HomeMy WebLinkAboutVer - More Info Received - 12/27/2012Final AA all os r%ro
UPDATE AND RECALIBRATION OF HARRIS LAKE
WATER QUALITY AND WATERSHED MODELS
M &N Project No. 7447 1 October 2012
Prepared for:
Progress Energy Carolinas, Inc.
410 S. Wilmington St., PEB 4A, Raleigh, NC 27601
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
TABLE OF CONTENTS
I.
Executive Summary .................................................................................
..............................1
11.
Introduction .............................................................................................
............................... 4
111.
GWLF Watershed Model .........................................................................
..............................7
IV.
CE- QUAL -W2 Input Revision and Development ....................................
.............................13
V.
CE- QUAL -W2 Model Calibration ............................................................
.............................29
VI.
Future Scenarios ....................................................................................
.............................63
VII.
Summary & Conclusions ........................................................................
.............................75
VIII.
References .............................................................................................
.............................83
1/11
motfatt & nichol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
TABLE OF FIGURES
Figure 1. Harris Lake Subwatersheds ............................................................. ..............................8
Figure 2. Existing Conditions Phosphorus Intensity ....................................... .............................11
Figure 3. Future Conditions Phosphorus Intensity ......................................... .............................12
Figure4. Branch Layout ................................................................................. .............................15
Figure 5. OASIS Output Showing Verification of Computed Daily Lake Elevations Compared to
Historic Daily Lake Elevations for Period Between 1999 - 2010 ............... .............................17
Figure 6. OASIS Output Showing Verification of Computed Monthly Lake Elevations Compared
to Historic Monthly Lake Elevations for Period Between 1987 - 1999 ...... .............................17
Figure 7. Approximate CE- QUAL -W2 Boundaries ......................................... .............................19
Figure 8. Water Quality Monitoring Stations in Harris Lake ......................... ............................... 31
Figure 9. Comparison of Computed and Observed Pool Elevations for Final Water Balance.... 32
Figure 10. Forebay Formed by Road crossing the Upper Portion of Branch 3 ...........................37
Figure 11. Harris Lake Forebay Aquatic Plants ........................................... ............................... 37
Figure 12. Forebays formed by Road Crossings in the Upper Portion of Branch 2 ....................38
Figure 13. Forebay Formed by Auxiliary Cooling Water Reservoir ................ .............................39
Figure 14. Computed and Observed Temperature Profiles for Station E2 .... .............................42
Figure 15. Computed and Observed DO Concentration Profiles at Station E2 ..........................47
Figure 16. Computed and Observed Time Series Concentrations of Nitrate at the Water Surface
forFour Stations ................................................................................... ............................... 52
Figure 17. Computed and Observed Time Series Concentrations of Ammonium at the Water
Surface for Four Stations ........................................................................ .............................53
Figure 18. Computed and Observed Time Series Concentrations of Ammonium at the Lake
Bottom for Four Stations ......................................................................... .............................54
Figure 19. Computed and Observed Time Series Concentrations of TKN at the Water Surface
forFour Stations ................................................................................... ............................... 55
Figure 20. Computed and Observed Time Series Concentrations of TN at the Water Surface for
FourStations .......................................................................................... .............................56
Figure 21. Computed and Observed Time Series Concentrations of TP at the Water Surface for
FourStations .......................................................................................... .............................57
ii 1/1S
motfatt & nichol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
Figure 22.Computed and Observed Time Series Concentrations of TOC at the Water Surface
forFour Stations ................................................................................... ...............................
58
Figure 23.
Computed and Observed Time Series Concentrations of Chl a at the Water Surface
forFour Stations ................................................................................... ...............................
59
Figure 24.
Time Series of Computed Phytoplankton Biomass on the Surface at the dam for the
ThreeAlgal Groups ................................................................................. .............................60
Figure 25.
Comparisons for Surface TN at the Dam ...................................... .............................65
Figure 26.
Comparisons for Surface TN at Segment 5 .................................. .............................65
Figure 27.
Comparisons for Surface TP at the Dam ...................................... .............................66
Figure 28.
Comparisons for Surface TP at Segment 5 .................................. .............................66
Figure 29.
Comparisons for Surface Chl a at the Dam .................................. .............................67
Figure 30.
Comparisons for Surface Chl a at Segment 5 .............................. .............................67
Figure 31.
Comparisons for Surface DO at the Dam ..................................... .............................68
Figure 32.
Comparisons for Surface DO at Segment 5 ................................. .............................69
Figure 33.
Comparisons for Bottom DO at the Dam ...................................... .............................69
Figure 34.
Comparisons for Bottom DO at Segment 5 .................................. .............................70
Figure 35.
Comparison of Computed Water Age at the Surface and at the Dam for Existing and
FutureConditions ................................................................................... .............................73
Figure 36.
Comparisons for Surface Chl a at the Dam ................................ ...............................
78
Figure 37.
Comparisons for Surface Chl a at Segment 5 .............................. .............................79
Figure 38.
Comparisons for Surface DO at Segment 5 ................................. .............................80
Figure 39.
Comparisons for Bottom DO at the Dam .................................... ...............................
80
1/11
motfatt & nichol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
REFERENCED TABLES
Table 1.
Existing and Future Land Use by Type in GWLF .............................. ..............................9
Table 2.
Comparison of TN and TP for Existing and Future Conditions by Year ........................10
Table 3.
Mean of Best Fitted Parameters Used to Compute Stream Temperatures ..................26
Table 4.
Modeled Water Quality State Variables ........................................... .............................29
Table 5.
Results of Web Soil Survey Comparative Soils Analysis ................. .............................36
Table 6.
Existing Conditions Average Annual Mass Loads ........................... .............................71
Table 7.
Future Regime A Average Annual Mass Loads ............................... .............................71
Table 8.
Future Regime B Average Annual Mass Loads ............................... .............................71
Table 9.
Hydrologic Factors Affecting Lake Residence Time ...................... ...............................
74
Table 10. Comparison of TN and TP for Existing and Future Conditions by Year ......................76
iv 1/11
motfatt & nichol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
Executive Summary
To support issuance of the 401 Water Quality Certification for the expansion at the Shearon
Harris Site, the CE- QUAL -W2 water quality model previously developed for Harris Lake was
updated and recalibrated for use in predicting future water quality conditions. The model was
used to simulate the combined effects of a higher pool, larger heat load for power plant cooling,
release targets, and introduction of makeup water from the Cape Fear River. The W2 model of
Harris Lake was supported by a Generalized Watershed Loading Function (GWLF) model that
predicted pollutant loads from stormwater runoff and other non -point sources throughout the
Lake's watershed. The GWLF model was also updated to reflect the most current land uses
within the urbanizing watershed.
The results of the GWLF show that the highest predicted loads occur during wet years, and the
lowest predicted loading occurs during dry years, as would be expected. It should be noted that,
while results vary from year to year, on average the GWLF model predicted no increase in non-
point source phosphorous loading for the future land use scenario.
Many updates and improvements were made to the CE- QUAL -W2 model inputs, including:
numerical grid structure; tributary discharge rates; precipitation; dam outflow discharge rates;
tributary inflow temperatures; point source loadings; non -point source loadings; cooling water
return flow temperatures and water quality; meteorology; and various calibration parameters.
Additionally, a newer version of the model code was used (version 3.6) rather than using the
older version (version 3.2) that was used in the previous modeling of Harris Lake. The newer
version allowed better representation of organic forms of nutrients.
Pool elevations computed by the revised model for the period 2001 — 2008 were compared to
observed pool elevations to verify that the model was accurately representing the correct water
balance. The revised model was then recalibrated for the years 2006 through 2008 for
temperature, various forms of nitrogen, total phosphorus, dissolved oxygen, algal chlorophyll a,
and total organic carbon. The results of the recalibration were subjected to an independent
peer review performed by HDRIHydroQual. The recommendations of the peer review were
used to further refine the model recalibration.
Three future scenarios were developed for evaluation with the recalibrated W2 model, and all
three scenarios were run with the same hydrologic base conditions. The three future scenarios
were:
• Existing Lake — Future Loads: This scenario examined the predicted water quality
conditions for the existing Lake, with current reservoir pool elevation and storage volume
in conjunction with future nonpoint source load conditions and projected future discharge
conditions at the Holly Springs WWTP.
1 1/11
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
• Future Regime A — This scenario examines all conditions associated with the expansion
of the HAR in conjunction with the expanded Lake, future nonpoint source loads, and
projected future conditions at the Holly Springs VWVfP. The operational scheme for
expanded HAR includes the addition of make -up water with no pumping when flows in
the Cape Fear River are less than 600 cfs.
• Future Regime B — This scenario examines all conditions associated with the expansion
of the HAR just as Future Regime A, but with the operational scheme for expanded HAR
including make -up water with no pumping when flows in the Cape Fear are less than
700 cfs.
The model predicts increases in total nitrogen (TN) and total phosphorus (TP) concentrations for
all future scenarios relative to existing conditions. The model shows there will be little difference
between Future Scenarios A and B with regard to in -lake TN and TP concentrations. Increases
in chlorophyll -a are predicted for all future scenarios relative to existing conditions. Peak values
spike to levels slightly greater than 23 µg /I near the end of the simulation period at both
locations. However, predicted chlorophyll -a concentrations are less than 20 µg /I most of the
time.
The model shows that the various scenarios cause very little change relative to each other in
the seasonal fluctuation of dissolved oxygen (DO) levels at the water surface. Model results for
bottom DO show that future scenarios result in a very slight increase the predicted duration of
anoxic periods during warm seasons, and that such periods of anoxia are more pronounced at
the dam than in segment 5. These results are not unexpected given the increased nutrient
concentrations and algal productivity predicted in conjunction with future conditions. The higher
incidence of bottom anoxia at the dam is also partly a function of greater water depth in that
vicinity for future lake conditions.
Analysis of lake water residence time was conducted using water age, a model- computed
parameter that serves as a surrogate for residence time. The water age for existing conditions
fluctuates between 200 -500 days for most of the 8 -year simulation, and between 400 -700 days
for the future conditions. Close examination of the timing of increases in water age in
conjunction with the timing of increases in nutrient levels within the lake and the timing of
increases in algal growth via chlorophyll -a indicate that periods of increased nutrient loads and
eutrophication correlate with water age.
Conclusions
The updated water quality modelling analysis described in this report represents several
improvements over previous analyses applied to Harris Lake, including:
• Revised and refined model bathymetry
2 1/11
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
• An updated and improved water balance
• Revised tributary nonpoint source nutrient loads
• Revised and improved meteorological data files
• Refined physical descriptions for the dam spillway
• Refined physical descriptions for the cooling water intake and discharge structures
• Simulation of real -world cooling water withdrawal and blow -down discharge rates
These improvements result in an improved simulation of the dynamics affecting water quality
conditions in Harris Lake, and this model provides a credible representation of the Harris Lake
system water quality.
The model indicates that for drier periods the future lake (addition of the two reactors and
associated lake changes and pumped water) will have higher nutrient and algal concentrations
compared to those associated with the existing lake with future loadings; but for wetter periods,
there will be little difference in -lake concentrations for the future lake compared to the existing
lake with future loads. The critical periods for increases in nutrient and algal concentrations (for
existing lake and future lake conditions) are during drought conditions that extend the lake water
residence time. Pumping water from the Cape Fear River is predicted to increase nutrient and
algal concentrations in the lake, but the amount of increase for any year will be dependent on
the local rainfall and hydrology. The potential water quality impacts, in terms of the increases in
algal productivity that could be associated with these increases, are at least partly mitigated by
the increased volume of the lake, which is nearly tripled.
The last year of the simulation (2008) represent one of the more stressful years for water quality
due to increased tributary inflows and pollutant loads after a drought year, increased pumping
from the Cape Fear River, and the Holly Springs wastewater discharge running at maximum
permitted levels. In spite of these more stressful conditions, chlorophyll a concentrations are
predicted to remain below 25 µg /I throughout the year.
Taken collectively, the model predictions indicate that the Shearon Harris Nuclear Power Plant
can be expanded and operated in the manner simulated herein without resulting in violations of
water quality standards in Harris Lake.
3 1/11
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
Intentional Blank Page
1/11
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
II. Introduction
1. Background Information
The Harris Advanced Reactor (HAR) project consists of adding two new nuclear power
generation units at the existing Shearon Harris Site. The net consumptive water usage for the
two new reactors will be approximately 42 million gallons per day (MGD). Having a sufficiently
sustained yield from Harris Lake to support the consumptive use of cooling water and meet
minimum flow requirements below Harris Lake Dam will require an expansion of Harris Lake
from the current pool elevation of 220 feet to 240 feet, as well as pumping of up to 86 MGD in
makeup water from the Cape Fear River. Progress Energy Carolinas (PEC) plans to construct
an intake structure and pump the makeup water from the Cape Fear River.
Basin -Wide Water Quality Assessment and Lake Assessment documents from North Carolina
Department of Water Quality (NCDWQ) have consistently noted that water quality is typically
good in Harris Lake, relative to other Piedmont reservoirs, despite the presence of some
nuisance aquatic macrophytes. The lake has typically high secchi depth readings and low
concentrations for nutrient parameters. Conversely, prior to reaching Buckhorn Dam, the Haw
and Deep Rivers that form the Cape Fear River flow through numerous cities and towns in the
state's most developed region, receiving non -point runoff and numerous large wastewater
discharges along the way. As a result, Cape Fear River water has higher concentrations of
nutrients and other pollutants relative to Harris Lake and its tributaries. The higher nutrient loads
have the potential to impact water quality in Harris Lake when substantial volumes of Cape Fear
River water are introduced to the Lake.
2. Purpose
The purpose of this water quality modeling of Harris Lake was to provide the information
necessary to demonstrate that the project will not have a significant impact on water quality and
support issuance of the 401 Water Quality Certification for two additional nuclear power
generation units at the Shearon Harris Site. This modeling analysis examined potential water
quality impacts to Harris Lake under various lake management and make -up water pumping
scenarios. The model simulations focused on conventional /eutrophication water quality
parameters including temperature, nutrients, phytoplankton, dissolved oxygen, and pH.
3. Modeling Approach
A two - dimensional, laterally- averaged, hydrodynamic and water quality model, CE- QUAL -W2
was applied to Harris Lake to forecast the water quality impacts of importing makeup water into
the lake from the Cape Fear River. In this particular application, a CE- QUAL -W2 model (W2
model) previously developed for simulation of Harris Lake was updated and recalibrated for use
in this study. The model was used to predict future water quality conditions in the lake resulting
5 1/11
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
from the combined effects of a higher pool, larger cooling load, release targets, and introduction
of makeup water from the Cape Fear River.
The W2 model of Harris Lake was supported by a Generalized Watershed Loading Function
(GWLF) model that predicted pollutant loads from stormwater runoff and other non -point
sources throughout the Lake's watershed. The watershed includes portions of the rapidly
growing and changing jurisdictions of southwestern Wake County, and since the GWLF model
was originally developed, local land use plans within these jurisdictions, as well as those for
Chatham and Harnett Counties have evolved. To reflect the most current land use planning
information available, the GWLF model was updated accordingly.
The methods for the update and application of both models, along with the results, are
described herein.
6 1/11
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
III. GWLF Watershed Model
An initial GWLF model of Utley Creek was developed by Tetra Tech in 2004 and was revised by
CH2M HILL to include Harris Lake. Soil properties were adjusted and land use (future and
existing) was taken into account in the revised model. Meteorological data from the Progress
Energy meteorology station was used. Further details can be found in the CH2M HILL report
(CH2M HILL, 2009).
Moffatt & Nichol updated the CH2M HILL model's soil properties, land use distributions, and
meteorological data in revising the model once again. The model subwatersheds delineated and
applied previously were also used in this watershed modeling analysis. The primary modification
of the GWLF model consisted of updating the land use distributions to reflect changes in the
watershed since the CH2M HILL modification.
1. Modeling Approach
The GWLF model uses daily precipitation, temperature, land use, and soil properties to
calculate monthly time series of watershed runoff, sediment, and nutrient loads from a given
drainage area. GWLF uses the SCS Curve Number method to calculate runoff volumes based
on rainfall amount. Pollutant loading rates are tied to build -up rates and wash -off for urban land
uses and runoff concentrations for rural land uses. The GWLF model was developed using the
values adapted from the larger Jordan Lake watershed assessment completed by Tetra Tech
(2003). In addition to sediment and nutrient loadings, groundwater recession coefficients and
evaporation rates were carried over from the Jordan Lake study. The nutrient build -up rates and
runoff concentrations utilized by Tetra Tech in the Jordan Lake study were derived from
numerous studies conducted within the immediate physiographic region including Line et al.,
2002; CH2M HILL, 2000; Greensboro, 2003; and U.S. EPA, 1983.
Parcels in the basin were identified using the most recent county parcel and zoning GIS layers
from Wake, Chatham, and Harnett Counties. Wake County GIS layers were utilized to identify
land use, number of buildings and type of buildings for each parcel. The parcels were sorted
based on current land use, using the presence of buildings as a starting point for determining
land use. The parcels in Harnett and Chatham counties were observed via 2010 aerial
photography and then classified appropriately. The aggregate land area attributed to each use
category (Forestry, Pasture, Row Crop, Barren, Urban Green Space, Wetlands, Water,
Residential — very high to very low, Office /Light Industrial, Commercial /Heavy Industrial) was
assigned an appropriate curve number and pollutant loading coefficient in GWLF. All areas not
included as tax parcels were considered to be roadway /right -of -way and were assigned the
curve number for a land use category with similar impervious area percentage (Office /Light
Industrial). Each parcel was grouped geographically into one of 17 GWLF sub - watersheds for
modeling purposes, as seen below in Figure 1.
