MODELING THE CIRCULATION AND WATER QUALITY
IN CHARLOTTE HARBOR ESTUARINE SYSTEM, FLORIDA
By
KIJFN PARK
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
2004
ACKNOWLEDGMENTS
I would like to thank my advisor, Dr. Y. Peter Sheng, for his guidance, support and
financial assistance throughout my study. In addition, much appreciation is owned to my
other committee members, Dr. Robert G. Dean, Dr. Robert J. Thieke, Dr. Louis H. Motz and
Dr. K. Ramesh Reddy, for their review of my dissertation.
I would like to thank the South Florida Water Management District and Southwest
florida Water Management District for sponsoring research project and providing data for
Charlotte Harbor estuarine system. I would like to express my thanks to Justin, Jeffery,
Yangfeng, Taeyoon, Jun, Vadim, and Vladimir whose help with class, research and writing
this dissertation. Many thanks go to Becky, Lucy, Sonna, Sidney, Kim and Halen for making
life easier.
I would like to dedicate this dissertation to my parents whose love and support made
this degree possible. Last, but not least, I would like to thank my wife, Kyung-Mi, who have
been praying for me to be faithful, kind, and honest all the time.
11
TABLE OF CONTENTS
Page
ACKNOWLEDGMENTS »
LIST OF TABLES vi
LIST OF FIGURES ix
ABSTRACT xvil
CHAPTER
1 INTRODUCTION 1
2 CHARLOTTE HARBOR CHARACTERIZATION 9
2.1 Climate U
2.2 Hydrodynamics 1 1
2.2.1 Tidal Stage, Discharge at Inlet, and Tidal Circulation 13
2.2.2 Freshwater Flow 14
2.2.3 Salinity 14
2.3 Water Quality 15
2.3.1 Nutrients 16
2.3.2 Dissolved Oxygen 18
2.3.3 Phytoplankton 19
2.4 Sediment 20
2.5 Light Environment 21
3 HYDRODYNAMIC AND SEDIMENT TRANSPORT MODEL 23
3.1 Governing Equation 24
3.1.1 Hydrodynamic Model 24
3.1.2 Sediment Transport Model 26
3.2 Boundary and Initial Conditions 27
3.2.1 Boundary Conditions 27
3.2.2 Initial Conditions 30
3.3 Heat-Flux at Air-Sea Interface 30
3.3.1 Short-Wave Solar Radiation 31
3.3.2 Long-Wave Solar Radiation 32
in
3.3.3 Sensible and Latent Heat Fluxes 33
4 WATER QUALITY MODEL 36
4.1 Mathematical Formulae 38
4.2 Phytoplankton Dynamics 40
4.2.1 Modeling Approach 40
4.2.2 Relationship between Phytoplankton and Nutrients 42
4.3 Nutrient Dynamics 43
4.4 Oxygen Balance 45
4.5 Effects of Temperature and Light Intensity on Water Quality Model 59
4.5.1 Temperature 59
4.5.2 Light intensity 61
4.6 Light Attenuation Model 61
4.7 Model Parameters and Calibration Procedures 69
5 APPLICATION OF CIRCULATION AND TRANSPORT MODEL 81
5.1 A High-Resolution Curvilinear Grid for Charlotte Harbor Estuarine System 82
5.2 Forcing Mechanism and Boundary Condition 86
5.3 Simulations for 1986 Hydrodynamics 94
5.3.1 Sensitivity and Calibration Simulations 99
5.3.2 Results of the 1986 simulation 101
5.4 Simulations for 2000 Hydrodynamic 118
5.4.1 Sensitivity and Calibration Simulations 118
5.4.2 Results of the 2000 Simulation 130
5.4.3 Application of the 2000 hydrodynamic Simulations 146
6 APPLICATION OF WATER QUALITY MODEL 167
6.1 Forcing Mechanism and Boundary Condition for Circulation 167
6.2 Initial and Boundary Condition for the Water Quality Model 172
6.3 Simulations of Water Quality in 1996 180
6.3.1 Calibration 180
6.3.2 Results for 1996 Water Quality Simulations 191
6.4 Simulations of Water Quality in 2000 212
6.4.1 Verification 212
6.4.2 Results of 2000 Water Quality Simulations 215
6.4.3 Application of 2000 Water Quality Simulations 235
7 CONCLUSION AND DISCUSSION 258
APPENDIX
A FLOW CHARTS FOR CH3DJMS 263
IV
B DMENSIONLESS EQUATIONS IN CURVILINEAR BOUNDARY-FITTED AND
SIGMA GRID 278
C COMPARISON OF WATER QUALITY MODELS 284
D NUMERICAL SOLUTION TECHNIQUES FOR WATER QUALITY MODEL . . 289
E NUTRIENT DYNAMICS 294
F SEDIMENT FLUX MODEL 306
G MODEL PERFORMANCE TEST WITH PARALLEL CH3D_IMS 309
H TIDAL BENCH MARKS FOR CHARLOTTE HARBOR 313
REFERENCES 314
BIOGRAPHICAL SKETCH 325
LIST OF TABLES
Table E^e
1.1 Component models of the CH3D-IMS 6
2.1 Ratios of nitrogen and phosphorous constituents 18
3.1 Mean latitudinal values of the coefficient X 33
4. 1 Average values of oxygen uptake rates of bottom 54
4.2 The spectrum of incident sunlight data 64
4.3 Coefficient ranges to use in stand along light model 67
4.4 The best fit light model coefficients for Charlotte Harbor estuarine system 67
4.5 Dixon and Gray's model coefficients for the Charlotte Harbor estuarine system.. . 67
4.6 Temperature adjustment functions for water quality parameters 70
4.7 Water quality parameters related to conversion rate 71
4.8 Water quality parameters related to phytoplankton and zooplankton 71
4.9 Water quality parameters in the nitrogen dynamics 72
4.10 Water quality parameters in the phosphorous dynamics 73
4.11 Water quality parameters in the oxygen balance 74
