Massachusetts Bay model
(BEM), and the Tolo Harbor model. The Peconic Estuary model and Massachusetts Bay model have a
mean dissolved oxygen error of near zero which, when considered by itself, may mislead the reader to
assume an almost perfect model-data match. However, by also considering the absolute mean error
statistic shown in Figure 11-1, the reader will understand that offsetting positive and negative "O-minus-
P" pairs of data points in Eq. 11-1 resulted in a mean error of near zero for these two models. The
magnitude of mean errors for the models in this comparative study were Massachusetts Bay (+0.03
mg/L), Peconic Estuary (+0.10 mg/L), Tolo Harbour (+0.16 mg/L), Christina River (+0.20 mg/L),
Chesapeake Bay (-0.50 mg/L), Delaware Inland Bays (-1.25 mg/L), New York Bight (-0.55 mg/L),
Gunston Cove (+1.8 mg/L), Eau Galle Reservoir 1981 (-1.1 mg/L), and Eau Galle Reservoir 1982
(-3.0 mg/L). Although the sample size is small, a common characteristic of eutrophication models seems
11 - Statistical Summary of Calibration and Validation
11-9
-------
to be a general inability to simulate minimum dissolved oxygen levels as inferred by the tendency toward
a negative value of the mean error statistic.
11.6.15 Comparison of Relative Errors
The comparisons of relative errors for dissolved oxygen, chlorophyll, total nitrogen, and total
phosphorus for several model studies are shown in Figures 11-3 to 11-6. Nearly 20 years ago, the median
relative error in a summary of dissolved oxygen models was reported as about 10% (Thomann 1982).
Despite tremendous improvements in model formulation, it is apparent that with the exception of the
Peconic Estuary and BEM models, the 10% relative error standard has not changed much in the past 2
decades. The median relative error in dissolved oxygen derived from models completed after 1982 was
about 9% (not including the Christina River and Peconic Estuary models). The relative error in dissolved
oxygen in the Christina River model was 6.3%, which ranks as 2nd best of the 20 models included in the
comparative study. The average relative error for dissolved oxygen for all the models is 15.1%. The
degree of realism in present-day eutrophication models has improved tremendously. This realism has
removed degrees of freedom available to the modeler to calibrate the model. In other words, some of the
calibration processes are not as amenable to subjective manipulation by the modeler as they were in the
past. For example, the 2-D and 3-D hydrodynamic models have eliminated the use of a dispersion
parameter to transport mass about an estuarine system. The predictive sediment model has eliminated the
oftentimes subjective specification of benthic nutrient fluxes and sediment oxygen demand. The use of
organic carbon as a state variable instead of BOD has eliminated flexibility in converting short-term
measures of BOD to long-term values, which impacts the rate of oxygen consumption.
The relative error of chlorophyll-a for the recent generation of water quality models (Peconic
Estuary, Long Island Sound, Massachusetts Bay, Chesapeake Bay, Tolo Harbour, and Delaware Inland
Bays) ranges from 27 to 75% (Figure 11-4). The Christina River model is the best of this group with a
relative error of 26.9%. The average relative error for chlorophyll-a for all the models is 35.2%. Some
of the older models (Chesapeake Bay Steady-State, Potomac Estuary, Gunston Cove, Eau Galle
Reservoir, and Lake Ontario) seem to give better results, with relative errors in chlorophyll-a ranging
from 10 to 32%. However, most of these models were applied over a short-term duration (a single
season) rather than a full year or multiple years, which removes the seasonal variation from the statistics
and helps to produce a lower relative error. Also, the Potomac River Estuary model was compared to
various data sets by readjusting the model parameters for each calibration run, which improves the fit. In
the Christina River model, the important chlorophyll type is attached algae (periphyton) rather than
floating algae. The floating algae biomass is small compared to the periphyton in the freshwater streams
of the Christina River Basin.
11-10
11 - Statistical Summary of Calibration and Validation
-------
The relative error of total nitrogen for the various models is shown in Figure 11-5. The Christina
River model has a relative error of 15.9%, which ranks as 8th best of the 15 models included in the
comparative study. The average relative error for total nitrogen for all the models is 17.1%. The relative
error of total phosphorus for the various models is presented in Figure 11-6. The Christina River model
has a total phosphorus relative error of 29.8%, which ranks 11th of the 16 models in the comparison
group. This value is only slightly larger than the average relative error for total phosphorus for all the
models (26.2%) in the comparison group. The larger than average relative error for total phosphorus in
the Christina River model may be due to one or more of the following reasons: (1) the NPDES point
source discharges were characterized based on monthly average flows and loads, whereas the in-stream
monitoring data were grab samples taken at a single point in time and therefore reflect any short-term
variations in the point source loading that the model would not be able to resolve; (2) phosphorus loads
from nonpoint sources are based on a constant concentration whereas during storm events, the
concentration would likely increase because of runoff from the watershed; and (3) uptake by aquatic
macrophytes. The model simulates floating algae and periphyton but no other types of macrophytes.
According to the Technical Guidance Manual for Performing Waste Load Allocations (USEPA
1990), acceptable relative error statistic criteria are 15% for dissolved oxygen and 45% for nutrient
parameters (nitrogen, phosphorus, and carbon). The overall relative error statistics for the Christina
River model were 6.3% for dissolved oxygen, 15.9% for total nitrogen, 29.9% for total phosphorus, and
32.6% for total organic carbon. The relative error statistics for the Christina River water quality model
meet the general guidance criteria published in USEPA (1990).