7 1/11
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
Legend
GWLF SLII) atel'shetls I�
Waterbodies
�IY
Harris Lake Watershed
0 0.75 1.5 3
Mmes
Ch ath am
County
s 1 �S
Sub8
Wa ke
County
Sub1
S gVSub10
Sub41+1� Sub13
Sub12
Sub16
` Sula15 - 1 Subs
—.S W 7
;r, — 5ub6
Chatham
County
a
Harnett
County
7
I a+ 0
1 IL
a'
4
v� a ++
{ .I
SubO
`Sub14
Sub3�
ub4
Figure 1. Harris Lake Subwatersheds
8
,/,1
moffatt & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
Future land use estimates were developed using the methodologies described above, with the
exception that the parcels were assigned the maximum build -out condition based on current
zoning assignments. For example, if a parcel in the existing model was assigned a Heavy
Industrial land use, but the site was not developed, it would have been assigned a Forestry or
Barren land use, with a representative curve number for that use. For the future condition, the
same parcel would be assigned to the Heavy Industrial category and the curve number would
change accordingly. The curve numbers used to define each land use category were
maintained for the future conditions model. Again, roadway /right -of -way area was assigned the
curve number for the Office /Light Industrial land use category. The results of the land use
classifications for existing and future conditions are found in Table 1. As would be expected in
an area impacted by suburban residential growth, the future land use shows decreased portions
of barren, forested, and agricultural land with commensurate increases in all categories of
residential land.
Table 1. Existing and Future Land Use by Type in GWLF
Monthly time series results from the GWLF model were used as a starting point for the W2
modeling.
9 1/11
moffati & mchol
Existing Conditions
Future Conditions
Land Use Type
Hectares
Watershed
Hectares
Watershed
Forestry
7535.7
41.0%
4399.2
24.2%
Pasture
418.0
2.3%
507.2
2.8%
Row Crop
2066.0
11.2%
52.5
0.3%
Barren
2582.8
14.0%
473.2
2.6%
Urban Green Space
91.9
0.5%
56.1
0.3%
Wetlands
318.7
1.7%
318.7
1.8%
Water
1662.6
9.0%
2874.2
15.8%
Very Low Density Residential
357.7
1.9%
1840.6
10.1%
Low Density Residential
1053.5
5.7%
2179.0
12.0%
Medium Low Density Residential
626.1
3.4%
2001.2
11.0%
Medium Density Residential
185.6
1.0%
973.6
5.3%
Medium High Density Residential
85.4
0.5%
167.1
0.9%
High Density Residential
58.3
0.3%
166.6
0.9%
Office /Light Industrial
0.5
0.0%
0.5
0.0%
Commercial /Heavy Industrial
964.4
5.2%
1175.8
6.5%
Monthly time series results from the GWLF model were used as a starting point for the W2
modeling.
9 1/11
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
2. GWLF Results
The results of the GWLF are summarized below in Table 2 and illustrated spatially in Figure 2
and Figure 3. The model results in Table 2 indicate that the highest predicted loads occur in the
year 2006 for TN and TP, and the lowest predicted loading occurs in year 2005 for both TN and
TP. Given the patterns of precipitation in the watershed, these results are not unexpected.
Comparing the results in Figure 2 and Figure 3 illustrate that predicted total phosphorous (TP)
loading rates remain steady or increase in 12 of 17 subwatersheds in conjunction with future
land use conditions relative to existing conditions. The increase occurs largely from conversion
of barren and forest land to suburban land uses in pockets throughout the watershed. Five
subwatersheds actually experience decreases in predicted loading rates when moving from
existing to future conditions. However, it is not unusual for GWLF to show decreases in overall
loads in subwatersheds where significant amounts of row crop and /or pasture land are
converted to medium to low density residential uses, which have lower nutrient loads associated
with them.
Monthly nutrient loads were output from GWLF and redistributed across daily flow records for
input into the CE- QUAL -W2
Table 2. Comparison of TN and TP for Existing and Future Conditions by Year
10 „„
moffati & mchol
TN (lb
/ac)
TP (lb /ac)
Year
Precipitation (in)
Existing
Future
Existing
Future
2001
31.82
3.31
4.22
0.56
0.61
2002
42.61
5.62
7.40
0.88
1.05
2003
46.11
5.61
6.16
0.94
0.88
2004
36.32
3.53
3.62
0.64
0.53
2005
28.34
2.31
2.68
0.38
0.38
2006
47.45
6.37
7.02
1.08
1.01
2007
30.98
3.69
4.69
0.58
0.66
2008
40.98
4.58
4.78
0.78
0.68
Average
38.08
4.38
5.07
0.73
0.73
10 „„
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
Legend - Existing
Phosphorus Intensity
0.00 - 0.10 Iblaclyr
0.11 - 0.25 Iblaclyr
0.26 - 0.50 Iblaclyr
0.51 - 0.75 Iblaclyr
0.76 - 1.00 Iblaclyr
1.00 Iblaclyr
0 0.75 1.5
CY rY Ti
calk
Su b2
N
i
i
3
1 Miles
5Al 6
5ub15
I
i
5
Su
5u h7
Chatham
County
SUM
Su b5
Harnett
County
5u b0
f
Subl4
Figure 2. Existing Conditions Phosphorus Intensity
11
moffatt & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
Figure 3. Future Conditions Phosphorus Intensity
12 1111
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
IV. CE- QUAL -W2 Input Revision and Development
The general inputs required by the CE- QUAL -W2 model (W2 model) are similar to those of most
time - varying simulation models and include: domain computational grid development and
associated geometry and bathymetry; boundary conditions, such as lake inflows and outflows;
initial conditions, such as initial lake concentrations; and model control parameters including
process parameters, such as rates, stoichiometric fractions, coefficients, etc. Harris Lake input
development is discussed in this section for the first two types of inputs noted above. The latter
two types of input (initial conditions and parameters) are discussed in the model calibration
chapter since these types of inputs predominantly pertain to model calibration.
The W2 model requires three separate time series input files for water -borne boundary
conditions, i.e., material loadings entering the lake. These three input files pertain to water flow
rates, constituent concentrations, and temperatures. Flow rates are expressed in units of cubic
meters per second (m3 /sec, or cros), concentrations in mg /L or g /m3, and temperatures in
degrees centigrade ( °C). These files can be input for lake- branch inflows (branch headwater
segments), tributaries that enter segments through the side of the segment (laterally),
precipitation, and distributed tributaries, which are tributaries that are spread among all the
segments within a branch. Branch inflows generally describe non -point sources that enter via
streams feeding into the branch headwater, such as rivers and creeks. Point sources are
described by the tributary inflow option. However, non -point source flows can also enter a
segment laterally as a tributary input, which is the case for one tributary entering Branch 1.
Water -borne input development methodologies that differ from those used in the
existing /original model (henceforth referred to as the CH2M model) are described in detail in the
subsequent sections below.
1. Flow Inputs
Flows for OASIS and CE- QUAL -W2
The revised W2 model utilizes the same branch and tributary subdivision of the lake bathymetry
as was used in the CH2M model. The lake is divided into five (5) branches and one (1) tributary
for a total of six (6) non -point sources. The tributary and each of the branches is assigned an
aggregate land area based on a particular grouping of a number of the 17 sub - watersheds
defined in the GWLF model. See Figure 4 for branch and tributary spatial orientation. The
groupings do not differ from those defined in the CH2M model. However, the total land area for
each branch and tributary was revised based on the updated sub - watershed calculations
described in Section III. The new branch and tributary land areas were used to subdivide inflow
quantity results from an OASIS hydrology model provided by Hydrologics, Inc. on an area basis.
The subdivided inflows are the direct inputs to the six nonpoint source flow time series used in
the W2 model. The CH2M model utilized a simple area - weighted, flow gauging method for
estimating nonpoint source flow inputs. That model utilized daily flow records from a USGS
13 1/11
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
gage in White Oak Creek (approximately 15 mi. north of Harris Lake) to scale flows based on
drainage area ratios to the aforementioned gage. The inflows developed for the OASIS model
discussed below have become the new basis for the determination of nonpoint source flow
inputs to the lake system.
14 1111
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
N
Legend
CE- DUAL -VV2 BOUndaries f�
1Naterbodies 4 .
0 0.75 1.5 3
Miles
Cknm
C RRIv
W,Mk.e
�S,h2 Branch`3r
ti Branch
M
,,915u hs � w� � �sh- �
in Dam e
eianM O
wetly D
H�nett
*Note that CE-0 UAL -VV2 Boundaries are approximate
Figure 4. Branch Layout
15
,/,1
moffatt & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
An update to the total inflow regime for Harris Lake was necessary to improve accuracy when
calibrating to known data and especially for evaluating varying future flow regimes in the
expanded reservoir. The original dataset used back - calculated inflows based from a
downstream gage at Buckhorn Creek near Corinth. A gage in Lillington was utilized and
subsequently adjusted for drainage area for the period before 1972 when the gage flows were
not available for Buckhorn Creek.
For the OASIS inflow update, Hydro Logics, Inc. used the Middle Creek gage in the adjacent
Neuse River Basin, which has a similar drainage area and correlates very well with the back -
calculated inflows to the lake. The Middle Creek gage was chosen as nearby gages in the Cape
Fear River basin do not correlate as well to the back - calculated flows. It is noted that
approximately 5% of the monthly back - calculated data were negative, (hydrologically and
physically impossible).
In particular, the rating on the Buckhorn Creek gage was "good" through the early 1990s, but
deteriorated after this, with "poor" ratings in the mid -1990s and post -2004 due to beaver dam
impacts. In addition, the accuracy of the net withdrawal data from Harris may have been less
accurate when only monthly data were available (1987 to 1999), compared to when daily data
sets were available (2000 to present).
The data sets with negative inflows values were adjusted to remove the negatives while still
preserving the volume over the adjustment period. This methodology ensured that the average
flow over that period was preserved. By eliminating the negative inflow values, a more accurate
correlation was made with the Middle Creek gage. The Middle Creek flows had first been
unimpaired prior to the Town of Cary, NC wastewater return upstream which has been in place
for the last 20 years. The impact of this wastewater return resulted in occasional negative
unimpaired flows, which were adjusted in a manner similar to the aforementioned negative flow
values.
The correlations between the inflow data sets were improved by excluding the back - calculated
data and using only the Buckhorn Creek gage data prior to the construction of Harris Lake in
1981. The resulting correlations were therefore based on overlapping data from Buckhorn Creek
(adjusted for the drainage area at the head of Harris Lake) and Middle Creek for the 1972 to
1980 period. The FILLIN program was used to compute these correlations and to "fill in" the
missing record for the Harris Lake inflows (from 1930 to 1971 and 1981 to present).
The computed inflows were passed through an OASIS hydrologic model. They were run to
match the historical discharge and net withdrawal from Harris Lake compared to the computed
elevation with the recorded historic lake elevations for the 220' existing operation level. Figure 5
and Figure 6 illustrate the results. Further details on the development of the OASIS model can
be found in Modeling the Proposed Operations of Harris Lake Using OASIS (HydroLogics,
2012).
16 1111
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
Figure 5
OASIS Output Showing Verification of Computed Daily Lake Elevations Compared to
Historic Daily Lake Elevations for Period Between 1999 -2010
Figure 6. OASIS Output Showing Verification of Computed Monthly Lake Elevations Compared to
Historic Monthly Lake Elevations for Period Between 1987 -1999
17 „„
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
Outflow at the Dam
The outflows resulting from the spill over the Harris Lake main dam during high flow periods are
not quantified with an input time series, rather they are governed by a spillway discharge rating
curve within the W2 model. When the lake reaches a stage above 220' in the existing scenario,
and above 240' in any future scenario, the dam will spill at a rate controlled by a weir rating
equation utilized in the W2 model. It is necessary to develop the coefficients for the weir
equation based on the rating curve information for the existing and future spillway provided by
PEC. The spillway rating curves for existing and future conditions are shown in the following
equations, respectively:
Q = 27.807 ( pool — 67.056) 1.463
( 1 )
Q = 62.845 ( pool — 73.152.363 (2 )
Where pool is the pool elevation in meters and Q is the discharge in cros.
The intake flows pumped from the Cape Fear River and the flows released from the low base
flow valve in the main dam are dictated by requirements for parameters set forth in various
future scenarios. These inputs are variable based on any number of future water management
scenarios deemed critical for evaluation. These inputs are compiled as daily time series files for
every future water pumping, release and spill management scenario. Further discussion of
outflows can be found in Section IV -1.
Point Source Flow Inputs
The Harris Lake watershed has four point source flow inputs that had to be specified in the W2
model for existing conditions and /or future scenario runs. The point source inputs for the
existing conditions are shown in Figure 7 and are as follows:
• A municipal wastewater effluent stream from the Town of Holly Springs WWTP
• A small, domestic wastewater effluent stream from the Harris Energy & Environmental
Center
• Blow -down discharge from Harris Nuclear Plant (HNP) Cooling Tower
• Tributary 1, which enters branch 1. However, this input is a non -point source runoff flow,
rather than a point source discharge.
All point source discharges and the Tributary 1 inflow were specified as tributary inputs, rather
than branch inputs, in the W2 model. In addition to the inputs listed above, the future scenarios
utilize the following additional point- source flow inputs:
• Inflow makeup water pumped from the Cape Fear River
• Blow -down discharge from the two (2) proposed units that comprise the Harris Advanced
Reactor project (HAR -2 & HAR -3) aggregated with the existing HNP reactor.
18 1111
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
Chatham Legend N
County Q CE- QUAL -W2 Boundaries
Waterbodies
Wake Harris Lake Watershed
County c
0 2,500 5,000 10,000
Feet
f
16 Enters
earon Ha is
clear Plan 20
17.
26 'CF5 ling1Wateril
.'�ri
22
!�qjw�l
Chatham
County
that CE- QUAL -W2 Bo daries are
21
31
Figure 7. Approximate CE- QUAL -W2 Boundaries
19
Harnett
unty
,111
moffatt & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
The Town of Holly Springs WWTP flow time series was generated based on actual NCDWQ
Discharge Monitoring Records (DMRs) provided directly from municipal officials. Flow rates
were taken from daily reported values in million gallons per day (MGD) and converted into a
monthly time series in cubic meters per second (cros). Monthly time series were used because
the nutrient concentration input data associated with the respective discharge source could only
be estimated at a monthly interval. Further discussion as to the reporting of concentration input
data can be found later in the point source nutrient concentration Section IV -2. Flow rates for
future operation of the Holly Springs WWTP were based on existing DMR and service area
population data and then extrapolated to a constant value of 6.12 MGD based on the estimated
service area population of the WWTP in year 2025.
The flow time series for the domestic wastewater stream from the Harris Energy &
Environmental Center was generated using an assumed constant value of 0.00013 cros, which
is equivalent to approximately 3000 gallons per day (GPD). This value was taken from the
CH2M model input file and was deemed valid for use in the revised model, as it is
representative of a typical domestic wastewater flow rate from a building of that size and use. It
will be noted that the flow rate for this point source is insignificant compared to the total system
inflow. The future flow for this source was assumed not to increase, and thus remains identical
to the existing input flow series file.
The blowdown water withdrawal and discharge rates for the existing Harris Nuclear Plant, as
well as future units HAR -2 & HAR -3, will be discussed together, as the inputs are connected
through a mass and flow balance affected by the reactor cooling system requirements. The net
difference between the intake and the blowdown flows represents the evaporative loss
experienced by the lake system due to the almost constant operation of the reactor and cooling
tower units. Moffatt and Nichol was provided with a database of daily average flow rates from
PEC detailing the reactor /cooling tower flow balance. The data from PEC provided the most
accurate available estimation of daily intake and discharge rates. A revision to the blowdown
flow inputs provided the mechanism to model the mixed discharge system utilized at the HNP,
by which cooler lake water (dilution water) is mixed with heated reactor water prior to returning
to the lake system. The inclusion of a dilution stream into the reactor water flow balance
resulted in the intake and blow -down discharge flow rates being much higher, on average, than
the constant flow rates utilized in the CH2M model. The inclusion of the time series (updated to
a daily interval) of flows in the revised model resulted in improvements to flow, temperature, and
constituent mass balances.
The estimation of the intake and blowdown discharge rates for the future operation scenarios
are governed by the data and assumptions used in developing the existing HNP flow balance
and are applied in conjunction with estimates for the operational flow requirements of the two
identical proposed reactors, HAR -2 & HAR -3. This data, provided in the Harris Advanced
Reactor Combined Operating License (COL) Application Environmental Report, was used to
20 1/11
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
develop daily time series inputs for the cooling water intake flows, discharge flows, and
evaporative losses of the proposed reactor system.
