The document summarizes a water quality modeling project for the Lower Cape Fear River estuary. It describes: 1) The water quality model used inputs like river flows, temperatures, and point source loads to simulate dissolved oxygen, nutrients, and organic matter. 2) The model was calibrated for 2004 using monitoring data and verified against 2005 data, with good agreement found for salinity, temperature, dissolved oxygen and other constituents. 3) The partitioning of organic matter into labile and refractory components improved the model's ability to simulate long-term biochemical oxygen demand measurements in the estuary.

1. LCFR Water Quality Modeling
Project Report
Jim Bowen, UNC Charlotte
LCFRP Advisory Board/Tech. Comm.
Meeting, October 30, 2008
Raleigh, NC

2. Outline of Presentation
• A Quick Review of the LCFR Model
• Summary of Model Report
• Questions/Suggestions

3. Basis of
Presentation
Technical
Report
Draft
(available on web)

4. LCFR Dissolved Oxygen Model
The big picture
Estuary Physical
Characteristics:
e.g. length,
width, depth,
roughness
EFDC Software
Adjustable Parameters:
(e.g. BOD decay, SOD,
reaeration)
Hydrologic Conditions
River
Flows,
Temp’s,
Conc’s
Tides Time
“Met” Data
Air temps,
precip,
wind,
cloudiness
Time
State Variables
nutrients
DO, organic C
Time

5. Dissolved Oxygen Conceptual Model
BOD Sources
Sediment
Cape Fear, Black &
NECF BOD Load
Muni & Ind.
BOD Load
decaying
phytopl.
Estuary Inflow
BOD Load

6. Dissolved Oxygen Conceptual Model
BOD Sources, DO Sources
Sediment
Cape Fear, Black &
NECF BOD Load
Ocean Inflows
Surface
Reaeration
Phytoplank.
Productivity
Muni & Ind.
BOD Load
decaying
phytopl.
MCFR Inflows
Estuary Inflow
BOD Load

7. BOD
Consumption
Dissolved Oxygen Conceptual Model
BOD Sources, DO Sources & Sinks
Sediment Sediment O2 Demand
Cape Fear, Black &
NECF BOD Load
Ocean Inflows
Surface
Reaeration
Input of NECF &
Black R. Low DO
Water
Phytoplank.
Productivity
Muni & Ind.
BOD Load
decaying
phytopl.
MCFR Inflows
Estuary Inflow
BOD Load

8. Steps in Applying a Mechanistic
Model
1. Decide on What to Model
2. Decide on Questions to be Answered
3. Choose Model
4. Collect Data for Inputs, Calibration
5. Create Input Files
6. Create Initial Test Application
7. Perform Qualitative “Reality Check”
Calibration & Debugging

9. Steps in Applying a Mechanistic
Model, continued
8. Perform quantitative calibration & model
verification
9. Design model scenario testing procedure
(endpoints, scenarios, etc.)
10. Perform scenario tests
11. Assess model reliability
12. Document results

10. Description of Model Application
Open Boundary
Elevation Cond.
Lower Cape Fear River
Estuary Schematic
Black River Flow
Boundary Cond.
Cape Fear R. Flow
Boundary Cond.
NE Cape Fear
Flow Boundary Cond.

11. Description of Model Application
• Flow boundary condition upstream (3 rivers)
• Elevation boundary condition downstream
• 20 lateral point sources (WWTPs)
• Extra lateral sources add water from tidal
creeks, marshes (14 additional sources)
• 37 total freshwater sources

12. Model State Variables
• Water Properties
– Temperature, salinities
• Circulation
– Velocities, water surface elevations
• Nutrients
– Organic and inorganic nitrogen, phosphorus, silica
• Organic Matter
– Organic carbon (labile particulate, labile and
refractory dissolved), phytoplankton (3 groups)
• Other
– Dissolved oxygen, total active metal, fecal coliform
bacteria

13. Water Quality Model Schematic

14. Data Collected to Support Model
• Data Collected from 8 sources
– US ACoE, NC DWQ, LCFRP, US NOAA, US
NWS, USGS, Wilmington wastewater authority,
International Paper
• Nearly 1 TB of original data collected
• File management system created to save and
protect original data

15. Observed Data Used to Create
Model Input Files
• Meteorological forcings (from NWS)
• Freshwater inflows (from USGS)
• Elevations at Estuary mouth (from
NOAA)
• Quality, temperature of freshwater
inflows and at estuary mouth (from
LCFRP, USGS, DWQ)
• Other discharges (from DWQ)