11.7 References for Section 11
Cerco, C. 1985. Water quality in a Virginia Potomac embayment: Gunston Cove. Virginia Institute of Marine
Science, Gloucester Point, VA.
Cerco, C., B. Bunch, M. Cialone, and H. Wang. 1993. Hydrodynamic and eutrophication model study of Indian
River-Rehoboth Bay Delaware. U.S. Army Engineer Waterways Experiment Station, Vicksburg, MS.
Cerco, C.F., and T.M. Cole. 1994. Three-Dimensional Eutrophication Model of Chesapeake Bay. Volume I: Main
Report. Technical Report EL-94-4. U.S. Army Corps of Engineers, Waterways Experiment Station, Vicksburg,
MS, May 1994.
Chau, K.W. and H. Jin. 1998. Eutrophication Model for a Coastal Bay in Hong Kong. ASCE J. Environ. Engr.
124(7):628-638.
Hall, R., and M. Dortch. 1993. New York Bight study report 2 - development and application of a
eutrophication/general water quality model. Technical Report EL-93-xx, U.S. Army Engineer Waterways
Experiment Station, Vicksburg, MS.
HydroQual. 1987. A steady-state coupled hydrodynamic/water quality model of the eutrophication and anoxia
process in Chesapeake Bay. Final report. HydroQual Inc., Mahwah, NJ.
11 - Statistical Summary of Calibration and Validation
11-11
-------
HydroQual. 1991. Water quality modeling analysis of hypoxia in Long Island Sound. HydroQual Inc., Mahwah,
NJ.
HydroQual. 1995. A water quality model for Massachusetts and Cape Cod Bays: calibration of the Bays
Eutrophication Model (BEM). HydroQual, Inc., Mahwah, NJ.
Tetra Tech. 1999. Three-dimensional hydrodynamic and water quality model of Peconic Estuary. For Peconic
Estuary Program, Suffolk County Department of Health Services, Riverhead, NY. By Tetra Tech, Inc., Fairfax,
VA.
Thomann, R., R. Winfield, and J. Segna. 1979. Verification analysis of Lake Ontario and Rochester Embayment
three-dimensional eutrophication models. EPA-600/3-79-094. U.S. Environmental Protection Agency, Duluth,
MN.
Thomann, R. 1982. Verification of water quality models. ASCE J. Environ. Engr. 108(EE5):923-940.
Thomann, R., and J. Fitzpatrick. 1982. Calibration and verification of a mathematical model of the eutrophication of
the Potomac Estuary. HydroQual Inc., Mahwah, NJ.
USEPA. 1990. Technical Guidance Manual for Performing Waste Load Allocations, Book III Estuaries, Part 2,
Application ofEstuarine Waste Load Allocation Models. EPA823-R-92-003. U.S. Environmental Protection
Agency, Office of Water. May 1990.
Wlosinski, J., and C. Collins. 1985. Confirmation of the water quality model CE-QUAL-RI using data from Eau
Galle Reservoir, Wisconsin. Technical Report E-85-11. U.S. Army Engineer Waterways Experiment Station,
Vicksburg, MS.
11-12
11 - Statistical Summary of Calibration and Validation
-------
Salinity
4.0 -
3.5 -
c 3.0 -
ro
3 2.5 -
O
.c
2.0 -
-------
2.0
1.0
0.0
JUL
JZZL
-1.0
-2.0
-3.0
CRB
PE
LIS
BEM
CB
DIB
TH
NYB
GC EGR81 EGR82
Legend:
CRB
Christina River Basin 1995 & 1997 (Tetra Tech 2000)
PE
Peconic Estuary (Tetra Tech 1999)
LIS
Long Island Sound (HydroQual 1991)
BEM
Massachusetts Bay (HydroQual 1995)
CB
Chesapeake Bay (Cerco and Cole 1994)
DIB
Delaware Inland Bays (Cerco et al. 1993)
TH
Tolo Harbour (Chau and Jin 1998)
NYB
New York Bight (Hall and Dortch 1993)
GC
Gunston Cove (Cerco 1985)
EGR81
Eau Galle Reservoir 1981 (Wlosinski and Collins 1985)
EGR82
Eau Galle Reservoir 1982 (Wlosinski and Collins 1985)
Figure 11 -2. Mean dissolved oxygen error for several model studies.
11-14
11 - Statistical Summary of Calibration and Validatioi
-------
Dissolved Oxygen
60%
50%-
40%-
o
LU
is
a;
q:
20%-
10%
o%
~nU~UUIIIIII.I~~
CRB LIS CB TH CBS65 CBS85 PR69 PR77 PR79 EGR81
PE BEM DIB NYB CBS84 PR68 PR70 PR78 GC EGR82
Figure 11-3. Relative error in dissolved oxygen for several water quality models.
Legend:
CRB
Christina River Basin 1995 & 97 (Tetra Tech 2000)
TH
Tolo Harbour (Chau and Jin 1998)
PE
Peconic Estuary (Tetra Tech 1999)
CBS
Chesapeake Bay Steady-State (HydroQual 1987)
LIS
Long Island Sound (HydroQual 1991)
PR
Potomac River (Thomann & Fitzpatrick 1982)
BEM
Massachusetts Bay (HydroQual 1995)
GC
Gunston Cove (Cerco 1985)
CB
Chesapeake Bay (Cerco & Cole 1994)
EGR
Eau Galle Reservoir (Wlosinski & Collins 1985)
DIB
Delaware Inland Bays (Cerco et al. 1993)
LO
Lake Ontario (Thomann et al. 1979)
Chlorophyll
80%-|
70%--- - -
60%--- ...................................