2. Nutrient/Water Quality Concentration Inputs
Nonpoint Source Concentration Inputs
Each of the six non -point source inflows requires a concentration input file for W2. The product
of flow rate and concentration constitutes mass loading (mass /time) entering the lake. These
files serve to translate nutrient mass loading values from the GWLF model outputs into
meaningful loading boundary conditions for the W2 model. Nutrient loadings from GWLF are
output at a monthly interval, and include both the dissolved and particulate fractions of the total
nitrogen and phosphorous mass loads. The GWLF total mass loads are then divided by the total
OASIS monthly flows, and converted into concentrations. As the W2 model does not generate
results directly from these bulk nutrient inputs, a number of stoichiometric calculations and
assumptions regarding nutrient fractioning were required to generate the inputs into sub - species
required by the W2 model. In addition to key nutrients, there are other physical and chemical
parameters that are required to complete a functional water quality input file. The methodologies
for processing the existing and future non -point source concentration input files are identical. All
of the concentration file inputs, including sources and /or assumptions for each are detailed
below.
TDS (Total Dissolved Solids)
A constant TDS concentration of 35 mg /L was developed in an iterative fashion during the initial
calibration to match observed TDS data in the lake from 2006 -2008.
ISS (Inorganic Suspended Solids)
A constant ISS concentration of 35 mg /L was assumed. There were no observed data for ISS.
Nitrate (NO3) and Ammonia (NH4)
The nitrogenous sub - species fractioning was based on the Ambient Water Quality Criteria
Recommendations: Eco- region IX (EPA. 2000). The total nitrogen mass load was subdivided as
5% NH4, 20% NO3 and 75% Total Organic Nitrogen (TON) for all non -point sources.
Phosphate (PO4)
The phosphorous sub - species was fractioned using the (EPA, 2000) fact sheet. The total
phosphorous mass load was subdivided as 20% PO4 and 80% Total Organic Phosphate (TOP).
21 1111
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
Dissolved Silica (DSI) and Particulate (PSI)
Assumptions remaining from the CH2M inputs were deemed satisfactory and remained in the
revised model. DSI remains at a 10:1 ratio to PSI as in the CH2M model.
Iron FE
An average of recorded historical values from the DWQ monitoring station at NC Highway 42
near Corinth, NC were used to verify the assumption for a constant input value of 1.0 mg /L for
FE.
Labile /Refractory Dissolved Organic Matter (LDOM, RDOM), Labile /Refractory Particulate
Organic Matter (LPOM, RPOM)
Inflowing organic matter was used for representing organic carbon for version 3.6 of the W2
model since organic N and P were accounted for directly as state variables with loading inputs.
As there are no observed inputs of organic carbon and since GWLF does not estimate this
variable, iterative model runs were conducted during model calibration to estimate the required
inputs for organic matter (and, thus organic carbon). Model- computed total organic carbon
(TOC) was compared with observed TOC values in the lake during 2006 — 2008. Adjustments
were made to the organic matter inputs until the model TOC results matched observed TOC
relatively well. TOC (and, thus total organic matter) was assumed to be split equally between
dissolved and particulate. Furthermore, organic matter (and organic carbon) was assumed to be
split as 20% labile and 80% refractory for dissolved and particulate forms. The 20 — 80 % split is
based on experience on modelling nonpoint source loadings into various water bodies.
Carbonaceous Biochemical Oxygen Demand (CBOD1, CBOD2, CBOD3)
Assumptions remaining from the CH2M inputs were deemed valid and remained in the revised
model.
Algae (ALG1, ALG2, ALG3)
Constant value inputs for algal concentrations were set based on the assumption of a small
base algal input of 0.2 mg /L from the non -point sources. This assumption is consistent with the
CH2M model.
Dissolved Oxygen (DO)
Dissolved oxygen was assumed saturated and calculated based on temperature inputs to
generate seasonably variable DO concentration values. The equation used was the same as
the one in the CH2M model.
22 1111
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
Total Inorganic Carbon (TIC), Alkalinity (ALK)
Assumptions remaining from the CH2M inputs were deemed satisfactory and remained in the
revised model.
Labile /Refractory Dissolved /Particulate Organic Matter as Phosphorous (LDOM P, RDOM P,
LPOM P, RPOM P)
Dissolved organic P (DOP) was computed from the difference in total dissolved P (stemming
from GWLF) and phosphate. Particulate organic P (POP) is the difference in TOP (see
phosphate above for estimating TOP) and DOP. Both DOP and POP were split as 20% labile
and 80% refractory. DOP and POP represent DOM as P and POM as P, respectively.
Labile /Refractory Dissolved /Particulate Organic Matter as Nitrogen (LDOM N, RDOM N,
LPOM N, RPOM N)
Dissolved organic nitrogen (DON) was computed based upon the difference in the total
dissolved N (stemming from GWLF) and the computed concentrations for dissolved inorganic N
(ammonium and nitrate) discussed above. Particulate organic N (PON) is the difference in TON
and DON, where TON was computed as discussed above under Nitrate and Ammonium. Both
DON and PON were split as 20% labile and 80% refractory. DON and PON represent DOM as
N and POM as N, respectively.
Point Source Concentration Inputs
The concentration files for non -point and point sources do not differ based on the number and /or
type of constituent inputs. However, the sources of the nutrient and physical parameter data do
differ from one point source file to another. The following is a description of the methods used to
generate the nutrient concentration files for the various point source inputs.
The existing Holly Springs WWTP constituent data was obtained from monthly DMR summary
reports provided from the municipality for every month in the 2001 -2008 period of record. The
data included Total Nitrogen and Total Phosphorus measurements up to a weekly interval
period, as well as weekday reporting of dissolved oxygen and ammonia - nitrogen. The TN and
TP values were averaged on a monthly basis. Per the conclusions of the Utley Creek
Watershed Report (Tetra Tech, 2004), nutrient loads from the discharge were reduced (TN
inputs by 26% and TP inputs by 58 %) to account for reductions occurring during stream
transport in the small impoundments along the creek during transport to the lake. The
reductions were based on a statistical analysis of TN and TP values measured just downstream
of the WWTP discharge compared to values measured just upstream of a main stem branch of
Harris Lake. These reductions were applied prior to performing the nutrient fractioning
calculations for TN and TP. The assumptions that differ for the existing Holly Springs WWTP are
as follows:
23 1111
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
• All nitrogen and phosphorus was assumed to be in the dissolved state.
• Total phosphorus (TP) was split as 75% Phosphate (PO4) and 25% DOP for exiting
conditions at the treatment plant.
• Total Nitrogen (TN) was assumed to be 81% Nitrate (NO3). Ammonium (NH4) was
based on monthly averages of the recorded DMR values. DON was computed from TN
minus NO3 minus NH4.
• Dissolved Organic Matter (DOM), as well as DOP and DON, was fractioned to 15%
Labile DOM and 85% Particulate DOM. This ratio applies to all other organic inputs.
• DMR data for dissolved oxygen was averaged on a monthly interval for model input.
The future HS WWTP concentration file was based on a speculative discharge limit letter
provided to the Town by NCDWQ. The letter outlined maximum allowable TN and TP values on
a mg /L basis. The letter specified that the proposed future discharge would be relocated to a
point in very close proximity to the lake, thereby eliminating the justification for any nutrient
reduction as was utilized for the existing inputs. To fully characterize the potential nutrient load
to the lake system, the maximum speculative limits for TN and TP were used as the base inputs
for the nutrient fractioning calculations. The assumptions that differ from the existing HS WWTP
inputs are as follows:
• Total phosphorus (TP) was split as 54% Phosphate (PO4) and 46% organic phosphorus
(DOP). The speculative letter limit for TP is 0.5 mg /L.
• Total Nitrogen (TN) was split as 78% Nitrate (NO3), 5% ammonia (NH4). DON was
computed as TN minus NO3 minus NH4. The speculative letter limit for TN is 5.0 mg /L.
The Harris Energy & Environmental Center input file was identical to the file used in the CH2M
model file for both existing and future scenarios. The values contained within were deemed to
be valid for the assumed level of treatment applied to the domestic wastewater stream.
The cooling tower blow -down discharge concentrations were calculated for both existing and
future scenarios to maintain a nutrient (and flow) mass balance through the plant. Calculations
were performed to account for evaporative losses and the mixing of heated process water with
ambient temperature dilution water. An explanation of the iterative processes used during model
calibration and scenario analyses can be found in Section V and VI of this report.
The Cape Fear River concentration file was developed using recorded USGS monitoring data
from 1988 to 2003. The 0210215985 USGS Station located approximately 2 miles upstream of
the proposed makeup water intake point was deemed the best available source of water quality
data. The data were recorded at monthly intervals over the period of record. The assumptions
used to generate the CFR concentration input file are as follows:
• TN, TON, DON, NO3, NH4, TP, PO4 and TOP are grouped into values from the same
month and averaged over the period of record. Averaging the values in this manner (all
24 1/11
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
January values averaged together, Feb, etc.) allowed for a degree of seasonal nutrient
variation to be introduced into the initial input. The 12 monthly constituent averages were
used to generate an input for one year's time. The values for the initial input year were
repeated annually for 8 years to create the input file.
• POP inputs were dictated by measured TPP minus (PO4 minus Ortho -P) multiplied by
31/95. DOP inputs were dictated by TP minus POP..PON inputs were dictated by TON
minus DON.
• Organic matter was used to represent organic carbon in version 3.6 of the model, as
stated previously. From stoichiometry, organic matter is about 45% organic carbon.
• Organic matter, including organic P or N, was split to be 20% labile and 80% refractory.
• Suspended Solids, dissolved solids, iron, dissolved oxygen were set using the averages
of the divided monthly recorded data.
• Algal constituents were assumed to have the same base input concentration as the non-
point sources.
3. Temperature Inputs
Nonpoint Source Temperature Inputs
The five branches and tributary inflows were assigned identical temperature inputs in the CH2M
model. To achieve a more accurate temperature calibration, the available meteorological data
was reviewed to determine if the existing stream temperature input data was valid. The lack of
historical water temperature observations coupled with an insufficient knowledge of the
methodologies used to derive the existing temperature data resulted in the need to verify and /or
revise the existing nonpoint source temperature inputs, as those inputs would affect the overall
temperature balance of the lake system. After confirming the validity of the existing
meteorological air temperature data utilized in the CH2M model, a nonlinear logistic equation for
estimating weekly stream temperatures (Mohseni et al. 1998) was used to develop the nonpoint
source temperature files. This equation is stated as,
Where,
=P+1 +er(Q a) �3)
Ts = weekly average stream temperature, deg C
Ta = weekly average air temperature, deg C
N = minimum weekly average stream temperature during the year, deg C
a = maximum weekly average stream temperature during the year, deg C
Y = function slope steepness parameter
25 1111
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
R = air temperature at the function inflection point, deg C
Observed stream temperatures were used by the equation developers with the above equation
and observed air temperatures for 573 streams across the U.S. to obtain the best fits for each of
the parameters in the above equation. The mean of the best fitted parameters are shown in
Table 3 below. These values were used along with Equation 3 to compute stream temperatures
for the Harris Lake tributaries.
Table 3. Mean of Best Fitted Parameters Used to Compute Stream Temperatures
a
R
Y
p
26.2
13.3
0.18
0.8
A rolling average of the prior 7 days air temperature for the Harris site was used to generate
daily data with trends similar to that observed in the existing CH2M temperature data. The six
nonpoint sources were assigned identical water temperature inputs, just as in the existing
model. Future model scenarios utilized the same temperature inputs as the existing, as the
estimations of the data were considered typical for any sequential 8 year period.
Point Source Temperature Inputs
The temperature data for the Holly Springs Municipal wastewater treatment plant was taken
from the CH2M input files, after verification of the data was completed. Future model scenarios
utilize the same temperature inputs as existing, to maintain consistency with the nonpoint
source data.
Temperature inputs for the Harris Energy & Environmental Center WWTP were copied from the
Holly Springs Data as that data was assumed to be consistent for all wastewater input streams
in the area.
Temperature estimates for reactor process water were based on recorded values from PEC
databases for existing lake conditions. The database of reactor process flows provided by PEC
included daily average temperature readings of the withdrawal and blowdown water over the
entire 2001 -2008 period of record. An assumption was made after conversations with PEC that
the temperature readings were taken after the heated water was mixed with ambient
temperature dilution water. This assumption validated the direct use of the PEC blowdown
values as model input temperatures for the blowdown flow.
Temperature values for future scenarios are not input directly into the model; rather an iterative
process was necessary to more accurately predict future temperatures. Every initial model run
for a future scenario generated a daily temperature time series output for the ambient
temperature withdrawal water. The absence of comprehensive data regarding the thermal
characteristics of the future reactors required an assumption to be made that the future process
26 1/11
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
would impart the same increase in temperature (delta T) to reactor process water as in the
existing. Another assumption was made that the mixing of ambient temperature withdrawal
water and heated reactor process water would occur at the same ratio as the existing process.
These two assumptions allowed the delta T calculated from the existing PEC data for daily
withdrawal and blow -down to be applied to the withdrawal water outputs for the initial future
runs. The new temperature values (model generated withdrawal water output T + calculated
daily delta T) were utilized as the final blow -down temperature time series for each future
scenario.
The Cape Fear River point source input temperature file was generated from historical data
obtained from the NCDWQ monitoring station #136160000 at NC Highway 42 in Corinth, NC.
Using historical monthly recorded temperatures from grab samples, an average temperature
value was developed for each month in the calendar year. Using the monthly average values, a
time series for the entire period of record was developed by repeating the monthly values
annually over the 8 year span of analysis.
Temperature inputs are not required for release or withdrawal water (reactor cooling water, dam
spill, and low flow releases) since the temperatures of these releases are computed during the
model simulation.
4. Updates to Model Grid and Bathymetry
The model grid segmentation was revised to provide more spatial definition where it was judged
that some segments were too large. Four upstream segments were split in half resulting in an
additional four segments, or 37 segments in all including the boundary segments. Segments 2
(Branch 1), 19 (Branch 3), 24 (Branch 4), and 28 (Branch 5) were split and bathymetric data
were added for the new segments. As a result, the first two segments for the headwaters of
these branches are now numbered 2 and 3, 20 and 21, 26 and 27, and 31 and 32.The revised
grid is shown in Figure 4 and Figure 7.
Two layers were also added to the grid bringing the total number of layers to 25. One layer was
added during calibration after it was realized that the lake was not quite deep enough at the
dam. The second layer was added to accommodate the pool rise for future conditions. Grid and
bathymetric revisions were carefully conducted to ensure that the correct lake capacity versus
elevation was maintained.
5. Meteorology
Local meteorology is included within a boundary condition file required by the W2 model to
compute surface heat exchange and solar radiation, which are used in heat source terms for
temperature simulation. Solar radiation is also used to compute underwater irradiance used for
plant growth. The meteorological inputs required include air temperature, dew point
temperature, wind speed and direction, cloud cover (0 to 10, where 0 is clear sky), and solar
27 1/11
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
radiation (W /m2). The meteorological input file from the CH2M model was developed from a
local station and deemed satisfactory for use in the present model, with the exception of wind
data, where it was revealed that data entry errors had resulted in some incorrect values in the
file. The meteorology data file was updated to reflect correct wind values. (Refer to discussion
of review and recalibration in Section IV -6.)
28 1111
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
V. CE- QUAL -W2 Model Calibration
While the CH2M HILL model was developed in version 3.2, in order to take advantage of
greater flexibility in nutrient parameter inputs, it was decided to change over to the latest release
version of W2, version 3.6. Version 3.6 has variable stoichiometry, which means that organic N
and P can be explicitly treated as state variables, and their respectively loadings can be directly
managed in model inputs. Additionally, version 3.6 allowed full control of loadings specifications
for organic N, P, and C. For these reasons, only results obtained with version 3.6 are presented
in this report. Likewise, the final scenario results were also conducted using version 3.6.
The modelled water quality state variables are shown in Table 4. These are the variables that
are transported within the lake and that require input loadings. There are also numerous derived
variables that are not transported but can be determined from the modelled state variables.
These include quantities such has pH, TKN, TN, TP, etc.
Table 4. Modeled Water Quality State Variables
Total dissolved solids (TDS)
Labile dissolved organic
Labile dissolved organic
matter (LDOM)
phosphorus (LDOP)
Water age
Refractory dissolved organic
Refractory dissolved organic
matter (RDOM)
phosphorus (RDOP)
Conservative tracer
Labile particulate organic
Labile particulate organic
matter (LPOM)
phosphorus (LPOP)
Inorganic suspended solids
Refractory particulate organic
Refractory particulate organic
(ISS)
matter (RPOM)
phosphorus (RPOP)
PO4 -P
Carbonaceous biochemical
Labile dissolved organic
oxygen demand (CBOD)
nitrogen (LDON)
NO3 -N
Algae 1 (diatoms)
Refractory dissolved organic
nitrogen (RDON)
Dissolved inorganic silica
Algae 2 (greens)
Labile particulate organic
nitrogen (LPON)
Particulate organic silica
Algae 3 (blue greens)
Refractory particulate organic
nitrogen (RPON)
Total iron
Total inorganic carbon (TIC)
Temperature
Dissolved oxygen (DO)
Alkalinity
29 1111
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
Data were collected between January 1, 2006 and December 31, 2008 from several water
quality monitoring stations located in the lake. The observed data were used during the
calibration process via comparison to model predictions. Monitoring site locations are shown in
Figure 8.