16. EFDC Input Files & Data Sources

17. Lower
Cape Fear
River
Program
Sites Used

18. USGS
Continuous
Monitoring
and DWQ
Special
Study
Stations
Used

19. New Cross-
Sections
Surveyed
by NC
DWQ

20. SOD
Monitoring
Stations
Performed
by NC
DWQ

21. LCFR Grid
• Channel
Cells in Blue
• Wetland
Cells in
White
• Marsh and
Swamp
Forest in
Green,
Purple

22. LCFR Grid Characteristics
• Grid based on NOAA bathymetry and
previous work by TetraTech
• Off-channel storage locations (wetland
cells) based on wetland delineations done
by NC DCM
• 1050 total horizontal cells (809 channel
cells, 241 wetland cells)
• 8 vertical layers for each horizontal cell
• Used a sensitivity analysis to locate and size
wetland cells

23. Model
Grid
Showing
Location
and Size
of
Wetland
Cells

24. Riverine
Swamps
and
Saltwater
Marshes in
Estuary
(NC DCM)

25. Input File Specification
• Inflows
• Temperatures and Water Quality
Concentrations at Boundaries
• Water quality mass loads for point sources
• Benthic fluxes
• Meteorological data

26. Riverine Inflow Specification
• Flows based on USGS flow data
• Flows scaled based upon drainage area ratios
• 17 total inflows
– 3 rivers, 14 estuary sources

27. Subwatersheds Draining Directly
to the Estuary

28. Subwatersheds
Draining
Directly to the
Estuary

29. Temperature and Concentration
Specification
• 5 stations used (3 boundaries, 2 in estuary)
• Combined USGS and LCFRP data
• Point source specification tied to closest
available data

30. Procedure for creating water quality
mass load file (WQPSL.INP)
• Used an automated procedure based upon
available data (LCFRP, DMR’s)
Use data interpolation and estimation to create a monitoring data set with no data
gaps, enter data into Excel spreadsheet, one spreadsheet for each source
For each source, create a data conversion matrix to estimate each
model constituent from the available parameters in the source data
For source data given as a concentration time history, multiply
concentrations by flows to get mass loads
Collect mass load time histories and reformat, then write into
WQPSL.INP file using Matlab script

31. An Example Conversion Matrix
(Cape Fear River Inflow)

32. Benthic fluxes and meteorological
data
• Used a prescriptive benthic flux model
• SODs time varying, but constant across
estuary
• SOD values based upon monitoring data
• Met data constant across estuary
• Met data taken from Wilmington airport

33. Model Calibration and
Confirmation
• 2004 calendar year used for model
calibration
• Nov 1, 2003 to Jan. 1 2004 used for model
startup
• 2005 calendar year used for confirmation
run (a.k.a. verification, validation run)

34. Streamflows during Model Runs
• 2004 dry
until October
• Early 2005
had some
high flows
• Summer
2005 was dry

35. Hydrodynamic Model Calibration
• Examined water surface elevations,
temperatures, salinities
• Used LCFRP and USGS data for
model/data comparisons of salinity
temperature
• Used USGS and NOAA data for model/data
comparisons of water surface elevation
• USGS data based on pressure measurements
not corrected for barometric changes

36. Monitoring
Stations Used
for
Hydrodynamic
Calibration

37. Simulation of Tidal Attenuation
in Estuary
• Varied wetland cell widths to determine
effect on attenuation of tidal amplitude
• Wider wetland cells gave more attenuation,
as expected
• Also tried different distribution of wetland
cells within estuary