2 50% — — J—¦
LU
> 40% |—| r~i
+J I
30%
20%
10%
0%
~
~
CRB LIS CB TH CBS84 PR68 PR70 PR78 GC EGR82
PE BEM DIB CBS65 CBS85 PR69 PR77 PR79 EGR81 LO
Figure 11-4. Relative error in chlorophyll for several water quality models.
11 - Statistical Summary of Calibration and Validation
11-15
-------
35%™
o
LU :
JS
-I—'
TO
0 20%
a.
1 o%
o%
JnU
CBS65 CBS84 CBS85 PR68 PR69 PR70 PR77 PR78
Figure 11 -6. Relative error in total phosphorus for several water quality models.
* Total phosphorus not available for BEM; relative error in DIP is shown instead.
** Total phosphorus not available for TH; relative error in DOP is shown instead.
11-16
11 - Statistical Summary of Calibration and Validation
-------
12 - SUMMARY AND CONCLUSIONS
12.1 Summary of EFDC Hydrodynamic and Water Quality Model Framework
The time-dependent, multidimensional Environmental Fluid Dynamics Code (EFDC) provided
the modeling framework for this study. EFDC solved prognostic equations for free-surface elevation,
velocity components, temperature, salinity, and turbulence energy. All equations were written in
curvilinear, coastline-fitted coordinate systems combined with a free-surface and bottom following
sigma-coordinate system. An imbedded turbulence closure submodel was employed to provide vertical
mixing coefficients for momentum, temperature, and salinity.
A high spatial resolution grid was employed to resolve the important physical processes
operating in the Christina River Basin. The horizontal grid spacing along the streams ranged from 500 to
about 1,000 meters to provide adequate resolution. The vertical direction was resolved by a single layer.
The model was driven by data sets of tidal elevations, salinity, and temperature at the north and south
open boundaries, as well as by winds, solar radiation, and point and nonpoint source discharges.
A suite of 22 state variables was required to model the eutrophication processes in the water
column (see Table 4-1). Three variables (salinity, water temperature, and total suspended sediment),
which are necessary for certain computations involving the 22 state variables, were provided by the
EFDC hydrodynamic model. The interactions among the state variables were shown in Figure 4-1.
Kinetic interactions affecting the state variables were described in over 80 partial differential
equations that required evaluation of more than 130 parameters. The kinetics described carbon,
phosphorus, nitrogen, and silica cycles as well as the dissolved oxygen balance. Algal production is the
primary source of carbon, although carbon also enters the system through external loads. Predation on
algae releases particulate and organic carbon to the water column, a portion of which undergoes first-
order dissolution to dissolved organic carbon, and the remainder settles to the bottom sediments. The
kinetic rates used for model calibration are provided in the listing of the model input data in Appendix G.
External loads provide the ultimate source of phosphorus to the system. Dissolved phosphate is
consumed by algae during growth and is released through respiration and predation as phosphate and
organic phosphorus. A portion of the particulate organic phosphorus hydrolyzes to dissolved organic
phosphorus, and the remaining balance settles to the bottom. Dissolved organic phosphorus in the water
column is mineralized to phosphate, a portion of which sorbs to inorganic solids and settles to the
12- Summary and Conclusions
12-1
-------
bottom. Within the sediment layer, particulate phosphorus is mineralized and recycled back into the
water column as dissolved phosphate.
External loads provide the primary source of nitrogen to the Christina River Basin system.
Inorganic nitrogen is consumed by algae and released as ammonia and organic nitrogen through
respiration and predation. A portion of the particulate organic nitrogen hydrolyzes to dissolved organic
nitrogen and the remaining balance settles to the bottom sediments. Dissolved organic nitrogen in the
water column is mineralized to ammonia. Depending on the concentration of oxygen in the water
column, a fraction of the ammonia is oxidized to nitrate through the nitrification process, or nitrate is lost
to nitrogen gas through denitrification. Particulate nitrogen settles to the bottom where it is mineralized
and recycled to the water column as ammonia. Nitrate moves in both directions across the sediment-
water interface depending on the relative concentrations in the water column and sediment pore water.
In the silica cycle, diatoms consume the available silica and recycle both available and
particulate biogenic silica through the actions of metabolism and predation. Particulate silica dissolves in
the water column or settles to the bottom. A portion of the settled particulate biogenic silica dissolves
within the sediments and returns to the water column as available silica. The sources and sinks of
dissolved oxygen in the water column are algal photosynthesis, algal respiration, atmospheric reaeration,
nitrification, and chemical oxygen demand.
12.2 Summary of Hydrodynamic Results
An extensive water quality database was used to calibrate a high-resolution, physically
comprehensive hydrodynamic and water quality model of the freshwater and tidal streams in the
Christina River Basin. The model was driven by data sets of tidal elevations, salinity, and temperature at
the north and south open boundaries on the Delaware River, as well as by winds, solar radiation, and
point and nonpoint source discharges.
The period May 1 to September 21, 1997, was chosen for model calibration because of available
detailed field measurements by Davis (1998) and because the month of August during this period was
characterized by stream flows approaching the 7Q10 flow rate. The month of August 1997 was used for
tide calibration because tide elevation data were available from two USGS tide gages on the Christina
River. Model-data comparisons included water surface elevations, stream flow, velocity, and chlorides.
It was apparent from the calibration results that the model is well suited to predict the hydrodynamic
characteristics of both the freshwater and the tidal streams in the Christina River Basin. The model was
validated during the period May 1 to September 21, 1995.