30 1111
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
Chatham
County
Wake
A County
Legend
Monitoring Stations N
— Waterbodies
CE- QUAL -W2 Boundaries
QHarris Lake Watershed
0 2,500 5,000 10,000
Feet
I Full
�q
�a
34
35
/Chatham
County
*Note that CE- QUAL -W2 Bo daries are
1
Harnett 1
County 19
Figure 8. Water Quality Monitoring Stations in Harris Lake
31
,/,1
moffatt & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
1. Water Balance
The first step in the model calibration process was to compare computed lake pool elevations
against observed elevations to verify an accurate water balance, which ultimately affects
properly computed hydrodynamics. The process was conducted several times due to changes
in tributary inflow rates and precipitation, as well as changes to the grid noted earlier. The final
water balance was based on inflows and precipitation furnished by Hydrologics, Inc., developed
for input into the OASIS Hydrologic model (refer to Section IV -1). The only outflows were those
computed within the W2 model for spillway releases. The final water balance also used the final
model grid and bathymetry which included the additional segments and layers discussed
previously.
The model was applied to the period January 1, 2001, through December 31, 2008, with only
hydrodynamics and temperature simulated, and water quality components turned off. The
computed pool elevation matched the observed elevations quite well as shown by the plot in
Figure 9 below. Computed lake pool levels are affected by tributary and branch inflows,
withdrawals, spillway releases, precipitation, and evaporation. Evaporation was computed within
the W2 model during the simulation.
Computed and Observed Pool Elevations
fig
E
e fib
O
n 67.5
w
67 •Observed
OF
O 66.5 - MAO
0
a 66
65.5
U 500 IUUU 1500 2000 2500 3000
Julian Day, starting on Jan 1,2W1
Figure 9. Comparison of Computed and Observed Pool Elevations for Final Water Balance
2. Temperature Calibration
Lake water temperature calibration was conducted next with other water quality variables turned
off. The period January 1, 2006, through December 31, 2008, was used for temperature
calibration. The primary parameters that were adjusted during this calibration process were the
water extinction coefficient for short wave solar radiation, the percentage short wave solar
radiation absorbed in the surface layer of the lake, and the wind sheltering coefficients. Two
sets of wind sheltering coefficients had to be used to achieve calibration, a winter value and a
summer value, where the winter value was higher (less sheltering). The winter value extended
32 1/1'll
mofiati & nichol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
from October 1 to April 1 of each year, and the summer value was for the other six months. The
two values that were finally set after all calibration (including water quality) were 0.95 and 0.65
for the winter and summer periods, respectively.
There are other parameters that can be adjusted that affect hydrodynamics and temperature,
but these were set to recommended default values. There are various options for computing
turbulent vertical mixing, and the W2 option was selected, which is also the recommended
default. All of the other vertical mixing options were tested, and the W2 option provided as good
as or better results than the others. Initial model calibration involved more than adjusting
parameters. It also required checking to ensure that all flow structure descriptions and related
geometric inputs for lake inflows and outflow were properly specified. Although a fairly good
temperature calibration was achieved, temperature had to be recalibrated when all of the other
water quality variables were activated since the inclusion of algae and organic matter had a
substantial effect on short wave solar radiation penetration and the resulting temperature
stratification.
3. Water Quality Calibration
Water quality variables were activated and simulated for the same 2006 — 2008 period, and
computed concentrations were compared with measured values for the period. Various water
quality kinetic rate parameters and coefficients were varied within reasonable bounds to bring
computed results into agreement with measured values. In addition to temperature, water
quality variables that were measured and examined during calibration included: dissolved
oxygen, chlorophyll -a, ammonium nitrogen (NH4 -N), nitrate + nitrite nitrogen (NO3 -N), TN, total
Kjeldahl nitrogen (TKN), TP, and TOC. Other variables were included in the model, but there
were no measured data for most of those variables. There were observed data for pH, alkalinity,
and total dissolved solids (TDS). Model alkalinity and TDS are conservative state variables, i.e.,
they have no reaction kinetics in the model. Thus, if a good water balance is achieved, and if the
inflowing concentrations are sufficiently accurate or adjusted for accuracy, the computed
concentrations for TDS and alkalinity should approximately match observed, and they did for
this study. Water pH is a reactive variable in the model, and is affected by total inorganic
carbon, which is affected by photosynthesis, respiration, and biodegradation. Although
computed pH was within reasonable agreement with observed values, it was not a focus for
calibration since pH is not a concern for this lake. There are various rate coefficients and
parameters associated with each of the target variables that had to be specified, but the key
parameters that were adjusted during calibration included:
• Water temperatures associated with calculating algae temperature rate multipliers for
growth of three algal groups (diatoms, greens, and blue greens)
• Algal maximum specific growth rate for the three algal groups
• Ratio of algal biomass to chlorophyll -a for the three algal groups
33 1111
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
• Algal half saturation concentrations of nitrogen and phosphorus for computing nutrient
limitation functions for growth of the three algal groups
• Optimal solar radiation for computing the light limitation function for growth of the three
algal groups
• Nitrification rate
• Sediment denitrification rate
• Organic benthic sediment mineralization rate
• Organic benthic sediment burial rate
The option for simulating benthic sediment diagenesis with a first -order mineralization process
was selected. To handle low - level, background SOD, a zero -order SOD was also specified. The
inclusion of the zero -order SOD, which was set to 0.5 g /m2 day, had a slight effect on dissolved
oxygen (DO) and little to no effect on other water quality variables.
In addition to striving to have computed concentrations agree with measured data, there was
also the goal of having the three algal groups following in succession with diatoms dominating
winter and early spring and green and blue green algae dominating during late spring to late -
summer and fall. The model was able to capture these trends.
4. Phosphorous Calibration
Calibration consisted of adjusting input parameters (kinetic coefficients and rates), running the
model, and then comparing the output to measured values. This process was repeated
numerous times until reasonably good agreement was achieved for all key variables. However,
during this process, it was recognized that excessively high total phosphorus (TP)
concentrations were being computed within the lake. There are only two permanent sinks for
TP, which include lake outflows and burial in benthic sediments. Since the lake outflow rates
were accurately modeled, and benthic burial is a slow process with limited effect for TP removal,
the primary cause of excessive TP was attributed to excessive loadings from the watershed.
The watershed loadings were developed from the GWLF model; thus, the nonpoint source loads
as delivered via the tributaries are the largest area of uncertainty in the input data.
The method to address this overestimation of total phosphorous from GWLF was to
successively iterate the W2 model with increasing scaling factors applied to adjust the tributary
nonpoint source loads downward until predicted TP levels approximated observed in -lake
concentration data. Through this iterative process, it was determined that an 80% reduction in
the nonpoint source phosphorous loads derived from the GWLF model produced the best fit
between the W2 model predictions and observed values for in -lake TP concentrations. This
same problem was encountered in the previous version of the Harris Lake model and was
addressed in this same fashion by the previous modelers (CH2M HILL, 2009).
34 1111
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
As mentioned previously in Section III, the Harris Lake GWLF model utilized the same nutrient
loading inputs as were developed for the GWLF -based watershed model that supported the
Jordan Lake TMDL study, where those model inputs resulted in predictions that closely matched
the loads delivered to the reservoir. To investigate potential reasons why the same inputs could
lead to a significant over - prediction of TP levels in Harris Lake, an analysis of the differences in
soil properties between the two watersheds was performed. Soils were examined because the
watershed for Jordan Lake drains primarily Piedmont Soils, while the watershed for Harris Lake
is located within the Triassic Basin and drains some transitional Coastal Plain soils.
The comparative soils analysis was performed using the NRCS Web Soil Survey
( http : / /websoiIsurvey.nres.usda.gov /app /HomePage.htm) by selecting sample areas within the
Jordan Lake watershed and the Harris Lake watershed and extracting key soil factors within
each sample area. Area weighted values for each of the soil parameters examined were
calculated and the results from the comparative analysis are shown below in Table 5. The Web
Soil Survey limits interactive map sample areas to 10,000 acres so only one sample area within
the Harris lake watershed could be accommodated (denoted by the Wake County entry in the
table), while 4 sample areas where selected to represent a broad geographic sample of the
1700 square mile Jordan Lake watershed. The results in Table 5 show that with the exception of
the sample area in Orange County, the Harris Lake sample area had a higher K- Factor than all
the sample areas in the Jordan watershed. In soil survey data, the K- Factor is the principle
measure of the propensity of the particles of a given soil to be eroded (dislodged by water).
The results in Table 5 also show that the soils from the Harris Lake watershed sample area
have a higher cationic exchange capacity (CEC) than the soils in all sample areas within the
Jordan Lake watershed. In soil survey data, the CEC is the measure of a soils capacity to bind
with ionic particles. In short, the results of the comparative analysis indicate that soils within the
Harris Lake watershed had a greater capacity to be eroded into streams, and once in those
streams, have a greater capacity to bind with phosphorous particles and settle. Such
distinctions in soil properties could explain why the Harris Lake watershed has a higher potential
to trap phosphorous before it is delivered to the lake, or facilitate rapid settling of bound
phosphorus as it enters the lake. The results of this analysis support the justification to apply
scaling factors to reduce the nonpoint TP loads while not reducing the TN loads as was done
with the previous version of the model, especially given the far greater propensity of
phosphorous to be bound to soil particles as opposed to nitrogen.
35 1111
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
Table 5. Results of Web Soil Survey Comparative Soils Analysis
Over and above the sediment dynamics in the Harris Lake watershed, the morphometry of the
lake itself is very likely to be contributing to reductions in the TP loads from its tributaries. The
largest portion of the mass load entering Harris Lake comes down Utley Creek and enters into
the arm designated as Branch 3 in the W2 model. Near the headwaters of Branch 3 New Hill
Holleman Road crosses on an elevated causeway with a narrow outlet to the main body of the
lake (Figure 10). This constriction and secondary impoundment amount to what could be
considered a treatment forebay on this arm of the lake. Close inspection of the aerial
photograph in Figure 10 reveals the green growth in the headwaters and around the edges of
the forebay, which consists of massive beds of large emergent aquatic plants, as shown in
Figure 11. The opportunities for settling and uptake by plants could result in substantial
reductions of the TP loads, as well as sediment and TN loads, entering this arm of the lake.
36 1111
moffati & mchol
Area of Interest
Area - weighted Soil
Parameters
County
Acres Sampled
K- factor
CEC
% Clay
Alamance
8886.8
0.24
4.6
15.1
Chatham
9435.6
0.17
3.3
13.1
Guilford
9928.8
0.2
4.3
15.7
Orange
9856.1
0.4
3.8
16.1
Wake
9808.7
0.3
4.9
13.9
Over and above the sediment dynamics in the Harris Lake watershed, the morphometry of the
lake itself is very likely to be contributing to reductions in the TP loads from its tributaries. The
largest portion of the mass load entering Harris Lake comes down Utley Creek and enters into
the arm designated as Branch 3 in the W2 model. Near the headwaters of Branch 3 New Hill
Holleman Road crosses on an elevated causeway with a narrow outlet to the main body of the
lake (Figure 10). This constriction and secondary impoundment amount to what could be
considered a treatment forebay on this arm of the lake. Close inspection of the aerial
photograph in Figure 10 reveals the green growth in the headwaters and around the edges of
the forebay, which consists of massive beds of large emergent aquatic plants, as shown in
Figure 11. The opportunities for settling and uptake by plants could result in substantial
reductions of the TP loads, as well as sediment and TN loads, entering this arm of the lake.
36 1111
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
Figure 10. Forebay Formed by Road crossing the Upper Portion of Branch 3
k�
1 • 1
Y'
�
Figure 11. Harris Lake Forebay Aquatic Plants
37 1111
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
In addition to the forebay formed by the roadway crossing on Branch 3, causeways constructed
for entry to the Shearon Harris Plant site form effective dual forebays on Branch 2 (Figure 12)
and the auxiliary cooling water impoundment acts as a large forebay on Branch 4 (Figure 13).
These forebays all have discernable areas of aquatic plant growth in their headwaters and offer
the same opportunities for sediment and nutrient load reduction through settling and uptake as
are likely realized by the forebay on Branch 3. Further, the arms of Harris Lake without
secondary impoundments also show significant areas of colonization by aquatic plants which
may be resulting in further load reductions.
Figure 12. Forebays formed by Road Crossings in the Upper Portion of Branch 2
38 1111
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
Figure 13. Forebay Formed by Auxiliary Cooling Water Reservoir
5. Cooling Water Withdrawal and Blowdown Discharge
The W2 model does not directly provide for the capability to simulate cooling water systems,
such as the one in Harris Lake. Cooling water can be withdrawn as either a lateral withdrawal or
an end -of- branch outflow, and heated return water can be introduced into the lake as a tributary
inflow, but the user must specify the temperature and water quality of the return, heated
discharge. This was the process used in this study. The temperatures and water quality
concentrations of the withdrawn cooling water were saved and processed within a spreadsheet
program to compute the temperatures /concentrations of the heated return flow, referred to as
the blowdown discharge at Harris Lake. Evaporative cooling water losses and the portion of
withdrawn water used for dilution had to be taken into account for computing the blowdown
discharge water temperatures and water quality concentrations. The spreadsheet computations
for the blowdown discharges were performed for daily values. Additionally, the spreadsheet
computations were mass - conservative; thus, withdrawn material mass and blowdown return
flow material mass were forced to be equal. However, the return flow concentrations were
higher than the withdrawal concentrations due to evaporative loss of cooling water.
The magnitude of evaporative loss plays an important role in the existing and future conditions
in Harris Lake. For example, the daily average tributary inflow (all six non -point source inflows)
for the period 2001 — 2008 is estimated to have been 1.89 cros. The projected daily average
39 1/11
moffall & nichol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
evaporative losses for the future Harris power plant with three units is estimated to be 2.35 cros,
which is greater than the existing non -point inflows. Thus, make -up water is needed from the
Cape Fear River. Evaporative losses on this scale result in blowdown water return flow
concentrations being substantially greater than withdrawal concentrations since material mass
must be conserved.
After each model run, within which significant adjustments had been made to calibration
parameters or other inputs, it was necessary to process the cooling water withdrawal water
quality into blowdown discharge water quality for the next iteration of the model. In addition to
the iterative procedure for properly handling the cooling water withdrawals and blowdown
discharge water quality, there was also the need to adjust initial conditions in the lake. After the
model was nearly calibrated, the output concentrations at the end of the three year simulation
were used to estimate the initial concentrations or initial conditions for the beginning of the
simulation. The calibration process was repeated with the revised initial conditions and
processed blowdown discharge inputs.
6. Recalibration
Following calibration of the model, the model was reviewed by water quality modeling
specialists from HDRIHydroQual. Their recommendations were evaluated, and a final set of
recommendations were accepted by HDRIHydroQual, M &N, and PEC. This set of final
recommendations for model calibration consisted of the following:
• Use corrected wind and meteorological inputs as provided by PEC and prepared
by HDRIHydroQual.
• Retain the zero -order SOD of 0.5 g /m2 /day but set the zero -order sediment
release rates for ammonium and phosphate to zero.
• Use algal Whalf saturation values for computing ammonium preference to values
of 0.015, 0.015, and 0.001 for the three algal groups, which are the same as
those used for algal uptake.
• Set the nitrification rate to 0.10 day'.
• Set RBOD = 4.52 and the CBOD decay rate to 0.05 day'.
The above parameters are described within the W2 user manual. All of the final calibration
results presented below, as well as scenario results that follow, were obtained by applying the
above calibration revisions.
7. Temperature Calibration Results
Computed and observed temperatures are plotted in Figure 14 as depth profiles for the dates of
observation and for the station at the dam (E2). Computed temperatures match observed better
on some dates than on others. These results are with all water quality variables activated and
calibrated. Inaccuracies in computed temperature are attributed to the sensitivity in vertical
40 1/11
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
mixing for Harris Lake, which is probably caused by the blowdown discharges. The model lake
was particularly sensitive to the wind sheltering coefficients. Improvements in computed
temperatures were obtained by adjusting the wind sheltering coefficients for some periods that
had more error, but these error improvements were usually accompanied by greater error for
other periods that had better accuracy before the adjustment. Thus, only the two values for the
wind sheltering coefficient were used for the winter and summer period. Overall, the
temperature calibration is quite good, particularly after applying the revised wind data.
41 1111
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
Ian 24, 2006, day 1850, Sta. E2
70
65
E
C 60
= Computed
.Q 55 • Observed
W
50
0 10 20 30
Temperature, deg C
Jul 10, 2006, day 2017, Sta. E"2
70
65
E
e 60
g Com puted
.w 55 • Observed
W
50
0 10 20 30
Temperature, deg C
Jan 31, 2007, day 2222, Sta. E2
70
65
E
e 60
—
.Q
Y
55
W
50 i
0 1u �C 30
Temperature, deg C
�Com puted
• Observed
May 10, 2006, day 1956, Ste. E2
fry
E
F' f0
9 - Computed
jj 55 • Observed
iY
0 10 20 30
TernperaW re, deg C
Nov 30, 2006, clay 2160, Sta. E2
65
E
F fQ
9 -Co m p ute d
55 • Observed
lu
50
{3 10 2.0 3C
Temperature, deg C
May 2, 2007, day 2313, Sta. E2
ry
b5
'c
—
.2
Y
i
W
50 ir
0 10
Temperature, deg
r-
Figure 14. Computed and Observed Temperature Profiles for Station E2
Co m p uteri
• Observed
42 1111
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
Jul 25, 2007, clay 2397, Sta. E2
70
65
E
e 60
•g —Computed
W 55 • Observed
50 '
Temperature, deg C
Jan 25, 2005, day 2551, Ste. E 2
70
65
d &D
_g 55 • obseR.v.
iu
50 T
10 i0 30
4emperature, degC
Mar 18, 2008, day 2634, Sta. E2
{ 0 -
65
e 60
= Com p ute d
yi 55 Observed
W
50
0 10 20 30
Temperatu re, deg C
Nov 27, 2007, day 2522, Sta. E2
70
65
E
C? 60
= Com pute tl
W 55 • Observed
50
0 1 ; iU 3u
Temperature, deg C
Feb 21, 2008, day 2608, Ste. E2
70
65
E
C? gp
9 �Corn FDLIte
.s 55 • 0I)SE Ived
iu
50 ,
0 10 20 30
Temperature, deg C
Apr 10, 2008, day 2657, Ste. E2
!V
65
E
C? gp
9 Com puted
-2 55 — • Observed
Lu
50
0 10 iU uu
Temperature, deg C
Figure 14 Continued
43 1111
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
May 21, 2008, day 2698, Sta. E2
70
65 -
E
e 60
Computed
4' 55 + Observed
W
50 y,
10 sU Ji1
Temperature, deg C
Jun 17, 2008, day 2725, Sta. E2
70
65
6Q
S Com putt d
_Q 55 �► Obmrved
W
u 10 20 30
Temperature, deg C
Jul 16, 2008, day 2754, Sta. E2
7a -
65 —
E
e 60
Computed
-9 55 • Observed
W
0 10 20 30
Temperature, deg C
Jun 3, 2008, clay 2711, Sta. E2
7C
65 ` —
E
6C ..