38. M2 Tidal Amplitude for Various
Cell Width Scenarios

39. M2 Tidal Amplitude for Various
Cell Distribution Scenarios

40. M2 Tidal Amplitude for Various
Cell Distribution Scenarios
Width * 2, v1 chosen as best overall
(in green)

41. Example Time Series Comparison
– Black at Currie (upstream), 2004

42. Example Time Series Comparison
– NECF at Wilmington, 2004

43. Example Time Series Comparison
– Cape Fear at Marker 12, 2004

44. Example Time Series Comparison –
Black at Currie (upstream), Jan. 04

45. Example Time Series Comparison
– Wilm. Tide Gage, Jan. 04

46. Example Time Series Comparison
– Cape Fear at Marker 12, Jan. 04

47. Example Time Series Comparison –
Salinities at Navassa, 2004

48. Example Time Series Comparison
– Salinities at NECF Wilm., 2004

49. Example Time Series Comparison
– Salinities at Marker 12, 2004

50. Calibration Statistics, Salinity

51. Salinity Scatter Plot

52. Temperature Scatter Plot

53. Calibration Statistics, Temperature

54. Water Quality Calibration
• Added a second category of dissolved
organic matter (refractory C, N, P)
• Split between labile and refractory based
upon longer-term BOD measurements from
LCFRP, IP, Wilmington wastewater
authority
• Accounted for effects of NBOD in these
tests

55. Water Quality Model Schematic

56. Water Quality Model Schematic
State Variables Usually
Used to Simulate
Organic Matter Load

57. Water Quality Model Schematic
State Variables Usually
Used to Simulate
Organic Matter Load
Additional State
Variables Used
(settling velocity = 0.0)

58. Partitioning Organic Matter into
Labile and Refractory Parts
• Fit data to 2 component model for BOD
exertion, using equation

CBOD(t) rBODu(1 ekdrt
) rBODu
kdr
kdr  kdl
(ekdlt
 ekdrt
) lBODu(1 ekdlt
)

59. Example: Long-term BOD, IP
discharge, 7/20/2003

60. Partitioning Organic Matter into
Labile and Refractory Parts
• Fit data to 2 component model for BOD
exertion, using equation

CBOD(t) rBODu(1 ekdrt
) rBODu
kdr
kdr  kdl
(ekdlt
 ekdrt
) lBODu(1 ekdlt
)

61. DOC Load (kg)
Cape Fear
Black
NE Cape Fear
Estuary Inflows
Point Sources
Loading Breakdown for DOC

62. RPOC Load (kg)
Cape Fear
Black
NE Cape Fear
Estuary Inflows
Point Sources
Loading Breakdown for
Refractory DOC

63. NH4 Load (kg)
Cape Fear
Black
NE Cape Fear
Estuary Inflows
Point Sources
Loading Breakdown for NH4

64. Also implemented time variable
SOD (varies w/ temperature)
y = 0.0578x - 0.9637
R² = 0.341
-1.8
-1.6
-1.4
-1.2
-1
-0.8
-0.6
-0.4
-0.2
0
-6 -4 -2 0 2 4 6 8 10
ln(SOD)
(T-20)
ln(SOD)
Linear (ln(SOD))

65. Example Time Series Comparison –
DO at Navassa, 2004

66. Example Time Series Comparison
– DO at NECF Wilm., 2004

67. Example Time Series Comparison
– DO at Marker 12, 2004

68. Calibration Statistics, DO

69. DO Scatter Plot

70. DO Percentile Plot

71. Calibration of Other WQ
Constituents
• Show some key constituents
– Ammonia, nitrate+nitrite, total phosphorus,
chlorophyll-a
• Show only at Navassa (more plots in report)
• Overall, water quality model predicts each
of the constituents well

72. Example Time Series Comparison –
Ammonia at Navassa, 2004

73. Example Time Series Comparison –
NOx at Navassa, 2004

74. Example Time Series Comparison –
TP at Navassa, 2004

75. Example Time Series Comparison –
Chl-a at Navassa, 2004

76. Confirmation Run Results
• Ran model for calendar year 2005, with
parameters determined from calibration
• USGS continuous monitoring data ended by
then, used LCFRP data instead
• Show time histories only at Navassa (more
in report)

77. Example Time Series Comparison –
Salinities at Navassa, 2005

78. Example Time Series Comparison –
Temperatures at Navassa, 2005

79. Example Time Series Comparison –
DO at Navassa, 2005

80. Model Fit Statistics, DO, 2005
Confirmation Run

81. DO Percentile Plot, Predicted vs.
Observed, 2005 Confirmation Run

82. Sensitivity Testing
• Examined effect of varying SOD on model
DO predictions and sensitivity of system to
changes in organic matter loading
• SOD had an significant impact on model
predictions
• Effect of changing SOD on effect of load
changes shown in next section (scenario
testing)

83. Scenario Tests - Methods
• In general, test effect of changing
wastewater input on water quality of system
• Changed loads only for oxygen demanding
constituents (DOC, RDOC, Ammonia
• Examine DOs during warm weather period
(April 1 – November 1) at 18 stations
spread across impaired area
• Look at predicted DOs in each layer
• 6 scenario tests done so far