72-2
12- Summary and Conclusions
-------
12.3 Summary of Water Quality Results
It can be debated that using a highly sophisticated, fully dynamic hydrodynamic and
eutrophication model such as EFDC is not warranted for a steady-state, low-flow study. However, this
calibrated model represents the first phase of a much larger project that will require the dynamic
capabilities of EFDC, namely, linking to an HSPF watershed runoff model of the Christina River Basin.
The calibration period for the water quality component (May 1 to September 21, 1997) included
substantial instream and point source monitoring data collected by PADEP, DNREC, USGS, and Davis
(1998) for model calibration. The model was validated during the period May 1 to September 21, 1995.
Comparison of the EFDC water quality model predictions with observations indicated the following
characteristics:
• Particular attention was given to reproduction of the August 1997 water column
concentrations of chlorides, chlorophyll, carbon, nitrogen, and phosphorus in the estuary and
freshwater stream reaches. Comparisons of predicted and observed data for all parameters
were considered to be reasonable in all 11 major stream reaches included in the model.
• The magnitude of the chlorophyll-a concentrations was replicated quite well in all 11 stream
reaches.
• The longitudinal concentration gradients of phosphorus, nitrogen, and organic carbon species
were replicated reasonably well throughout the system.
• The daily average dissolved oxygen concentrations as well as the daily range in DO agreed
well with the observations.
• The model results during the validation period, which experienced flow rates below 7Q10
values, agreed with the observations in a manner similar to the calibration period.
• Although no data were available to compare with the predicted sediment oxygen demand and
benthic nutrient flux rates during the calibration and validation periods, the fact that the
water column concentrations of oxygen and nutrients compared well with the data provides
an indirect confirmation that the predictive sediment submodel is operating reasonably well.
Also, SOD data were available from July and August 1996 at two locations in the tidal
Christina River and one location in the tidal Brandywine Creek, and the model agreed
reasonably well with these limited observations.
12.4 Sources of Uncertainty
In modeling any large and complex system the size of the Christina River Basin, there will
always be many possible uncertainties in the model input data (e.g., boundary conditions, loadings, and
kinetic rate parameters). The EFDC model incorporates 22 water quality state variables, and 19 of those
were required for the Christina River Basin application. The nutrients require the partitioning of carbon,
12- Summary and Conclusions
12-3
-------
nitrogen, and phosphorus organic material into dissolved and particulate forms, and the particulate matter
is further split into labile and refractory matter. This detailed information was not available from the
point source monitoring records or from the STORET monitoring data. To fill in these data gaps, an
assumption was made that the organic matter from point and nonpoint sources was generally partitioned
into 50% dissolved, 25% labile particulate, and 25% refractory particulate. In addition, the only nutrients
generally available from discharge monitoring records were ammonia nitrogen and total phosphorus.
The rules for estimating the other nitrogen and phosphorus species were presented in Table 6-4,
Table 6-5, and Table 7-4.
The estimated loadings from nonpoint sources are also subject to uncertainty. For this low-flow
study, nonpoint source loading rates were computed using a constant concentration reflective of
conditions in low-flow periods in the summer and estimates of daily discharge rates. In reality, the
concentrations of the various water quality parameters will also change in accordance with storm events
due to associated runoff from the watersheds. When the HSPF watershed runoff model of the Christina
River Basin is completed, the uncertainty in the nonpoint source loading rates should be reduced because
the watershed model will be computing calibrated washoff loads. It was beyond the scope of this study
to estimate the watershed runoff loading during rain events.
Detailed information on stream geometry was available only at selected locations, including the
four areas sampled in the August 1997 study (Davis 1998), as well as at several cross-sections in the tidal
Christina River, Red Clay Creek, and some of the other smaller tributaries. Very crude information on
stream geometry was obtained from HEC-2 cross-section data obtained from FEMA. However, since the
HEC-2 data were developed for flood studies, they do always include sufficient detail to resolve the low-
flow stream channels. Vegetative cover shields many portions of the stream reaches from direct
sunshine, which can have a profound effect on localized chlorophyll photosynthesis. The present model
incorporates a light reduction factor due to vegetative shading that is adjustable for each grid cell.
However, for this calibration the shade factor has been set to 1.0 (i.e., no light reduction due to shade
cover) for all grid cells because detailed information on areas affected by canopy cover was not available.
The WWTPs have permit limits for CBOD5 and the discharge monitoring records report
CBOD5, whereas the model uses organic carbon instead of CBOD. A special study was conducted in
August and September 1999 in which 14 of the largest dischargers in the Christina River Basin were
asked to collect effluent data and analyze it for CBOD5, CBOD20, total organic carbon, and dissolved
organic carbon content. These data were used to determine the CBODu/CBOD5, DOC/TOC, and
TOC/CBODu ratios for the rules listed in Table 7-4.
12-4
12- Summary and Conclusions
-------
12.5 Conclusions
The dynamic simulation of eutrophication in a freshwater and estuarine system is a very
complicated and computationally intensive endeavor because a large number of chemical, biological, and
biochemical processes interact, and the reaction rates and external inputs vary with time. In addition, the
flow rates and associated circulation are also time-varying, having time scales ranging from minutes to
months or even years in the case of sediment flux recovery.
The present EFDC hydrodynamic and water quality model of the freshwater and tidal streams in
the Christina River Basin represents the current state of the art in eutrophication modeling. The original
scope of work for this project was designed around a model framework based on EPA's WASP model.