0 Computed
55 + Observed
W
�y,�y
V 10 20 M
Temperature, deg C
Jul 2, 2008, day 2740, Sta. E2
�v
g . Coriptit "'.
r
55 vii
W
5u '
Temperature, deg C
Aug 6, 2008, clay 2775, Sta, E
7c
65
60
Com p ute d
55 • Observed
W
50
Temperature, deg C
Figure 14 Continued
44 1111
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
Aug 20, 2008, day 2789, Sta. E2
C
65 .
g — Computed
& 55 6 CfbsErt,Ed
W
50
0 10 20 30
Temperate re, deg C
Sep 23, 2008, day 2823, Sta, E
70
65
E
e 6C.
.g Conn p ut� d
W 55 0bsErt,ed
5^ T
Temperature, deg C
Nov 12, 2008, day 28 r 3, Sta. E2
70
65
E
e 60
0 — Com p ute d
3Q 55 s Observed
W
50
0 10 20 510
Temperature, deg C
Sep 9, 2008, day 2809, S } a. E2
70
65
E
e 60
g — Cv- !I:We.l
.2 • C)bserved
50
0 10 20 30
Temperature, deg C
Oct 22, 2008, day 2852, Sta. E2
7v
e5 —
e 60 .
A Computed
4
!z �' • O b--,E rve d
W
5C
0 10 20 s
Temperature, deg C
Dec 3, 2008, day 2894, Sta. E2
7�j
65
E
e 60
0 — Computed
s 55 • 0 bse rve d
W
5Q '
0 1 11 20 3^
Temperature, deg C
Figure 14 Continued
45 ,d„
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
8. Dissolved Oxygen Calibration Results
Computed and observed DO concentrations are plotted in Figure 15 as depth profiles for the
dates of observation and for the station at the dam (E2).The model is most accurate for late
spring and early summer conditions when the lake is the most stratified and the hypolimnion is
anoxic. The model is less accurate in the late fall and winter.
There are two primary causes of differences in computed DO from observed. One cause is
prediction of vertical density stratification (due to temperature), with either too much or not
enough vertical mixing at times.
The second primary cause of DO inaccuracy in this model is the formulation of the first -order
diagenesis sediment model within W2. This formulation does not include the release of chemical
oxygen demand (COD) when benthic organic matter is decomposed during times when the
overlying hypolimnetic water is anoxic. Anaerobic organic matter decomposition results in the
reduction of other components, such as nitrate, iron and manganese, and sulfate. Ultimately
methanogenesis occurs with the release of methane. Such reductions can result in the release
of oxygen demanding reaction products, such as sulfide and methane. The presence of
sediment - released COD in the water column results in lower DO following fall turn -over of the
lake. Comparison of the plots in Figure 15 shows that this is the situation in most cases since
computed DO is usually greater than observed during late fall through winter.
Despite the slight over - predictions in late fall and winter, the model performs well with regard to
dissolved oxygen during periods in which critical conditions resulting from oxygen depletion are
likely to be of greatest concern to the ecological health of the reservoir (spring and summer).
46 1111
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
Jars 24, 2006, stay 1850, Sta. E2
70
65
E F___
e 60
Q -Com p ute J
_w 55 • a bse rve d
W
50
0 5 10 15
Dissolved oxgen, mgfl
Jul 10, 2006, clay 2017, Sta. E2
70
65
E
e' 60
= Computed
_w 55 0 Obsen.-v.
W
50
0 5 1�j 15
Dissolved oxygen, mg/L
Jan 31, 2007, day 2222, Sta. E2
70
65
E
e 60
0 Conn pute d
55
W 50W • Obsertied
50
0 5 10 15
Dissolved oxygen, mgjL
May 10, 2006, day 1956, Sta. E2
-V
65
E
60 -
Q Cor7 JsLIU tl
55 - Dbs°
W
50 � .
0 5 ! _5
Dissolved oxygen, mg�L
Nov 30, 2006, clay 2160, Sta. E2
-V
e5
C 60
= - Computed
55 • Obsenred
W
5C
5 f0 15
Dissolved oxygen, mgfL
May 2, 2007, day 2313, Sta, E2
70
65 0
E
e 60
0 Computed
.9 55 • Observed
W
50
5 !v 15
Dissolved oxygen, mgf L
Figure 15. Computed and Observed DO Concentration Profiles at Station E2
47 1111
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
Jul 25, 2007, day 2397, Sta.. c2
7�D
65 —
E
e 6o -
= Computed
-w 55 - • Obmwed
W
50 ,
U 5 _ 15
Dissolved oxygen, mg/L
Jan 25, 2008, day 2581, Sta. E2
rw
b5 '
E
C �v
g Co m p ute d
3� 5 5 - • O bse rte d
W
50 �.
C 5 10 15
Dissolved oxygen, mg/L
Mar 18, 2008, day 2634, Sta, E2
7G -
65
E
C? 60
= Computed
_Q 55 • 0 bse we d
W *�
fl 5 10 15
Dissolved oxygen, mg/L
Nov 27, 2007, day 2522, Sta. E2
70
b5
'c
60
.2 Co m p ute d
� 55 • O bse rte d
W
5v —
U 5 ! 15
Dissolved oxygen, mg/L
Feb 21, 2008, day 2608, Sta. E2
70 -
65
C
C $0
0 Conn puted
-w 55 Observed
W
50 ,
0 5 10 15
Dissolved oxygen, mg/L
Apr 10, 2008, clay 2657, Sta. E2
7G -
b5 –
'c
e
60
= Computed
g 55 Observed
W
fl 5 10 15
Dissolved oxygen, mg/L
Figure 15 Continued
48 1111
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
May 21, 2008, day 2698, Sta. E2
70
65
g 60
Computed
4' 55 1M 0bsen-ed
W
50 r
0 5 10 15
Dissolved oxygen, mg/L
Jun 17, 2008, day 2725, Sta. E2
70
65
E • dF
60
= Com p uteri
9Q 55 • observed
w
50
0 5 !C 15
Dissolved oxygen, mg/L
Jul 16, 2008, day 2754, Sta. E2
70
65
E
60
•2 Com puteri
3Q 55 • Observed
W
50
0 5 1 C 15
Dissolved oxygen, mg/L
Jun 3, 2005, day 2731; Sta. E2
70
65
E •
g 60
+, Com pute Cl
.2 55 Observed
W
50
0 5 10 15
Dissolved oxygec, nWL
Jul 2, 2005, day 2740, Sta. E2
70
65
E
60
= Com p ute d
.2 55 M Observed
W
50
0 5 1u 15
Dissolved oxygen, mg/L
Aug 6, 2008, day 2775, Sta. E2
70
65 1 IL
C 60
•2 Com pute {I
.w 55 • 0hseR.ed
W
50
0 5 10 15
Dissolved oxygen, mg/L
Figure 15 Continued
11,1
moffall & nichol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
Aug 20, 2008, day 2789, Sta. E2
70
65
d 60
= �Computecl
-w 55 Observed
W
0 5 1C 15
Dissolved oxygen, mg/L
Sep 23, 2008, day 2823, Sta. E2
70
65
E
g 60 •
2 * �Computecl
3w 55 observed
W
50 -
�
Dissolved oxygen, mol-
Nov 12, 2008, day 2873, Sta. E2
70
65
E
d 60
= �Computecl
-w 55 0 O b se rve d
W
5C 0 T
U 5 10 15
Dissolved oxygen, mg /L
Sep 9, 2008, day 2809, Sta. E2
70
65
E
d 60
•g -Computed
9
3Q 55 • ObseRed
W
50
0 5 1 15
Dissolved oxygen, mg/L
Oct 22, 2008, day 2852, Sta, E2
65
60
g
2 - Computed
-9 55 • observed
W
5C
5 1C -5
Dissolved oxygen, mg /L
Dec 3, 2008, day 2894, Sta. E2
70 .
65
E
d 60
g - Computed
_Q 55 * Obse rtre d
W
5 "_0 15
Dissolved oxygen, mg/L
Figure 15 Continued
50 1111
moffall & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
9. Nutrient and Chlorophyll -A Calibration Results
Nutrients and chlorophyll are plotted as time series for four of the observation stations, E2, H2,
P2, and S2. Surface values are plotted for each constituent. Additionally bottom values are
plotted for ammonium since it is released from benthic sediments during anoxic conditions.
Station E2 is in branch 1 at the dam. Station H2 is in branch 5 in a segment that starts 2.3 km
upstream of the intersection of branch 5 with branch 1. Station P2 is in branch 1 in a segment
that starts 4.96 km upstream of the dam. Station S2 is in branch 3 in a segment that starts 2.4
km upstream of the intersection of branch 3 with branch 1. Locations of these stations are
illustrated in Figure 8.
Calibration for the nitrogen components are presented first. Computed and observed time series
concentrations for nitrate at the water surface are shown in Figure 16. Computed and observed
time series concentrations for ammonium at the water surface and near the lake bottom are
shown in Figure 17 and Figure 18, respectively. Computed and observed time series
concentrations for TKN and TN at the water surface are presented in Figure 19 and Figure 20,
respectively. Computed nitrogen generally compares well with observed data. The model
accurately predicts the annual cycles of nitrate similar to the observed data. Ammonium levels
were anticipated to be low, similar to observed, and this was verified by the model. The higher
values of ammonium are on the lake bottom due to sediment release during anoxia. Computed
and observed TKN and TN concentrations are similar. Most of the TKN is composed of organic
N. Although accurate N predictions are important for representing the N cycle and have
implications for DO, N is not the limiting nutrient for algal growth in Harris Lake. P is limiting
whenever there is sufficient light.
51 1111
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
Station E2, Surface
0.4
J
4Q
E 0.3
9F' •
°= 0.2
:3 • • - Computed
0.1 1 • Ghserved
Z 1,827 2,192 2,557 2 ,922
Julian Day
Station P2, Surface
0.4
J
E 0.3
ti
±' 0.2 • •
' - Computed
0.1 a Observed
a+ 0
Z 1,827 2,192 2,557 2,922
Julian Day
Station H2, Surface
0.4
J
E 0.3
9F'
:3 - Computed
0.1 • * Ghserved
{, 0
a �
Z IA27 2,192 2,557 2,922
Julian Day
Figure 16. Computed and Observed Time Series Concentrations of Nitrate at the Water Surface for
Four Stations
52 1111
moffati & nichol
Station 52, Surface
4
J
1
b4
E 0.3
i' 0.2
*'
- Computed
0.1
a Observed
.+ 0
Z 1,827
2,192 2,557 2,922
Julian Day
Figure 16. Computed and Observed Time Series Concentrations of Nitrate at the Water Surface for
Four Stations
52 1111
moffati & nichol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
4.2
J
C 4.I5
j 4.1
C
0 4.45
C
d 4
IA27 2,192 2,557
Julian Day
Station E2, Surface
4.2
J
C 4.I5
4.1
C
0 4.45
C
d �
1,827 2,192 2,557
Julian Day
2 ,922
�• Computed
• Observed
Station P2, Surface
2,922
�• Computed
+ observed
4.2
J
F
C 4.I5
j 4.1
C
€ 4.45
C
.g 4
1,827 2,192 2,557
Julian Day
Station H2, Surface
4.2
J
C 4.I5
3 4.1
C
0 4.45
E
d 4
I,827 2,192 2,557
Julian Day
2 ,922
- Computed
• Observed
Station S2, Surface
2 ,922
- Computed
a observed
Figure 17. Computed and Observed Time Series Concentrations of Ammonium at the Water
Surface for Four Stations
53 1111
moffall & nichol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
Station H2, Bottom
J
� U.5
E
0.4
E 0.3 f
0 4.2 — —
•
E 0.1 - – — . ;�•
E •
O • •P •
1,827 2,192 2,557 2,922
Julian Day
Station S2, Bottom
J
F a -5
E D -4
0.3
1
42 —
O 1 • • �►
E o.1
E is a
2,557 2 ,922
Julian Day
Computed
Observed
CQm p uteri
a Observed
Figure 18. Computed and Observed Time Series Concentrations of Ammonium at the Lake Bottom
for Four Stations.
54 1111
moffati & mchol
Station E2, Bottom
-j D.6
DO O.5
E ¢4
E 0.3
•
O0.2
Co-m puted
E b -1
•
• Observed
E
O
• •
1,827
2,192 2,557 2,922
1,827
Julian Day
Station H2, Bottom
J
� U.5
E
0.4
E 0.3 f
0 4.2 — —
•
E 0.1 - – — . ;�•
E •
O • •P •
1,827 2,192 2,557 2,922
Julian Day
Station S2, Bottom
J
F a -5
E D -4
0.3
1
42 —
O 1 • • �►
E o.1
E is a
2,557 2 ,922
Julian Day
Computed
Observed
CQm p uteri
a Observed
Figure 18. Computed and Observed Time Series Concentrations of Ammonium at the Lake Bottom
for Four Stations.
54 1111
moffati & mchol
Station P2, Bottom
-j
pQ U -5
E O -4
U.3
j
G 0.2
• Computed
—
E o.1
+ O b se rve d
U
s
• ■ •� •
1,827
2,192 "55r
Julian Day
Station H2, Bottom
J
� U.5
E
0.4
E 0.3 f
0 4.2 — —
•
E 0.1 - – — . ;�•
E •
O • •P •
1,827 2,192 2,557 2,922
Julian Day
Station S2, Bottom
J
F a -5
E D -4
0.3
1
42 —
O 1 • • �►
E o.1
E is a
2,557 2 ,922
Julian Day
Computed
Observed
CQm p uteri
a Observed
Figure 18. Computed and Observed Time Series Concentrations of Ammonium at the Lake Bottom
for Four Stations.
54 1111
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
1.5
J 1.25
LEO
� 1
E 0.75
0.5
0.25
0
1,a27 2,292 2,557
Julian Day
Station E2, Surface
2,922
—Cam p uteri
• Observed
Station P2, Surface
1.5
1.25
J
� 1
E 0.75 A + . •
•
05 •
0.25
0
1,827 2,292 2,557 2,922
Julian Day
Cam p uteri
• Observed
Station H2, Surface
1.5
J
-
E 1
• • ••
� 0.75
F OS M
0.25
0
1,927 2,292 2,557 2,922
Julian Day
— Computed
• Observed
Station S2, Surface
1.5
J 1.25 • •
tio 2 • •
E 0.75
0.5 •
0.25
0
1,327 2,292 2,557 2,922
Julian Day
— Computed
• Observed
Figure 19. Computed and Observed Time Series Concentrations of TKN at the Water Surface for
Four Stations
55 1111
moffati & Withal
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
Station E2, Surface
J 2
bA
E 1.5 •
a 1
4 • • • - Computed
C u.5 • Observed
« a
0
H 1,827 2,192 2,557 2,922
Julian Day
Station P2, Surface
J 2
84
E 1.5
Q1 a!
1 `
o • • • • • • Computed
+' .5
'� T Observed
ate+ 4
0
H 1,827 2,192 2,557 2,922
Julian Day
Station H2, Surface
2
J
E 1.5
G1 1 - •
O • `• •• Computed
y 0.5 • Observed
� a
a+
H 1,827 2,192 2,557 2,922
Julian Day
Station S2, Surface
J 2
E 1.5
v 1 •
o -Com puted
c • Observed
ate+ O
D
F 1,827 2,192 2,557 2,922
Julian Day
Figure 20. Computed and Observed Time Series Concentrations of TN at the Water Surface for
Four Stations.
Computed and observed time series concentrations of TP are shown in Figure 21. About half of
the model TP is refractory dissolved organic P, and approximately 30% is labile particulate
organic P. Labile dissolved organic P is about 16% of TP, and refractory particulate organic P is
about 6% of TP. Orthophosphate is a minute fraction of TP since it is taken up by phytoplankton
as soon as it is available. Some of the TP is removed by sediment burial during algal die -off and
settling periods. The observed TP data is rather constant, around 0.03 mg /L, which may have
been the detection limit. The manifestation of the detection limit may mask some of the
variability in actual TP levels. Some discrepancy in computed TP compared with observed TP
occurs near the end of 2008 when there were some large storm events that brought TP into the
model lake, a result not fully reflected in the observed data. These events are discussed further
in Section V -10 below.