4.12 The relationship between water quality parameters and model constituents 79
5.1 Descriptions of 1986 and 2000 river boundary conditions for Charlotte Harbor
estuarine system 87
5.2 Locations of tidal stage and velocity-salinity measured stations by USGS 96
VI
5.3 The effect of removing selected boundary conditions on the accuracy of simulated
water level, velocity and salinity in July 1986. Value shown are average RMS
differences vs. baseline simulation at all data stations 99
5.4 The effect of varying bottom roughness, z0, on the accuracy of simulated water level,
velocity and salinity in July 1986. Value shown are average RMS errors at all
data stations 100
5.5 The effect of varying horizontal diffusion, AH, on the accuracy of simulated water
level, velocity and salinity in July 1986. Value shown average RMS errors at all
data stations 101
5.6 A summary of boundary conditions and model parameters used in the 1986
simulation 102
5.7 Calculated RMS errors between simulated and measured water level for 1986
simulation 102
5.8 Calculated RMS errors between simulated and measured current velocity for 1986
simulation 104
5.9 Calculated RMS errors between simulated and measured salinity for 1986
simulation 105
5.10 The effect of horizontal grid resolution, on the accuracy of simulated water level
and salinity. Values shown are average RMS errors for 2000 calibration at all
available stations. Values shown in parenthesis are % RMS error normalized by
maximum values 121
5.1 1 The effect of vertical grid resolution, on the accuracy of simulated water level and
salinity. Values shown are average RMS errors for 2000 calibration at all
available stations. Values shown in parenthesis are % RMS error normalized by
maximum values 122
5.12 The effect of varying bottom roughness, z0, on the accuracy of simulated water level
and salinity in 2000. Values shown are average RMS errors all data stations. . 123
5.13 The effect of varying salinity advection scheme on the accuracy of simulated water
level and salinity in 2000. Values shown are average RMS errors at all data
stations 128
5.14 The effect of modifying bathymetry on the accuracy of simulated water level and
salinity in 2000. Values shown are average RMS errors for all data stations. . 129
vn
5. 15 A summary of boundary conditions and model parameters used in the 2000
simulation ljyj
5.16 Calculated RMS errors between simulated and measured water level for 2000
simulation "1
5.17 Calculated RMS errors between simulated and measured salinity for 2000
simulation "3
5.18 The effect of hydrologic alternations on 2000 water level. Value shown are
average RMS differences with baseline simulation for all selected stations. . . 154
5.19 The effect of hydrologic alternations on 2000 salinity. Value shown are average
RMS differences with baseline simulation for all selected stations 155
6.1 Locations of water quality measured stations 173
6.2 Sediment types for Charlotte Harbor water quality simulations 174
6.3 Water quality parameters, baseline values, and ranges used in the sensitivity
analysis 1"
6.4 Sensitivity analysis results in RMS difference w.r.t. baseline for 1996 water quality
calibration simulation 184
6.5 The water quality model coefficients used for the Charlotte Harbor simulation. . . 185
6.6 The temporally averaged water quality species concentrations for baseline 2000
simulation 240
6.7 Normalized RMS differences of water quality species concentrations at 1 1 stations
during April to June 2000, showing the effect of no causeway islands 240
6.8 Normalized RMS differences of water quality species concentrations at 11 stations
during April to June 2000, showing the effect of no ICW 241
6.9 Normalized RMS differences of watei quality species concentrations at 1 1 stations
during April to June 2000, showing the effect of no causeway islands and ICW241
C.l Comparison of water quality models 287
G. 1 CPU time for the parallel, shared memory, CH3D procedures on SGI origin
platform 311
H.l Tidal datum referred to Mean Low Low Water (MLLW), in meter 313
viii
LIST OF FIGURES
Figure Page
1.1 Charlotte Harbor estuarine system and its subarea boundaries 2
2.1 Drainage basins of Charlotte Harbor estuarine system 10
2.2 Seasonal wind pattern in Florida 12
2.3 Average monthly concentration of dissolved oxygen in upper Charlotte Harbor, site
CH-006, 1976-84 19
4.1 The connection between nitrogen, phosphorous and carbon cycle 44
4.2 CBOD cycle and DO cycle 46
4.3 Comparison of wind-dependent re-aeration formula 49
4.4 The relationship between POM flux and SOD flux related in the oxidation and
reduction of organic matter in sediment column 53
4.5 Effect of dissolved oxygen on sediment consumption and SOD release 57