The framework was changed to the EFDC model because it provides several advances over a WASP
model application. First, the coupling of the model to a three-dimensional, time-varying hydrodynamic
model provides more realistic circulation physics of the tidal waters in the system. Second, the water
quality model itself includes an expanded suite of 22 state variables (the EPA WASP model includes
only 8 state variables). Third, the coupling to a fully predictive sediment process model allows the
simulation of sediment oxygen demand and nutrient fluxes. Fourth, the model simulates the growth of
attached algae (periphyton), which is the primary force driving the large diel dissolved oxygen swings
observed in certain stream reaches (WASP does not include a periphyton state variable).
The EFDC hydrodynamic and water quality model of the Christina River Basin meets or exceeds
the goals specified at the initiation of the project. Even though a number of potential sources of
uncertainty were outlined in Section 12.4, the model exhibits a high degree of correspondence to
observations monitored in the estuary and stream reaches. The calibration and validation statistics for
the Christina River water quality model were presented in Chapter 11. According to the Technical
Guidance Manual for Performing Waste Load Allocations (USEPA 1990), acceptable relative error
statistic criteria are 15% for dissolved oxygen and 45% for nutrient parameters (nitrogen, phosphorus,
and carbon). The overall relative errors of the Christina River model were 6.3% for dissolved oxygen,
15.9% for total nitrogen, 29.9% for total phosphorus, and 32.6% for total organic carbon. Based on the
calibration and validation results, the model is considered to be adequately calibrated and is acceptable as
a tool for TMDL management of nutrient, dissolved oxygen, and eutrophication issues in the Christina
River Basin.
12- Summary and Conclusions
12-5
-------
This page intentionally left blank
12- Summary and Conclusions
-------
13-REFERENCES
Ambrose, R.B., T.A. Wool, and J.L. Martin. 1993. The water quality analysis and simulation program, WASP5:
Part A, model documentation version 5.1. U.S. Environmental Protection Agency, Environmental Research
Laboratory, Athens, GA, 210 pp.
Andrews, D.G., and M.E. Mclntyre. 1978. An exact theory for of nonlinear waves on a Lagrangian flow. J. Fluid
Mech. 89:609-646.
Banks, R.B. and F.F. Herrera. 1977. Effect of wind and rain on surface reaeration. ASCE J. of the Environ. Engr.
Div. 103(EE3):489-504.
Barrow, N.J. 1983. A mechanistic model for describing the sorption and desorption of phosphate by soil. J. of Soil
Science 34:733-750.
Bennett, A.F. 1976. Open boundary conditions for dispersive waves. J. Atmos. Sci. 32:176-182.
Bennett, A.F., and P.C. Mcintosh. 1982. Open ocean modeling as an inverse problem: tidal theory. J. Phys. Ocean.
12:1004-1018.
Bennett, J.R., and A.H. Clites. 1987. Accuracy of trajectory calculation in a finite-difference circulation model. J.
Comp. Phys. 68:272-282.
Blumberg, A.F., and L.H. Kantha. 1985. Open boundary condition for circulation models. J. Hydr. Engr.
111:237-255.
Blumberg, A.F., and G.L. Mellor. 1987. A description of a three-dimensional coastal ocean circulation model.
Three-Dimensional Coastal Ocean Models, Coastal andEstuarine Science, Vol. 4, ed. N.S. Heaps, pp. 1-19.
American Geophysical Union.
Boni, L., E. Carpene, D. Wynne, and M. Reti. 1989. Alkaline phosphatase activity in Protogonyaulax tamarensis.
J. of Plankton Research ll(5):879-885.
Boudreau, B.P. 1991. Modelling the sulfide-oxygen reaction and associated pH gradients in porewaters.
Geochimica et Cosmochimica Acta, 55:145-159.
Bowie, G.L., W.B. Mills, D.B. Porcella, C.L. Campbell, J.R. Pagenkopf, G.L. Rupp, K.M. Johnson, P.W.H.Chan,
S.A. Gherini, and C.E. Chamberlin. 1985. Rates, constants and kinetics formulations in surface water quality
modeling (2nd edition). EPA/600/3-85/040, U.S. Environmental Protection Agency, Environmental Research Lab.,
Athens, GA, 455 pp.
Boynton, W., W. Kemp, and C. Keefe. 1982. A comparative analysis of nutrients and other factors influencing
estuarine phytoplankton production. Estuarine Comparisons, Academic Press, pp 69-90.
Burban, P.Y., W. Lick, and J. Lick. 1989. The flocculation of fine-grained sediments in estuarine waters. J.
Geophys. Res. 94:8323-8330.
Burban, P.Y., Y.J. Xu, J. McNeil, and W. Lick. 1990. Settling speeds of floes in fresh and seawater. J. Geophys.
Res. 95:18,213-18,220.
Carritt, D.E. and S. Goodgal. 1954. Sorption reactions and some ecological implications. Deep-Sea Research
1:224-243.
Caupp, C.L., Brock, J.T., and Runke, H.M. 1991. Application of the dynamic stream simulation and assessment
model (DSSAM III) to the Truckee River below Reno, Nevada: Model formulation and program description.
13- References
13-1
-------
Report prepared by Rapid Creek Water Works for Nevada Division of Environmental Protection, Carson City and
Washoe County Dept. Of Comprehensive Planning, Reno, NV.
Cerco, C.F., and T. Cole. 1993. Three-dimensional eutrophication model of Chesapeake Bay. J. Environ. Engnr.
119:1006-1025.