56 1111
moffall & nichol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
Station H2, Surface
J 0.1
0.08
V7 0.06
2
0 0.04 Computed
-C • • • we=
M 0 b se n'e d
4
a 0 0
M 1,827 2,192 2,557 2,'322
++
4
~ Julian Day
Station S2, Surface
J o.l
E o.o8
•
oP o.06
7
0 0.04 Computed
1 •
m 0.02 r [Observed
0
s
a o
M 1,827 2,192 2,557 2,922
4
~ Julian Day
Figure 21. Computed and Observed Time Series Concentrations of TP at the Water Surface for
Four Stations
Computed and observed TOC time series concentrations at the water surface are plotted in
Figure 22 for the four stations. The model compares quite closely with observed data for this
variable, but this was not difficult to achieve after converting inputs over to version 3.6 of the
model where organic matter could be used to represent organic C while using other inputs to
handle organic N and P loadings. Most (over %) of the organic C loading to the lake was
represented as refractory, which is typically the case for watershed loadings and is also the
reason the observed and model TOC is rather constant at about 8 to 10 mg /L. Phytoplankton
production contributes to labile organic C in the model.
57 1111
moffati & nichol
Station E2, Surface
❑.1
J
100.0a
E
•
N �.
4
•
2
• •
0 0.04
- -Computed
s
„ 0
•
-02
0
+ C b se we d
1
a ❑
2,192 2,557 92"
A 1,827
2,192 2.55 2.92
a+
0
r
Julian Day
Station H2, Surface
J 0.1
0.08
V7 0.06
2
0 0.04 Computed
-C • • • we=
M 0 b se n'e d
4
a 0 0
M 1,827 2,192 2,557 2,'322
++
4
~ Julian Day
Station S2, Surface
J o.l
E o.o8
•
oP o.06
7
0 0.04 Computed
1 •
m 0.02 r [Observed
0
s
a o
M 1,827 2,192 2,557 2,922
4
~ Julian Day
Figure 21. Computed and Observed Time Series Concentrations of TP at the Water Surface for
Four Stations
Computed and observed TOC time series concentrations at the water surface are plotted in
Figure 22 for the four stations. The model compares quite closely with observed data for this
variable, but this was not difficult to achieve after converting inputs over to version 3.6 of the
model where organic matter could be used to represent organic C while using other inputs to
handle organic N and P loadings. Most (over %) of the organic C loading to the lake was
represented as refractory, which is typically the case for watershed loadings and is also the
reason the observed and model TOC is rather constant at about 8 to 10 mg /L. Phytoplankton
production contributes to labile organic C in the model.
57 1111
moffati & nichol
Station P2, Surface
0.1
J
Do
E Q.
oP 0.06
AN
7
4
•
0 0.04
Computed
-C
CL 0.02
•• •••m
0
a Clbserved
r
a 0
M 1,827
2,192 2,557 92"
a+
4
~
Julian Day
Station H2, Surface
J 0.1
0.08
V7 0.06
2
0 0.04 Computed
-C • • • we=
M 0 b se n'e d
4
a 0 0
M 1,827 2,192 2,557 2,'322
++
4
~ Julian Day
Station S2, Surface
J o.l
E o.o8
•
oP o.06
7
0 0.04 Computed
1 •
m 0.02 r [Observed
0
s
a o
M 1,827 2,192 2,557 2,922
4
~ Julian Day
Figure 21. Computed and Observed Time Series Concentrations of TP at the Water Surface for
Four Stations
Computed and observed TOC time series concentrations at the water surface are plotted in
Figure 22 for the four stations. The model compares quite closely with observed data for this
variable, but this was not difficult to achieve after converting inputs over to version 3.6 of the
model where organic matter could be used to represent organic C while using other inputs to
handle organic N and P loadings. Most (over %) of the organic C loading to the lake was
represented as refractory, which is typically the case for watershed loadings and is also the
reason the observed and model TOC is rather constant at about 8 to 10 mg /L. Phytoplankton
production contributes to labile organic C in the model.
57 1111
moffati & nichol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
Station E2, Surface
-j 12
to
E s • �.
c
O 6
Computed
m 4
u
U 2 • dbsen-ed
M a
to
8 1,827 2,142 2,557 2,922
M
Julian Day
Station H2, Surface
12
roc 1U
E • +
C a �
-E Computed
M d
U
U 2_ + C b se rve d
'C I
M
E,
`p 1,827 2,142 2,557 2 ,922
m
n Julian Day
Station P2f Surface
J 12
hq 10
E
dab • •
a
e
o
-0
Computed
n 4
U
2 2
+ O b se rve d
C
M 0
`p 1,827
2,192 2,557 2 ,922
A
Julian Day
Station H2, Surface
12
roc 1U
E • +
C a �
-E Computed
M d
U
U 2_ + C b se rve d
'C I
M
E,
`p 1,827 2,142 2,557 2 ,922
m
n Julian Day
Figure 22.Computed and Observed Time Series Concentrations of TOC at the Water Surface for
Four Stations
Computed and observed chlorophyll -a (Chl a) time series concentrations at the water surface
are plotted in Figure 23 for the four stations. Computed Chl a values are in general agreement
with the observed values on an average basis. However, Chl a values are over - predicted in
branch 3 (Station S2) due to the higher TP and phosphate computed in this branch. These over -
predictions are most likely due to loadings entering branch 3 from the Holly Springs VWVfP. An
assumption was made during model calibration that 75% of the TP released by this treatment
plant is phosphate. Later during scenario testing, it was determined that this fraction was
probably too high for the level of treatment presently at this plant and as planned for future
expansions. Most of the TP released from the treatment plant should probably be in the form of
refractory dissolved organic P, which would lower Chl a calibration results at Station S2.
However, the available phosphorous data from the treatment plant did not reflect the level of
detailed chemical species analysis necessary to characterize a more realistic split between
organic P and phosphate.
58 1/11
moffati & nichol
Station S2, Surface
J 12
10
E
E a
e
o fr
-2
Computed
n 4
a
U 2
+ O b se rve d
C
A d
0 1,827
2,192 2,557 2,9-'
M
¢
Julian Day
Figure 22.Computed and Observed Time Series Concentrations of TOC at the Water Surface for
Four Stations
Computed and observed chlorophyll -a (Chl a) time series concentrations at the water surface
are plotted in Figure 23 for the four stations. Computed Chl a values are in general agreement
with the observed values on an average basis. However, Chl a values are over - predicted in
branch 3 (Station S2) due to the higher TP and phosphate computed in this branch. These over -
predictions are most likely due to loadings entering branch 3 from the Holly Springs VWVfP. An
assumption was made during model calibration that 75% of the TP released by this treatment
plant is phosphate. Later during scenario testing, it was determined that this fraction was
probably too high for the level of treatment presently at this plant and as planned for future
expansions. Most of the TP released from the treatment plant should probably be in the form of
refractory dissolved organic P, which would lower Chl a calibration results at Station S2.
However, the available phosphorous data from the treatment plant did not reflect the level of
detailed chemical species analysis necessary to characterize a more realistic split between
organic P and phosphate.
58 1/11
moffati & nichol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
Station E2, Surface
40
30 —
A 20 • •
• Computed
s
V • * �hse rve it
• •
O •
1,627 2,192 2,557 2,922
Julian Day
Figure 23. Computed and Observed Time Series Concentrations of Chi a at the Water Surface for
Four Stations
Computed phytoplankton biomass (dry weight basis) concentrations versus time are plotted in
Figure 24 for the three algal groups for the water surface at the dam. There are no observed
data to accompany these results, but the results serve to show the magnitudes of phytoplankton
biomass and the seasonality of blooms for the three groups. The three groups bloom within their
appropriate seasons, where diatoms dominate in the winter to early spring, and the other two
groups dominate in the other seasons when the water is warmer and there is more sun light.
Overall, the magnitude of the biomass in the blooms is reasonable.
59 1111
moffati & nichol
Station P2, Surface
35
30
25
20
•
25 •
• A Jim
Compute d
� 20
5
• • a Observed
0
•
1,827
2,192 2,557 2,922
Julian Day
Figure 23. Computed and Observed Time Series Concentrations of Chi a at the Water Surface for
Four Stations
Computed phytoplankton biomass (dry weight basis) concentrations versus time are plotted in
Figure 24 for the three algal groups for the water surface at the dam. There are no observed
data to accompany these results, but the results serve to show the magnitudes of phytoplankton
biomass and the seasonality of blooms for the three groups. The three groups bloom within their
appropriate seasons, where diatoms dominate in the winter to early spring, and the other two
groups dominate in the other seasons when the water is warmer and there is more sun light.
Overall, the magnitude of the biomass in the blooms is reasonable.
59 1111
moffati & nichol
Station H2, Surface
35
30
25
20
•
m 15
— Computed
•
:E 10
5
+ Chserved
0
•
1,827
2,192 2,557 2 ,922
Julian Day
Figure 23. Computed and Observed Time Series Concentrations of Chi a at the Water Surface for
Four Stations
Computed phytoplankton biomass (dry weight basis) concentrations versus time are plotted in
Figure 24 for the three algal groups for the water surface at the dam. There are no observed
data to accompany these results, but the results serve to show the magnitudes of phytoplankton
biomass and the seasonality of blooms for the three groups. The three groups bloom within their
appropriate seasons, where diatoms dominate in the winter to early spring, and the other two
groups dominate in the other seasons when the water is warmer and there is more sun light.
Overall, the magnitude of the biomass in the blooms is reasonable.
59 1111
moffati & nichol
Station 52, Surface
35
30
25
� 20
m 15
IND Computed
X 1p
5
■ • Observed
•
1,827
2,192 2,557 2,9?'
Julian Day
Figure 23. Computed and Observed Time Series Concentrations of Chi a at the Water Surface for
Four Stations
Computed phytoplankton biomass (dry weight basis) concentrations versus time are plotted in
Figure 24 for the three algal groups for the water surface at the dam. There are no observed
data to accompany these results, but the results serve to show the magnitudes of phytoplankton
biomass and the seasonality of blooms for the three groups. The three groups bloom within their
appropriate seasons, where diatoms dominate in the winter to early spring, and the other two
groups dominate in the other seasons when the water is warmer and there is more sun light.
Overall, the magnitude of the biomass in the blooms is reasonable.
59 1111
moffati & nichol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
Blue Greens
J
1.5
1
w
r�
� u.5
.4
m
0
1/1/06 6/30/06 12/27/06 6/25/07 12/22/07 6/19/08 12/16/08
Figure 24. Time Series of Computed Phytoplankton Biomass on the Surface at the dam for the
Three Algal Groups
60 1r1i1
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
10. Discussion of Calibration Results
The results of the calibration for nutrients and algal productivity, particularly TP, and
chlorophyll -a show increased levels in predicted values during the 2007 and 2008 model years
that are not necessarily reflected in the recorded data. The model predictions at Station S2
diverge (over - predicted) from observed Chl -a values in November 2007 (Julian Day 2500) and
again in the spring of 2008 until the end of the 2008.
The over - predicted Chl -a during the latter part of 2008 can be attributed, in part, to two tropical
weather systems that passed through central North Carolina in late August and early September
2008. On August 27 -28, 2008, the remnants of Tropical Storm Faye, which had routed out of the
Gulf of Mexico, passed across central North Carolina resulting in storm event rainfall totals of 4-
6 inches in the vicinity of Harris Lake. Shortly after that event, on September 6, Tropical Storm
Hanna came ashore near the North Carolina - South Carolina border and tracked north -
northwest through central North Carolina resulting in storm event rainfall totals of 5 -7 inches
with higher localized precipitation thought a substantial portion of the Piedmont including the
Harris Lake watershed (meteorological accounts adapted from the NOAA Raleigh National
Weather Service Forecast Office at http://www.erh.noaa.gov/rah/events/).
Comprehensive analysis of the contributing factors showed that increased nutrient and chl -a
levels in the last year can also be attributed to the nutrient inputs for the Holly Springs WWTP.
As detailed in Section IV -2, the total nitrogen (TN), total phosphorus (TP) and total flow rate
inputs were developed from discharge monitoring reports (DMR(s)) provided by the municipality.
A total mass load annual summary of TN and TP was generated for each calendar year of
analysis from 2001 to 2008. The mass loads for the Holly Springs WWTP were then aggregated
with total mass load values from all other nutrient sources and compared on a percentage
contribution basis. DMR data show that the service area experienced steady growth through
2001 -2008, as evidenced by the increasing flow rates. Reported TN and TP values do not
increase steadily, but rather show a substantial increase in the years of 2006 -2008. From 2001-
2005, the average annual mass load of TP was 0.35 metric tons, compared to an average
annual mass load of 1.62 metric tons during the span of 2006 -2008. Subsequently, the HS
WWTP percentage contribution compared to all nutrient sources more than tripled (from -11 %
to -35 %) for the 2006 -2008 period of analysis.
It stands to reason that the four fold increase in total annual TP mass load from 2001 -2005 to
the 2006 -2008 period is responsible for the predicted increases seen in Chl -a during the last two
years of analysis. The higher loadings from HS WWTP seem to inhibit the model predictions,
from cycling to seasonably low levels of chl -a in 2007 and 2008. This influence can be seen
more clearly at station S2, which is downstream from the HS WWTP.
However, it is important to note that the increased P levels predicted in the model in response to
increases in the Holly Springs WWTP load and the spikes resulting from tropical weather events
61 1/11
mofiati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
are not manifested in the actual TP data from the lake. This dampening of the actual response
in the real -world reservoir may be resulting from the reductions occurring during transport and
delivery of TP loads as discussed in Section IV -4.
Despite the divergence near the end of the calibration period, overall, the calibration shows
good agreement between computed and observed results and is considered satisfactory for
making projections about future lake water quality with a raised pool and different loading
conditions.
62 1111
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
VI. Future Scenarios
Three future scenarios were developed for evaluation with the recalibrated W2 model, and all
three scenarios were run with the same hydrologic base conditions (rainfall and tributary
inflows) as the 2001 -2008 period reflected in the calibration of the water balance. Specifically,
all future scenarios rely on the tributary flows and precipitation data files developed by
HydroLogics, Inc., for input into the OASIS model for the 2001 -2008 period described previously
in Section IV -1. One scenario examines predicted water quality conditions in the absence of the
expansion of the Shearon Harris plant and without the associated expansion of the Lake, and
two scenarios examine predicted water quality conditions in conjunction with different
operational regimes associated with an expanded Shearon Harris plant with an expanded
reservoir and make -up water pumped from the Cape Fear River
The detailed descriptions of each of the scenarios are as follow:
• Existing Lake — Future Loads: This scenario examines the predicted water quality
conditions for the existing Lake, with current reservoir pool elevation and storage volume
in conjunction with future nonpoint source load conditions and projected future discharge
conditions at the Holly Springs VW TP. The purpose of this scenario was to evaluate the
hypothetical future conditions that might occur in the absence of the expansion of the
HNP and without the associated expansion of the Lake. No minimum release form Harris
Lake is considered in this scenario.
• Future Regime A — This scenario examines all conditions associated with the expansion
of the HNP in conjunction with the expanded Lake, future nonpoint source loads, and
projected future conditions at the Holly Springs VW TP. The operational scheme for
expanded HNP includes the addition of make -up water with no pumping when flows in
the Cape Fear River are less than 600 cfs.
• Future Regime B — This scenario examines all conditions associated with the expansion
of the HNP just as Future Scenario A, but with the operational scheme for expanded
HNP including make -up water with no pumping when flows in the Cape Fear are less
than 700 cfs.
Both Future Regimes A and B include the same pattern of minimum releases from the Harris
Lake dam, developed in a collaborative effort involving PEC, its consultants, and regulatory and
resource management agencies. This was to ensure minimum stream flows and protect aquatic
habitat conditions in Buckhorn Creek below the dam and the reach of the Cape Fear River
between the make -up water intake and the confluence of Buckhorn Creek. The daily time series
for the minimum release strategy and the schedules for pumping of Cape Fear River make -up
water for the two pumping schemes were both output from the OASIS model and obtained from
HydroLogics, Inc.
63 1111
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
It should be noted that the Holly Springs discharge is treated as being fully delivered to the lake,
with no external load reductions in the future scenarios (refer to Section 2) due to the fact that
the discharge point is being moved closer to the lake, and as such, it will no longer benefit from
the reductions occurring during transport to the Lake. However, in the future state of the
expanded lake, the secondary impoundment, or forebay, formed by the road crossing on the
upper Utley Creek arm will still exist, and is likely to continue resulting in some effective
reduction of the nutrient loads from Holly Springs as they enter the lake.
For all future scenarios, the same 80% reduction scaling factor was applied to the future
nonpoint source phosphorus loads generated by GWLF that was applied for calibration of
existing conditions (refer to Section 4).
1. Time Series Results from Future Scenarios
Predicted water quality results from the three future scenarios described above are illustrated in
the following time series plots. For each water quality parameter, the time series corresponding
to existing conditions (existing lake — existing loads) for 2001 -2008 is also included in the plot to
examine the predicted degree of change associated with each future scenario. Results for each
parameter are shown for the W2 model segment 13, located nearest to the dam and
corresponding to observed data station E2, and model segment 5, corresponding to observed
data station P2 (refer to Figure 8 for an illustration of model segments and data stations). Model
segment 5 is the segment at the upper end of the main arm of the Lake where the upper arms
all come together. Predicted total nitrogen (TN), total phosphorous (TP), chlorophyll a, surface
layer dissolved oxygen (DO), and bottom layer DO are illustrated in the figures below.