84. Six Scenario Tests Done so Far
1. Changes in Flow (and load) of Brunswick Co.
WWTP
2. Removal of load from all WWTPs, and from 3
(IP, Wilm NS & SS)
3. Removal of Ammonia load from all WWTPs
4. Increase all WWTPs to maximum permitted
load
5. Reduction in load from rivers, tidal creeks,
wetlands
6. Reduction in loads for various SOD values

85. 1. Changes in Flow (and load) of
Brunswick Co. WWTP
• Base case flow
= 0.38 MGD
• Three increased
flows
1. 4.3 times
base
2. 12.1 times
base
3. 39.1 times
base

86. 2. Removal of load from all WWTPs,
and from 3 (IP, Wilm NS & SS)
• Completely
removed
CBOD &
ammonia load
from all
WWPTS
• Tried turning
off just IP, just
Wilm NS & SS

87. 3. Removal of Ammonia load
from all WWTPs
• Removed
ammonia load
from all 20
WWTP inputs
• No changes to
CBOD load

88. 4. Increase all WWTPs to
maximum permitted load
• Increased all
flows and loads
to maximum
permitted
values
• Assumed
constant load at
maximum
permitted value

89. 5. Reduction in load from rivers,
tidal creeks, wetlands
• Manipulated
concentrations
(& loads) of all
17 freshwater
inputs (3 rivers,
14 estuary
sources)
• Reduced loads
by 30% and
50%

90. 6. Reduction in loads for various
SOD values
• Varied SOD
above and
below
calibrated value
• Observed effect
of turning all
WWTP loads
off for each
SOD case

91. Summary & Conclusions
• Successfully created a simulation model of
dissolved oxygen in Lower Cape Fear River
Estuary
• Model testing included calibration, confirmation,
and sensitivity analyses
• Scenario tests used to investigate system
sensitivity to changes in organic matter and
ammonia load
• System found to be only moderately sensitive to
changes in WWTP load

92. Additional Work Ongoing
• Working to finalize modeling report and
other publications
• Will work with DWQ personnel to
incorporate model results into TMDL
• Training DWQ personnel to run LCFR
model and analyze additional scenarios

93. Additional Work Ongoing
• Working to finalize modeling report and
other publications
• Will work with DWQ personnel to
incorporate model results into TMDL
• Training DWQ personnel to run LCFR
model and analyze additional scenarios
Questions?

94. Additional Work Ongoing
• Working to finalize modeling report and
other publications
• Will work with DWQ personnel to
incorporate model results into TMDL
• Training DWQ personnel to run LCFR
model and analyze additional scenarios
• Additional analyses done that are not in
report

95. Effect on DO of deepening
navigation channel
• Entrance
channel
deepened from
40 to 44 feet
• Remainder of
channel (up to
CF Mem. Br.)
deepened from
38 to 42 feet
0
0.2
0.4
0.6
0.8
1
3 3.5 4 4.5 5 5.5 6
Base Case
Dredged Channel
Cumulative
Frequency
Dissolved Oxygen (mg/L)
April through October Simulated Dissolved Oxygen Concentrations
in the Impaired Area, LowerCape Fear RiverEstuary

96. Effect of Changing River Load
and SOD
• Considers
possible
cleanup of
sediments
• SOD lowered
by same
percentages
(30% and 50%)
as riverine
loading 0
0.2
0.4
0.6
0.8
1
3 3.5 4 4.5 5 5.5 6 6.5 7
Base Case
30% Reduction River Load
30% Reduction River Load & SOD
50% Reduction River Load
50% Reduction River Load & SOD
Cumulative
Frequency
Dissolved Oxygen (mg/L)
April through October Simulated Dissolved Oxygen Concentrations
in the Impaired Area, LowerCape Fear RiverEstuary:
Clean Rivers Scenario

97. Analysis of DO deficit in the
impaired region
• Examined summer
average DOs
(surface) at 3 sites
in impaired region
• Used linear
sensitivity analysis
to attribute deficit
to either WWTPs,
SOD, or river loads 0
2
4
6
8
10
NECF at Wilm.
CF at HB
CF at Nav
WWTP deficit
River Load Deficit
SOD deficit
Avg. Conc.
Summer
Average
DO
(mg/L)
Location