Cheng, R.T., V. Casulli, and J.W. Gartner. 1993. Tidal, residual, intertidal mudflat (TRIM) model and its
applications to San Francisco Bay, California. Estuarine, Coastal, and Shelf Science 36: 235-280.
Chrost, R.J. and J. Overbek. 1987. Kinetics of alkaline phosphatase activity and phosphorus availability for
phytoplankton and bacterioplankton in Lake PluBsee (North German eutrophic lake). Microbial Ecology 13:229-
248.
Cline, J.D. and F.A. Richards. 1969. Oxygenation of hydrogen sulfide in seawater at constant salinity, temperature
andpH. Environmental Science & Technology, 3(9):838-843.
Cole, T.M., and E.M. Buchak. 1994. CE-QUAL-W2: A two-dimensional laterally averaged, hydrodynamic and
water quality model, Version 2.0, Report ITL-93-7. U. S. Army Corps of Engineers, Waterway Experiment Station,
Vicksburg, MS.
Davis, J. 1998. Measurement of community photosynthesis and respiration rates for selected reaches of the
Christina Watershed. Prepared for Pennsylvania Department of Environmental Protection and Delaware
Department of Natural Resources and Environmental Control, March 1998.
Diaz, R.J. and R. Rosenberg. 1995. Marine benthic hypoxia: a review of its ecological effects and the behavioural
responses of benthic macrofauna. Oceanography and Marine Biology: an Annual Review 33:245-303.
DiToro, D.M. 1980. Applicability of cellular equilibrium and Monod theory to phytoplankton growth kinetics.
Ecological Modelling 8:201-218.
DiToro, D.M., P.R. Paquin, K. Subburamu, and D.A. Gruber. 1990. Sediment oxygen demand model: methane and
ammonia oxidation. ASCE J. Environ. Engr. 116(5):945-986.
DiToro, D.M., and J.F. Fitzpatrick. 1993. Chesapeake Bay sediment flux model. Contract Report EL-93-2.
Prepared by HydroQual, Inc. for U. S. EPA Chesapeake Bay Program, U. S. Army Engineer District, Baltimore,
MD, and U.S. Army Engineer Waterways Exp. Station.
Frick, W.E. 1984. Non-cmpirical closure of the plume equations, Aliuos. Environ. 18:653-662.
Froelich, P.N. 1988. Kinetic control of dissolved phosphate in natural rivers and estuaries: a primer on the
phosphate buffer mechanism. Limnol. and Oceanogr. 33(4, part 2):649-668.
Galperin, B., L.H. Kantha, S. Hassid, and A. Rosati. 1988. A quasi-equilibrium turbulent energy model for
geophysical flows. J.Atmos. Sci. 45:55-62.
Genet, L. A., D.J. Smith, and M.B. Sonnen, M.B. 1974. Computer program documentation for the dynamic estuary
model. Prepared for U.S. Environmental Protection Agency, Systems Development Branch, Washington, D.C.
Grant, W.D., and O.S. Madsen. 1986. The continental-shelf bottom boundary layer. Annual Review of Fluid
Mechanics, ed. M. Van Dyke et al., pp. 365-306, Annual Review, Inc.
Giordani, P. and M. Astorri. 1986. Phosphate analysis of marine sediments. Chemistry in Ecology 2:103-112.
13-2
13- References
-------
Hamrick, J.M. 1992a. A Three-Dimensional Environmental Fluid Dynamics Computer Code: Theoretical and
Computational Aspects, Special Report 317. The College of William and Mary, Virginia Institute of Marine
Science. 63 pp.
Hamrick, J.M. 1992b. Estuarine environmental impact assessment using a three-dimensional circulation and
transport model. Estuarine and Coastal Modeling, Proceedings of the 2nd International Conference, M. L.
Spaulding et al, Eds., American Society of Civil Engineers, New York, 292-303.
Hamrick, J. M. 1994: Linking hydrodynamic and biogeochemcial transport models for estuarine and coastal waters.
Estuarine and Coastal Modeling, Proceedings of the 3rd International Conference, ed. M.L. Spaulding et al.,
pp. 591-608. American Society of Civil Engineers, New York.
Hamrick, J.M. 1996. A User's Manual for the Environmental Fluid Dynamics Computer Code (EFDC), Special
Report 331. The College of William and Mary, Virginia Institute of Marine Science. 234 pp.
Hamrick, J.M., and T.S. Wu. 1996. Computational design and optimization of the EFDC/HEM3D surface water
hydrodynamic and eutrophication models. Computational Methods for Next Generation Environmental Models,
ed. G. Delich. Society of Industrial and Applied Mathematics, Philadelphia.
Hecky, R. and P. Kilham. 1988. Nutrient limitation of phytoplankton in freshwater and marine environments: a
review of recent evidence on the effects of enrichment. Limnol. And Oceanogr. 33(4):796-822.
Horner, R.R., E.B. Welch, M.R. Seeley, and J.M. Jacoby. 1990. Responses of periphyton to changes in current
velocity, suspended sediment, and phosphorus concentration. Freshwater Biol. 24(2):215-232.
HydroQual. 1991. Draft Water Quality Modeling Analysis of Hypoxia in Long Island Sound. Prepared for
Management Committee Long Island Sound Estuary Study and New England Interstate Water Pollution Control
Commission. Prepared by HydroQual, Inc., Mahwah, NJ. Job Number: NENGOO 12. July 1991.
Jirka, G. H., and R.L. Doneker. 1991. Hydrodynamic classification of submerged single-port discharges, J. of
Hydr.Engr. 117:1095-1112.