Nitrogen
The time series for TN in Figure 25 and Figure 26 show predicted increases in TN
concentrations for all future scenarios relative to existing conditions with TN levels exceeding 1
mg /I at the dam and in segment 5 for Future Scenarios A and B toward the end of the simulation
period. Beyond showing some overall increase, neither location shows a substantial difference
between Future Scenarios A and B with regard to in -lake TN concentrations.
64 1111
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
Figure 25. Comparisons for Surface TN at the Dam
Comparisons for Surface TN at the Diana
1.4
1.2
1
J
f�
-0.8
E
I
2
F
6.4
Existing Lake, Existing Loads
o.6
2
Existing Lake, Future Lads
7V=,?-
— — Future Conditions, Regime A
Future Conditions, Regime B
0.4 -
O
0 504 1006 15oC 2000 2500 30DC
Existing Lake, Existing Loads
Julian Day
Figure 25. Comparisons for Surface TN at the Dam
Figure 26. Comparisons for Surface TN at Segment 5
Phosphorus
The time series for TP in Figure 27 and Figure 28 show predicted increases in TP for all future
scenarios relative to existing conditions. In all scenarios, the year -to -year patterns of fluctuation
in TP levels are consistently repeated, indicating that the hydrology of the watershed and the
reservoir are a key factor in determining TP levels. As with the TN time series, future loadings
65 1111
moffati & nichol
Comparisons for Surface TN at Segment 5
1.2
0A
J
to
to
E
o.6
7V=,?-
0.4 -
Existing Lake, Existing Loads
02
Existing Lake, Future Loads
— — Future Conditions, Regime A
– – Future Conditions, Regime B
0
0 500 1Ccc 1500 _C�c 2506 3000
Julian Day
Figure 26. Comparisons for Surface TN at Segment 5
Phosphorus
The time series for TP in Figure 27 and Figure 28 show predicted increases in TP for all future
scenarios relative to existing conditions. In all scenarios, the year -to -year patterns of fluctuation
in TP levels are consistently repeated, indicating that the hydrology of the watershed and the
reservoir are a key factor in determining TP levels. As with the TN time series, future loadings
65 1111
moffati & nichol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
cause an increase in TP concentrations, and introducing make up water from the Cape Fear
River results in additional increases in TP concentrations.
Comparisons for Surface TP at the Dam
0.17
Existing Lake, Existing Loads
— Existing Lake, Future Loads
0.10 — — Future Conditions, Regime A
......• Future Conditions, Regime B
0.08
r.
J '
m � r
E 0.06 U JA
F �f
0.04
0.02
0.00
0 500 1000 1500 2000 2500 3000
Julian Day
Fininra 97 f_mmnaricnnc fnr SIIrFnra TP a4 4ha rlam
rigure zts. comparisons Tor 5urrace i P at Segment 5
Chlorophyll -A
The Chlorophyll -a time series for all three future scenarios illustrated in Figure 29 and Figure 30
show increases in chlorophyll -a relative to existing conditions. Peak values spike to levels
slightly greater than 23 µg /I near the end of the simulation period at both locations. However,
66 1 /1Iq
mofiatt & nichol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
predicted chlorophyll -a concentrations are less than 20 µg /I most of the time. Also, the scenario
of existing lake — future loads exhibits predicted chlorophyll -a concentrations similar to those of
future lake, indicating that the additional nutrient concentrations associated with the make -up
water would not cause substantially different algal concentrations in the expanded reservoir.
Figure 29. Comparisons for Surface Chi a at the Dam
Dissolved Oxygen
Figure 30. Comparisons for Surface Chi a at Segment 5
67 1111
moffati & nichol
Comparisons for Surface Chia at Segment 5
ZS
Existing Lake, Existing Loads
Existing Lake, Future Loads
— — Future Conditions, Regime A
20
...... Future Conditions. Regime 6
ti
15
E
10
5
r wq 1
0
0
500 1 C X 1506 260D 250D 3000
Julian Day
Dissolved Oxygen
Figure 30. Comparisons for Surface Chi a at Segment 5
67 1111
moffati & nichol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
Dissolved oxygen time series are shown for the water surface layer in Figure 31 and Figure 32
and for the bottom layer in Figure 33 and Figure 34. The surface time series indicate that the
various scenarios cause very little change in the seasonal fluctuation of DO levels at the
surface. Surface DO levels at the dam show a greater frequency and magnitude of downward
spikes in surface DO during late summer to early fall than predicted levels in Segment 5. Such
spikes are also more frequent and greater magnitude for the three scenarios compared with
existing conditions. However, it should be noted that the downward spikes are generally less
frequent and less severe in the expanded reservoir under Future Scenarios A and B, relative to
those predicted to occur in the Existing Lake — Future Loads scenario. Downward DO spikes
can be caused by lake turn -over when anoxic hypolimnetic water is mixed with epilimnetic water
causing a drop in surface DO. Also, surface DO can drop as a result of diurnal fluctuations
caused by algal respiration during night hours. Most of the variation in surface DO is due to
seasonal variations in water temperature.
Bottom DO time series show that future scenarios result in a very slight increase the predicted
duration of anoxic periods during warm seasons, and that such periods of anoxia are more
pronounced at the dam than in segment 5. These results are not unexpected, given the
increased nutrient concentrations and algal productivity predicted in conjunction with future
conditions. The higher incidence of bottom anoxia at the dam is also partly a function of greater
water depth in that vicinity for future lake conditions.
ComParisonsfor Surface DO at the Dam
14
10
J
aU S
E
0 6
d
Existing Lake, Existing Laads
2 Existing Lake, Future Loads
Future Conditions, Regime A
• • • • Future Conditions, Regime 8
O 1 -
0 500 low 1500 2600 2500 HC
Julian Day
Figure 31. Comparisons for Surface DO at the Dam
6s „„
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
Figure 32. Comparisons for Surface DO at Segment 5
Comparisons for Surface DO at Segment 5
Existing Lake, Existing Loads
14
Comparisons for Bottom DO at the Dam
12 -
A
A
1¢
18
A h
••. • • . Future Conditions, Regime &
12
1
$
E
O
10
0
5
J
119
a
E
Existing Lake, Existing Leads
2
Existing Lake, Future Loads
— Future Conditions, Regime A
•••... Future Conditions, Regime 0
0
tu
0 500 1000 153C 2500 300C
Julian Day
Figure 32. Comparisons for Surface DO at Segment 5
Figure 33. Comparisons for Bottom DO at the Dam
69 1111
moffati & nichol
Existing Lake, Existing Loads
Comparisons for Bottom DO at the Dam
Existing Lake, Future Loads
1¢
— — Future Conditions, Regime A
••. • • . Future Conditions, Regime &
12
10
J
119
E
n�
tu
a
I
x
I
-A
h
o
0
500 low 1500 2000
2500 3000
Julian Day
Figure 33. Comparisons for Bottom DO at the Dam
69 1111
moffati & nichol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
Figure 34. Comparisons for Bottom DO at Segment 5
2. Analysis of Time Series Predictions
The following sections describe analyses that are offered to provide insights into the key factors
driving the results shown in the time series predictions.
Mass Load Summary
To better understand the effects that changes in nutrient loading were having on model
predictions, the average annual mass loads from the various primary sources which were input
into the model were calculated for existing and future scenarios, and the results are shown
below in Table 6 to Table 8. It should be noted that the nonpoint source loads in the tables are
reflective of the W2 tributary inputs after the reductions discussed in Section V so they could be
evaluated in terms of their relative impact on model predictions. The loads associated with the
Existing Lake — Future Loads scenario are not shown as a table because they are identical to
those presented Table 7 and Table 8 for Future Scenarios A and B except that there is no load
from the Cape Fear River.
70 1111
moffati & nichol
Existing Lake, Existing Loads
Comparisonsfor Bottom DO at Segment 5
Existing Lake, Future Loads
14
— — Future Conditions, Regime A
Future Conditions, Regime B
12
10
J
E
O'
fl
S
1
4
i
r i i
2
I
0 500 1000 1500 2000
2500 3000
Julian Day
Figure 34. Comparisons for Bottom DO at Segment 5
2. Analysis of Time Series Predictions
The following sections describe analyses that are offered to provide insights into the key factors
driving the results shown in the time series predictions.
Mass Load Summary
To better understand the effects that changes in nutrient loading were having on model
predictions, the average annual mass loads from the various primary sources which were input
into the model were calculated for existing and future scenarios, and the results are shown
below in Table 6 to Table 8. It should be noted that the nonpoint source loads in the tables are
reflective of the W2 tributary inputs after the reductions discussed in Section V so they could be
evaluated in terms of their relative impact on model predictions. The loads associated with the
Existing Lake — Future Loads scenario are not shown as a table because they are identical to
those presented Table 7 and Table 8 for Future Scenarios A and B except that there is no load
from the Cape Fear River.
70 1111
moffati & nichol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
Table 6. Existing Conditions Average Annual Mass Loads
Table 7. Future Regime A Average Annual Mass Loads
Average Annual Mass Loads
(METRIC TONS) - Existing
Conditions
Average Annual Mass Loads (%
TOTAL) - Existing Conditions
Total Nitrogen
(TN)
Total Phosphorus
(TP)
Total Nitrogen
(TN)
Total
Phosphorus (TP)
INFLOW SOURCE
Non -Point Sources
87.99
2.96
92%
78%
Holly Springs WWTP
7.55
0.83
8%
22%
TOTAL
1 95.59
3.80
100%
100%
Table 7. Future Regime A Average Annual Mass Loads
Table 8. Future Regime B Average Annual Mass Loads
Average Annual Mass Loads
(METRIC TONS) - Future Regime A
Average Annual Mass Loads (%
TOTAL) - Future Regime A
Total Nitrogen
(TN)
Total Phosphorus
(TP)
Total Nitrogen
(TN)
Total
Phosphorus (TP)
INFLOW SOURCE
Non -Point Sources
103.79
2.97
53%
24%
Holly Springs WWTP
42.29
4.23
22%
34%
Cape Fear River - Regime A
49.60
5.09
25%
41%
TOTAL
195.74
12.30
100%
100%
Table 8. Future Regime B Average Annual Mass Loads
71 „„
moffati & mchol
Average Annual Mass Loads
(METRIC TONS) - Future Regime B
Average Annual Mass Loads
(% TOTAL) - Future Regime B
Total Nitrogen
(TN)
Total Phosphorus
(TP)
Total Nitrogen
(TN)
Total
Phosphorus
(TP)
INFLOW SOURCE
Non -Point Sources
103.79
2.97
53%
24%
Holly Springs WWTP
42.29
4.23
22%
34%
Cape Fear River - Regime B
49.46
5.06
25%
41%
TOTAL
195.60
12.27
100%
100%
71 „„
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
Most importantly, Table 7 and Table 8 show that the future scenarios result in significant
increases in nutrient loading to the lake from all sources. With the introduction of make -up water
from the Cape Fear River under Future Regime A and B, the TP load is more than triple that of
existing conditions on average during the simulation period. However, the stage- storage
relationship of the lake shows that the existing reservoir pool volume of approximately 65,000
acre -feet will be increased to a pool volume of more than 179,000 acre -feet when the storage
elevation is raised to 240 feet. The expanded reservoir will have a pool volume 2.75 times the
existing storage volume. Obviously, from a mass and volume perspective, the increased
loading, when considered relative to the increase in storage volume, cannot alone account for
the increases in predicted TP concentrations for future scenarios, which are shown in Figure 27
and Figure 28 to be approximately double those of existing conditions at times. For this reason,
factors affecting hydrology and residence time are examined below.
3. Hydrology and Residence Time (Water Age)
Analysis of lake water residence time was conducted using water age, which is a state variable
that can be simulated in W2 by using one of the general constituent options. The existing
conditions and future Regime A scenarios were run with water age activated to compare the
water residence times for the existing lake /pool with the future lake /pool.
The two water age results are shown in Figure 35 for existing and future lake conditions as a
time series at the water surface in Segment 13, the segment at the dam. Water ages are
substantially higher for most of the simulation for the future lake conditions. The model -
computed water age for existing conditions fluctuates between 200 -500 days for most of the 8-
year simulation, and between 400 -700 days for the future conditions.
The water age increases substantially for existing lake conditions during the last year (2008) of
the simulation because the daily average inflow (non -point and point) for 2008 is 1.29 cros, or
about 60% of the 8 -year daily average inflow of 2.16 cros. Thus, the residence time should
increase during 2008 as the model water age results indicate. The water age decreases slightly
during 2008 compared to the previous year for the future lake conditions because the total daily
average inflow (including pumped CFR water) for 2008 is 3.66 cros, compared to the 8 -year
daily average inflow of 3.49 cros. Thus, the residence time should drop some during 2008, as
verified by the water age results.
72 1111
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
Figure 35. Comparison of Computed Water Age at the Surface and at the Dam for Existing and
Future Conditions
Outside of the modelling analysis of water age, average residence times were calculated for the
lake to compare existing and future conditions. An average residence time of 696 days was
computed for the future conditions. This computation was based on dividing the future lake
volume, 2.1 E8 m3, by the daily average inflow of 3.49 cros. This inflow included the six tributary
inflows, the two point source inflows, and the pumped flows from the CFR.
A similar calculation was conducted for the existing pool elevation of 220 feet, which has a lake
volume of 7.7E7 m3„ or about a third of the future volume. The total daily average inflow rate for
the existing conditions is 2.16 cros, without the CFR pumped inflows. This calculation yields an
average residence time for the existing lake conditions of 412 days.
Close examination of the timing of increases in water age (Figure 35) in conjunction with the
timing of increased in TP levels within the lake (Figure 27 and Figure 28) and the timing of
increases in algal growth via chlorophyll a (Figure 29 and Figure 30) indicate that periods of
increased nutrient loads and eutrophication in the model prediction track to some degree with
water age. Residence time can be a key factor driving model prediction in Harris Lake. This is
logical considering that the long -term, steady -state lake concentration of a conservative
constituent is the constituent loading rate (mass /time) divided by the long -term average total
flow rate (volume /time) through the lake. Theoretical residence time is the lake volume divided
by the long -term average total lake flow. Thus, lake phosphorus concentration should directly
correlate to lake water residence time since both are inversely proportional to flow rate through
the lake.
73 1 /1Ill
mofiatt & nichol
Water Age Comparison
90❑
800
v,
a
700
60❑
Q
500
AN
"
400
W
300
200
• Future conditions, regime A
100
Existing conditions
0
0 500 1000 15M) 2000 2500 3000
Julian Day
Figure 35. Comparison of Computed Water Age at the Surface and at the Dam for Existing and
Future Conditions
Outside of the modelling analysis of water age, average residence times were calculated for the
lake to compare existing and future conditions. An average residence time of 696 days was
computed for the future conditions. This computation was based on dividing the future lake
volume, 2.1 E8 m3, by the daily average inflow of 3.49 cros. This inflow included the six tributary
inflows, the two point source inflows, and the pumped flows from the CFR.
A similar calculation was conducted for the existing pool elevation of 220 feet, which has a lake
volume of 7.7E7 m3„ or about a third of the future volume. The total daily average inflow rate for
the existing conditions is 2.16 cros, without the CFR pumped inflows. This calculation yields an
average residence time for the existing lake conditions of 412 days.
Close examination of the timing of increases in water age (Figure 35) in conjunction with the
timing of increased in TP levels within the lake (Figure 27 and Figure 28) and the timing of
increases in algal growth via chlorophyll a (Figure 29 and Figure 30) indicate that periods of
increased nutrient loads and eutrophication in the model prediction track to some degree with
water age. Residence time can be a key factor driving model prediction in Harris Lake. This is
logical considering that the long -term, steady -state lake concentration of a conservative
constituent is the constituent loading rate (mass /time) divided by the long -term average total
flow rate (volume /time) through the lake. Theoretical residence time is the lake volume divided
by the long -term average total lake flow. Thus, lake phosphorus concentration should directly
correlate to lake water residence time since both are inversely proportional to flow rate through
the lake.
73 1 /1Ill
mofiatt & nichol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
To better understand factors affecting the residence times in the W2 model, particularly with
regard to future scenarios, annual average rainfall totals and make -up water pumping totals are
summarized in Table 9 below. In general, the data illustrate that dry years lead to noticeable
increases in residence time. Closer inspection of the fluctuation of hydrologic inputs in
conjunction with increases in predicted chlorophyll a indicate that the year with highest algal
concentrations in the future scenarios, the last year, occurs after the driest year (hydrologic year
2007) in which pumping from the Cape Fear was curtailed to a minimum. Then in 2008, rainfall
and stream flows increase, delivering increased non -point source loads and the highest volume
of any of the eight years in the simulation period is pumped from the Cape Fear River. As a
result, the W2 model predicts a year of higher algal concentrations. However, despite this
combination of worst -case scenario conditions, chlorophyll a levels remain below 25 µg /I near
the dam, even with the sharp upward spikes induced by tropical storm events.