Jirka, G.H., and P.J. Akar. 1991. Hydrodynamic classification of submerged multiport-diffuser discharges, J. of
Hydr. Engr. 117:1113-1128.
Kang, I.S., and L.G. Leal. 1992. Orthogonal grid generation in a 2D domain via the boundary integral technique.
J. Comp. Phys. 102:78-87.
Knorr, D.F. and G.W. Fairchild. 1987. Periphyton, benthic invertebrates and fishes as biological indicators of
water quality in the East Branch Brandywine Creek. Proceedings of the Pennsylvania Academy of Science
61(l):62-66.
Kremer, J.N. and S.W. Nixon. 1978. A coastal marine ecosystem: simulation and analysis. Ecological Studies 24,
Springer-Verlag, New York. 217 pp.
Lebo, M.E. 1991. Particle-bound phosphorus along an urbanized coastal plain estuary. Marine Chemistry 34:225-
246.
Lee, J.H.W., and V. Cheung. 1990. Generalized Lagrangian model for buoyant jets in a current. J. Environ. Engrg.
116:653-662.
Lijklema, L. 1980. Interaction of ortho-phosphate with iron (III) and aluminum hydroxides. Environmental
Science & Technology, 14(5):537-541.
13- References
13-3
-------
Madsen, P. A., and J. Larsen. 1987. An efficient finite-difference approach to the mild-slope equation. Coastal
Engr. 11:329-351.
Mancini, J.L. 1983. A method for calculating effects, on aquatic organisms, of time varying concentrations. Water
Research 17(10): 1355-1362.
Matisoff, G. 1982. Mathematical models of bioturbation, ed. P.L. McCall and M.J.S. Tevesz, pp. 289-330. Animal-
Sediment Relations: The Biogenic Alteration of Sediments, Plenum Press, NY.
Mellor, G.L., and T. Yamada. 1982. Development of a turbulence closure model for geophysical fluid problems.
Rev. Geophys. Space Phys., 20:851-875.
Millero, F.J. 1986. The thermodynamics and kinetics of the hydrogen sulfide system in natural waters. Marine
Chemistry 18:121-147.
Mobley, C.D., and R.J. Stewart. 1980. On the numerical generation of boundary-fitted orthogonal Curvilinear
coordinate systems. J. Comp. Phys. 34:124-135.
Morel, F. 1983. Principles of Aquatic Chemistry. John Wiley & Sons, New York, NY. 446 pp.
Morse, J.W., F.J. Millero, J.C. Cornwell, and D. Rickard, D. 1987. The chemistry of the hydrogen sulfide and iron
sulfide systems in natural waters. Earth-Science Reviews 24:1-42.
Moustafa, M.Z., and J.M. Hamrick. 1994. Modeling circulation and salinity transport in the Indian River Lagoon.
Estuarine and Coastal Modeling, Proceedings of the 3rd International Conference, ed. M. L. Spaulding et al.,
pp. 381-395. American Society of Civil Engineers, New York.
Mclntire, C.C. 1973. Periphyton dynamics in laboratory streams: a simulation model and its implications.
Ecological Monographs. 43:399-420.
Nixon, S. 1981. Remineralization and nutrient cycling in coastal marine ecosystems. Estuaries and Nutrients, ed.
Neilson and Cronin. pp. 111-138. Humana Press, Clifton, NJ.
NOAA. 1998. Tide Tables 1998. National Oceanographic and Atmospheric Administration, National Ocean
Service, Silver Spring, MD.
O'Connor, D.J. and W.E. Dobbins. 1958. Mechanism of reaeration in natural streams. Transactions of the
American Society of Civil Engineers, 123(2934):641-684.
Odum, E.P. 1971. Fundamentals of ecology (3rd edition). W.B. Saunders Co., Philadelphia, PA. 574pp.
Oey, L.-Y., G.L. Mellor, and R.I. Hires. 1985. A three-dimensional simulation of the Hudson Raritan estuary. J.
Phys. Oceanogr. 15:1676-1720.
Park, K., A.Y. Kuo, J. Shen, and J.M. Hamrick. 1995. A three-dimensional hydrodynamic-eutrophication model
(HEM3D): description of water quality and sediment processes submodels, Special Report 327. The College of
William and Mary, Virginia Institute of Marine Science. 113 pp.
Parsons, T.R., M. Takahashi, and B. Hargrave. 1984. Biological oceanographic processes (3rd edition). Pergamon
Press. 330 pp.
Pfeifer, R.F., and McDiffett, W. 1975. Some factors affecting primary production of stream communities.
Archives ofHydrobiol. 75:306-317.
13-4
13- References
-------
Press, W.H., B.P. Flannery, S. A. Teukolsky, and W.T. Vetterling. 1986. Numerical recipes: the art of scientific
computing. Cambridge University Press. 818 pp.
Redfield, A.C., B.H. Ketchum, and F.A. Richards. 1963. The influence of organisms on the composition of sea-
water, ed. M.N. Hill, Chapter 2, pp 26-77. The Sea - Ideas and Observations on Progress in the Study of the Seas:
Vol. 2, Composition of Sea Water, Comparative and Descriptive Oceanography, Interscience Publishers.
Reif, A.G. 1999. Physical, chemical, and biological data collected for selected streams in Chester County,
Pennsylvania, 1981-1994.
Rennie, S., and J.M. Hamrick. 1992. Techniques for visualization of estuarine and coastal flow fields. Estuarine
and Coastal Modeling, Proceedings of the 2nd International Conference, ed. M. L. Spaulding et al., pp. 48-55.