Table 9. Hydrologic Factors Affecting Lake Residence Time
MODEL
JULIAN
DAY
JULIAN
DAY
REGIME A PUMP
VOLUME
REGIME B PUMP
VOLUME
RAINFALL TOTAL
YEAR
BEGIN
END
ANNUAL AVG MGD
ANNUAL AVG MGD
CENTIMETERS /YEAR
2001
1
365
25
26
81.36
2002
366
730
49
49
110.82
2003
731
1095
12
12
127.68
2004
1096
1461
29
28
88.22
2005
1462
1826
32
30
77.81
2006
1827
2191
35
36
112.49
2007
2192
2556
6
6
75.89
2008
2557
2922
55
55
99.47
Average
30.35
30.27
96.72
74 1111
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
VII. Summary & Conclusions
1. Background and Objectives
The Harris Advanced Reactor (HAR) project consists of adding two new nuclear power
generation units at the existing Shearon Harris Site. The net consumptive water usage for the
two new reactors will be approximately 42 million gallons per day (MGD). Having a sufficient
sustained yield from Harris Lake to support the consumptive use of cooling water and meet
minimum flow requirements below Harris Lake Dam will require an expansion of Harris Lake
from the current pool elevation of 220 feet to 240 feet. It will also require pumping of up to 86
MGD in makeup water from the Cape Fear River.
Basin -Wide Water Quality Assessment and Lake Assessment documents from North Carolina
Department of Water Quality (NCDWQ) have consistently noted that water quality is typically
good in Harris Lake, relative to other Piedmont reservoirs. Conversely, the Cape Fear River
water has higher concentrations of nutrients and other pollutants relative to Harris Lake and its
tributaries. When substantial volumes of Cape Fear River water are introduced to the Lake there
is a potential to impact water quality.
To support issuance of the 401 Water Quality Certification for the expansion at the Shearon
Harris Site, the CE- QUAL -W2 water quality model previously developed for Harris Lake was
updated and recalibrated for use in predicting future water quality conditions. The model was
used to simulate the combined effects of a higher pool, larger heat load for power plant cooling,
release targets, and introduction of makeup water from the Cape Fear River. The W2 model of
Harris Lake was supported by a Generalized Watershed Loading Function (GWLF) model that
predicted pollutant loads from stormwater runoff and other non -point sources throughout the
Lake's watershed. The GWLF model was also updated to reflect the most current land uses
within the urbanizing watershed.
2. GWLF Model Update
An initial GWLF model of Utley Creek was developed by Tetra Tech in 2004 and was revised by
CH2M HILL to include Harris Lake. Soil properties were adjusted and land use (future and
existing) was taken into account in the revised model. Meteorological data from the Progress
Energy meteorology station was used. Further details can be found in the CH2M HILL report
(CH2M HILL, 2009).
As part of the current effort, Moffatt & Nichol refined the CH2M HILL model's soil properties,
land use distributions, and meteorological inputs. For the land use update, parcels in the basin
were identified using the most recent county parcel and zoning GIS layers from Wake,
Chatham, and Harnett Counties. Wake County GIS layers were utilized to identify land use,
number of buildings and type of buildings for each parcel. The aggregate land area attributed to
each use category (Forestry, Pasture, Row Crop, Barren, Urban Green Space, Wetlands,
75 1/11
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
Water, Residential - very high to very low, Office /Light Industrial, Commercial /Heavy Industrial)
was assigned an appropriate curve number and pollutant loading coefficient in GWLF. All areas
not included as tax parcels were considered to be roadway /right -of -way and were assigned the
curve number for a land use category with similar impervious area percentage (Office /Light
Industrial). Each parcel was grouped geographically into one of 17 GWLF sub - watersheds for
modeling purposes. Future land use estimates were developed using the maximum build -out
condition based on current zoning assignments.
The results of the GWLF are summarized below in Table 10, which shows that the highest
predicted loads occur during wet years, and the lowest predicted loading occurs during dry
years, as would be expected. It should be noted that, while results vary from year to year, on
average the GWLF model predicted no increase in non -point source phosphorous loading for
the future land use scenario. Decreases in overall nutrient loads are not unusual when
significant amounts of row crop and /or pasture land are converted to medium to low density
residential uses (which have lower nutrient loading rates), as is projected to occur in the Harris
Lake watershed.
Table 10. Comparison of TN and TP for Existing and Future Conditions by Year
3. CE- QUAL -W2 Model Update and Recalibration
Many updates and improvements were made to the model inputs, including: numerical grid
structure; tributary discharge rates; precipitation; dam outflow discharge rates; tributary inflow
temperatures; point source loadings; non -point source loadings; cooling water return flow
temperatures and water quality; meteorology; and various calibration parameters. Additionally,
a much later version of the model code was used (version 3.6) rather than using the older
version (version 3.2) that was used in the previous modeling of Harris Lake. The newer version
allowed better representation of organic forms of nutrients.
76 1111
moffati & mchol
TN (lb /ac)
TP (lb /ac)
Year
Precipitation (in)
Existing
Future
Existing
Future
2001
31.82
3.31
4.22
0.56
0.61
2002
42.61
5.62
7.40
0.88
1.05
2003
46.11
5.61
6.16
0.94
0.88
2004
36.32
3.53
3.62
0.64
0.53
2005
28.34
2.31
2.68
0.38
0.38
2006
47.45
6.37
7.02
1.08
1.01
2007
30.98
3.69
4.69
0.58
0.66
2008
40.98
4.58
4.78
0.78
0.68
Average
38.08
4.38
5.07
0.73
0.73
3. CE- QUAL -W2 Model Update and Recalibration
Many updates and improvements were made to the model inputs, including: numerical grid
structure; tributary discharge rates; precipitation; dam outflow discharge rates; tributary inflow
temperatures; point source loadings; non -point source loadings; cooling water return flow
temperatures and water quality; meteorology; and various calibration parameters. Additionally,
a much later version of the model code was used (version 3.6) rather than using the older
version (version 3.2) that was used in the previous modeling of Harris Lake. The newer version
allowed better representation of organic forms of nutrients.
76 1111
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
Pool elevations computed by the revised model for the period 2001 — 2008 were compared to
observed pool elevations to verify that the model was accurately representing the correct water
balance. The revised model was then recalibrated for the years 2006 through 2008 for
temperature, various forms of nitrogen, total phosphorus, dissolved oxygen, algal chlorophyll a,
and total organic carbon. The results of the recalibration were subjected to an independent
peer review performed by HDRIHydroQual. The recommendations of the peer review were
used to further refine the model recalibration, resulting in greater improvement to model
accuracy relative to the initial effort by M &N.
4. Future Scenarios and Results
Three future scenarios were developed for evaluation with the recalibrated W2 model, and all
three scenarios were run with the same hydrologic base conditions. The three future scenarios
were:
• Existing Lake — Future Loads: This scenario examined the predicted water quality
conditions for the existing Lake, with current reservoir pool elevation and storage volume
in conjunction with future nonpoint source load conditions and projected future discharge
conditions at the Holly Springs WWTP.
• Future Regime A — This scenario examines all conditions associated with the expansion
of the HAR in conjunction with the expanded Lake, future nonpoint source loads, and
projected future conditions at the Holly Springs WWTP. The operational scheme for
expanded HAR includes the addition of make -up water with no pumping when flows in
the Cape Fear River are less than 600 cfs.
• Future Regime B — This scenario examines all conditions associated with the expansion
of the HAR just as Future Regime A, but with the operational scheme for expanded HAR
including make -up water with no pumping when flows in the Cape Fear are less than
700 cfs.
Both Future Regimes A and B include the same pattern of minimum releases from the Harris
Lake dam. Each scenario was run with the revised and recalibrated model for the period 2001 —
2008. Additionally, the revised and recalibrated model was run for the same eight year period
with existing lake conditions and existing (present day) loadings; thus, this run was merely a
five -year extension of the recalibration run. The results of the three future scenarios were
plotted together as time series (concentration versus time) with the results for the existing lake —
existing loads condition for comparison. Results were compared for the segment adjacent to
the dam and in segment 5, which corresponds to the area around observation station P2.
The model predicts increases in total nitrogen (TN) and total phosphorus (TP) concentrations for
all future scenarios relative to existing conditions. The model shows there will be little difference
between Future Scenarios A and B with regard to in -lake TN and TP concentrations. Increases
77 1/11
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
in chlorophyll -a are predicted for all future scenarios relative to existing conditions. Peak values
spike to levels slightly greater than 23 µg /I near the end of the simulation period at both
locations. However, predicted chlorophyll -a concentrations are less than 20 µg /I most of the
time. Chlorophyll -a prediction for the various scenarios are shown in Figures Figure 36.
Comparisons for Surface Chl a at the Dam and Figure 37. Comparisons for Surface Chl a at
Segment 5. The scenario of existing lake — future loads exhibits predicted chlorophyll -a
concentrations similar to those of future lake, indicating that the additional nutrient
concentrations associated with the make -up water would not cause substantially different algal
concentrations in the expanded reservoir.
Figure 36. Comparisons for Surface Chi a at the Dam
7s „„
moffati & nichol
Comparisons for Surface Chl a at the Dam
25
Existing Lake, Existing Loads
Existing Lake, Future Loads
20
— — Future Conditions, Regime A
- • • • • Future Conditions, Regime 6
II
_ 15
A
U 10
5
0
0 5vv 1000 _5 or 2400 2500 3oco
Julian Day
Figure 36. Comparisons for Surface Chi a at the Dam
7s „„
moffati & nichol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
Figure 37. Comparisons for Surface Chi a at Segment 5
The model shows that the various scenarios cause very little change relative to each other in
the seasonal fluctuation of dissolved oxygen (DO) levels at the water surface (Figure 38.
Comparisons for Surface DO at Segment 5). Model results for bottom DO show that future
scenarios result in a very slight increase the predicted duration of anoxic periods during warm
seasons, and that such periods of anoxia are more pronounced at the dam (Figure 39Figure 39)
than in segment 5. These results are not unexpected given the increased nutrient
concentrations and algal productivity predicted in conjunction with future conditions. The higher
incidence of bottom anoxia at the dam is also partly a function of greater water depth in that
vicinity for future lake conditions.
79 1111
moffati & nichol
Comparisons for SLirface Chi, a at Segment 5
25
Existing Lake, Existing Loads
Existing Lake, Future Loads
— — Future Conditions, Regime A
20
Future Conditions, Regime
15
—
to
CSJ 1a
5
-Y-r v T
a
1-
0
5CC 1090 1500 2000 2500 3000
Julian Day
Figure 37. Comparisons for Surface Chi a at Segment 5
The model shows that the various scenarios cause very little change relative to each other in
the seasonal fluctuation of dissolved oxygen (DO) levels at the water surface (Figure 38.
Comparisons for Surface DO at Segment 5). Model results for bottom DO show that future
scenarios result in a very slight increase the predicted duration of anoxic periods during warm
seasons, and that such periods of anoxia are more pronounced at the dam (Figure 39Figure 39)
than in segment 5. These results are not unexpected given the increased nutrient
concentrations and algal productivity predicted in conjunction with future conditions. The higher
incidence of bottom anoxia at the dam is also partly a function of greater water depth in that
vicinity for future lake conditions.
79 1111
moffati & nichol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
Figure 38. Comparisons for Surface DO at Segment 5
Comparisons for Surface DO at Segment 5
Existing Lake, Existing Loads
14
Comparisons for Bottom DO at the Dam
12
1¢
18
— — Future Conditions, Regime A
J
12
1
$
E
O
10
0
5
J
�
8
a
E
Existing Lake, Existing Leads
2
Existing Lake, Future Loads
— Future Conditions, Regime A
•••... Future Conditions, Regime 0
0
tu
0 500 1000 153C 2507 300o
Julian Day
Figure 38. Comparisons for Surface DO at Segment 5
Figure 39. Comparisons for Bottom DO at the Dam
In order to better understand the water quality dynamics of the lake further analyses of mass
loading and water quality response were conducted. The analyses showed that, with the
introduction of make -up water from the Cape Fear River under Future Regime A and B, the TP
load is more than triple that of existing conditions on average during the simulation period.
However, the expanded reservoir will have a pool volume 2.75 times the existing storage
80 1/11
moffati & nichol
Existing Lake, Existing Loads
Comparisons for Bottom DO at the Dam
Existing Lake, Future Loads
1¢
— — Future Conditions, Regime A
••. • • . Future Conditions, Regime &
12
10
J
�
8
E
n�
tu
a
I
z
I
-A
h
n
0
500 low 1500 2600
2500 3000
Julian Day
Figure 39. Comparisons for Bottom DO at the Dam
In order to better understand the water quality dynamics of the lake further analyses of mass
loading and water quality response were conducted. The analyses showed that, with the
introduction of make -up water from the Cape Fear River under Future Regime A and B, the TP
load is more than triple that of existing conditions on average during the simulation period.
However, the expanded reservoir will have a pool volume 2.75 times the existing storage
80 1/11
moffati & nichol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
volume. Obviously, from a mass and volume perspective, the increased loading, when
considered relative to the increase in storage volume, cannot alone account for the increases in
predicted nutrient and chlorophyll -a concentrations for future scenarios. For this reason, factors
affecting hydrology and residence time were also examined in detail.
Analysis of lake water residence time was conducted using water age, a model- computed
parameter that serves as a surrogate for residence time. The water age for existing conditions
fluctuates between 200 -500 days for most of the 8 -year simulation, and between 400 -700 days
for the future conditions. Close examination of the timing of increases in water age in
conjunction with the timing of increased in nutrient levels within the lake and the timing of
increases in algal growth via chlorophyll -a indicate that periods of increased nutrient loads and
eutrophication correlate with water age.
5. Conclusions
The updated water quality modelling analysis described in this report represents several
improvements over previous analyses applied to Harris Lake, not the least of which include:
• Revised and refined model bathymetry
• An updated and improved water balance
• Revised tributary nonpoint source nutrient loads
• Revised and improved meteorological data files
• Refined physical descriptions for the dam spillway
• Refined physical descriptions for the cooling water intake and discharge structures
• Simulation of real -world cooling water withdrawal and blow -down discharge rates
These improvements result in an improved simulation of the dynamics affecting water quality
conditions in Harris Lake, and this model provides a credible representation of the Harris Lake
system water quality.
The model indicates that for drier periods the future lake (addition of the two reactors and
associated lake changes and pumped water) will have higher nutrient and algal concentrations
compared to those associated with the existing lake with future loadings; but for wetter periods,
there will be little difference in -lake concentrations for the future lake compared to the existing
lake with future loads. The critical periods for increases in nutrient and algal concentrations (for
existing lake and future lake conditions) are during drought conditions that extend the lake water
residence time. Pumping water from the Cape Fear River is predicted to increase nutrient and
algal concentrations in the lake, but the amount of increase for any year will be dependent on
the local rainfall and hydrology. The potential water quality impacts, in terms of the increases in
81 1111
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
algal productivity that could be associated with these increases, are at least partly mitigated by
the increased volume of the lake, which is nearly tripled.
The last year of the simulation (2008) represent one of the more stressful years for water quality
due to increased tributary inflows and pollutant loads after a drought year, increased pumping
from the Cape Fear River, and the Holly Springs wastewater discharge running at maximum
permitted levels. In spite of these more stressful conditions, chlorophyll a concentrations are
predicted to remain below 25 µg /I throughout the year.
Taken collectively, the model predictions indicate that the Shearon Harris Nuclear Power Plant
can be expanded and operated in the manner simulated herein without resulting in violations of
water quality standards in Harris Lake.
82 1111
moffati & mchol
MODELING REPORT FINAL
Update & Recalibration of Harris Lake Watershed and Water Quality Models October 2012
VIII. References
CH2M HILL. 2008. Technical Memorandum: Development and Calibration of a CE- QUAL -W2
Model for Harris Lake, prepared for Western Wake Partners.
CH2M HILL. 2009. Technical Memorandum: Development and Calibration of a CE- QUAL -W2
Model for Harris Lake including 2008 Update, prepared for Western Wake Partners.
Cole, T.M., and Wells, S.A. 2003. CE- QUAL -W2: a two - dimensional, laterally- averaged,
hydrodynamic and water quality model, version 3.2, user manual, Instruction report EL -03 -1
(draft), prepared for U.S. Army Corps of Engineers.
Haith, D.A. and L.L. Shoemaker, 1987. Generalized Watershed Loading Functions for Stream
Flow Nutrients. Water Resources Bulletin, 23(3), pp. 471 -478.
HydoLogics, Inc. 2011. Modeling the Proposed Operations of Harris Lake Using OASIS,
prepared for Progress Energy Carolinas.
NC DWQ. 2005. B. Everett Jordan Reservoir, North Carolina Nutrient Management Strategy
and Total Maximum Daily Load. Public Review Draft, North Carolina Division of Water
Quality, Raleigh, NC.
Tetra Tech .2003. B. Everett Jordan Lake TMDL Watershed Model Development. Prepared for
the N.C. Division of Water Quality. Raleigh. NC.
Tetra Tech. 2004.Modeling Analysis of the Holly Springs WWTP Nutrient and BOD Discharge
Impact on the Utley Creek Watershed, Prepared for the Town of Holly Springs and the N.C.
Division of Water Quality. Raleigh. NC.
83 1111
moffati & mchol
1111
moffatt & nichol
Creative People, Practical Solutions
1616 East Millbrook Road, Suite 160
Raleigh, North Carolina 27609
P: 919 - 781 - 4626 1 F: 919 - 781 - 4869