American Society of Civil Engineers, New York.
Robbins, J.A., T. Keilty, D.S. White, and D.N. Edgington. 1989. Relationships among tubificid abundances,
sediment composition and accumulation rates in Lake Erie. Canadian J. of Fisheries & Aquatic Sciences,
46(2):223-231.
Rosati, A.K., and K. Miyakoda. 1988. A general circulation model for upper ocean simulation. J. Phys. Ocean,
18:1601-1626.
Ross, P.J. 1983. Dynamics of periphyton communities. Periphyton in Freshwater Ecosystems, ed. R.G. Wetzel,
pp. 5-10. Junk, Boston, MA, 5-10.
Runke, H.M. 1985. Simulation of the lotic periphyton community of a small mountain stream by digital computer.
PhD thesis, Utah State University, Logan, Utah.
Ryskin, G. andL.G. Leal. 1983. Orthogonal mapping. J. Comp. Phys. 50:71-100.
Sand-Jensen, K. 1983. Physical and chemical parameters regulating the growth of periphyton communities.
Periphyton in Freshwater Ecosystems, ed. R.G. Wetzel, pp. 63-71. Junk, Boston, MA.
Senior, L.A. 1999. Background concentrations for Christina River Model. U.S. Geological Survey, Malvern, PA.
Memorandum to M.R. Morton, Tetra Tech, Inc. dated August 6, 1999.
Shen, J., J. Boon, and A.Y. Kuo. 1999. A numerical study of a tidal intrusion front and its impact on larval
dispersion in the James River estuary, Virginia. Estuary 22(3A):681-692.
Smolarkiewicz, P.K., and T.L. Clark. 1986. The multidimensional positive definite advection transport algorithm:
further development and applications. J. Comp. Phys. 67:396-438.
Smolarkiewicz, P.K., and W.W. Grabowski. 1990. The multidimensional positive definite advection transport
algorithm: nonoscillatory option. J. Comp. Phys. 86:355-375.
Smolarkiewicz, P.K., and L.G. Margolin. 1993. On forward-in-time differencing for fluids: extension to a
curvilinear framework. Mon. Weather Rev. 121:1847-1859.
Steele, J.H. 1962. Environmental control of photosynthesis in the sea. Limnol. and Oceanogr. 7(2): 137-150.
Stevenson, R.J., and Glover, R. 1993. Effects of algal density and current on ion transport through periphyton
communities. Limnol. and Oceanogr., 38(6):1276-1281.
Strickland, J. 1960. Measuring the production of marine phytoplankton. Fisheries Research Board of Canada
Bulletin 122:172.
13- References
13-5
-------
Stumm, W. and J.J. Morgan. 1981. Aquatic chemistry, an introduction emphasizing chemical equilibria in natural
waters (2nd edition). John Wiley & Sons, Inc. 780 pp.
Thomann, R.V., N.J. Jaworski, S.W. Nixon, H.W. Paerl, and J. Taft. 1985. The 1983 algal bloom in the Potomac
Estuary. The Algal Bloom Expert Panel, pepared for the Potomac Strategy State/EPA Management Committee, US
Environmental Protection Agency, Region III, Philadelphia, PA.
Thomann, R.V. and J. A. Mueller. 1987. Principles of surface water quality modeling and control. Harper & Row,
Publishers, Inc. 644 pp.
Troup, R. 1974. The interaction of iron with phosphate, carbonate and sulfide in Chesapeake Bay interstitial
waters: a thermodynamic interpretation. Ph.D. Dissertation, Johns Hopkins University, MD. 114 p.
USEPA. 1990. Technical Guidance Manual for Performing Waste Load Allocations, Book III Estuaries, Part 2,
Application ofEstuarine Waste Load Allocation Models. EPA823-R-92-003. U.S. Environmental Protection
Agency, Office of Water. May 1990.
USEPA. 1995. Technical Guidance Manual for Developing Total Maximum Daily Loads, Book II: Streams and
Rivers, Part 1: Biochemical Oxygen Demand/Dissolved Oxygen andNutrients/Eutrophication. EPA 823-B-95-007.
U.S. Environmental Protection Agency, Office of Water. September 1995.
Warwick, J.J., D. Cockrum, and M. Horvath. 1997. Estimating non-point source loads and associated water quality
impacts. ASCEJ. Water Res. Plan, and Manage. 123(5):302-310.
Westrich, J.T. and B.A. Berner. 1984. The role of sedimentary organic matter in bacterial sulfate reduction: the G
model tested. Limnol. and Oceanogr. 29(2):236-249.
Wezernak, C.T. and J.J. Gannon. 1968. Evaluation of nitrification in streams. ASCEJ. Sanitary Engr. Div.
94(5):883-895.
Whitford, L.A. and G.J. Schumacher. 1964. Effect of current on respiration and mineral uptake of Spirogyra and
Oedogonium. Ecology 45:168-170.
Wu, T.S., J.M. Hamrick, S.C. McCutechon, and R.B. Ambrose. 1996. Benchmarking the EFDC/HEM3D surface
water hydrodynamic and eutrophication models. Computational Methods for Next Generation Environmental
Models, ed. G. Delich. Society of Industrial and Applied Mathematics, Philadelphia.
Yamamoto, S., J.B. Alcauskas, and T.E. Crozler. 1976. Solubility of methane in distilled water and seawater. J. of
Chemical and Engineering Data 21(l):78-80.
13-6
13- References
------- |