4.6 The scatter plots for kd(PAR) during calibration period with best fit coefficients . . 68
4.7 Systematic calibration procedure 80
5.1 Boundary-fitted grid (92 x 129) used for numerical simulation for Charlotte Harbor
estuarine system 84
5.2 Bathymetry in the boundary-fitted grid for Charlotte Harbor estuarine system (92 x
129) 85
5.3 Tidal forcing and river discharges for 1986 simulations of Charlotte Harbor
circulation 89
5.4 Wind velocity for 1986 simulations for Charlotte Harbor circulation 90
ix
5.5 Tidal forcing and river discharges for 2000 simulations of Charlotte Harbor
circulation 91
5.6 Wind velocity for 2000 simulation of Charlotte Harbor circulation 92
5.7 Air temperature for 2000 simulation of Charlotte Harbor circulation 93
5.8 Locations of the available 1986 water level and discharge measurement stations of
USGS 97
5.9 Locations of available 1986 velocity and salinity measurement stations of USGS . 98
5.10 Comparison between simulated and measured water level in July 1986 108
5. 1 1 Comparison between simulated and measure spectra of water level in July 1986 1 10
5.12 Comparison between simulated and measured current velocity in July 1986 ... 1 1 1
5.13 Comparison between simulated and measured salinity in July 1986 115
5.14 Typical flow pattern of Charlotte Harbor estuarine system during one tidal cycle for
August 6, 1986 H6
5.15 The 29-day residual flow and salinity for Charlotte Harbor estuarine system during
July 2 to July 30, 1986 117
5.16 Locations of the available 2000 water level and salinity measured stations at
Caloosahatchee River operated by SFWMD 119
5.17 The comparison of the coarse grid (71x92) and the fine grid (92x129) for Charlotte
Harbor estuarine system 120
5.18 A comparison between simulated and measured salinity at Shell Point using
ultimate QUICKEST, QUICKEST, and upwind advection schemes 124
5.19 A comparison between simulated and measured salinity at Fort Myers using
ultimate QUICKEST, QUICKEST, and upwind advection schemes 125
5.20 A comparison between simulated and measured salinity at BR31 using ultimate
QUICKEST, QUICKEST, and upwind advection schemes 126
5.21 Simulated longitudinal-vertical salinity along the Caloosahatchee River at slack-
water before flood tide on September 7, 2000 127
5.22 Comparison between simulated and measured water level in 2000 136
x
5.23 Comparison between simulated and measured salinity at S79 in 2000 137
5.24 Comparison between simulated and measured salinity at BR31 in 2000 138
5.25 Comparison between simulated and measured salinity at Fort Myers in 2000. . . 139
5.26 Comparison between simulated and measured salinity at Shell Point in 2000. . . 140
5.27 Comparison between simulated and measured salinity near Sanibel Causeway in
2000 141
5 28 Comparison between simulated and measured temperature at Fort Myers in 2000.
142
5.29 Typical flow pattern of San Carlos Bay during ebb and flood tide for August, 7 on
2000 143
5.30 One-year residual flow in San Carlos Bay in 2000 144
5.31 One-year residual salinity distribution in San Carlos Bay in 2000 145
5.32 The locations of Sanibel Causeway and IntraCoastal Waterway and stations for
comparing the effect of hydrologic alterations 148
5.33 The comparison of bathymetry and shoreline for each hydrologic alteration case
scenarios which are Baseline, the absence of IntraCoastal Waterway, and the
absence of causeway l4y
5.34 The comparisons of water level for three cases at three selected stations: ST05 (Pine
Island Sound), ST08 (San Carlos Bay), and ST10 (Caloosahatchee River mouth).
150
5.35 The comparisons of surface and bottom salinity for three cases at three selected
stations: ST05 (Pine Island Sound), ST08 (San Carlos Bay), and ST 10
(Caloosahatchee River mouth) 1*1
5.36 The comparisons of surface residual flow and salinity fields for three cases. ... 152
5.37 The comparisons of bottom residual flow and salinity fields for three cases. ... 153
5.38 The vertical-longitudinal salinity profiles along the axis of the Caloosahatchee River
during wet season in 2000 l->9
XI
5.39 The vertical-longitudinal salinity profiles along the axis of the Caloosahatchee River
during dry season in 2000 160
5.40 Time histories of river discharge at S-79 and the locations of 1, 10, and 20 ppt
surface salinity along the Caloosahatchee River during 2000 simulation period.161
5.41 Time histories of river discharge at S-79 and the 1 ppt salinity location along
Caloosahatchee River 162
5.42 The 1-day averaged 20 ppt surface salinity location and 30-day averaged 10 ppt
surface salinity location during 2000 simulation period 163
5.43 The locations of 10 ppt surface salinity due to river discharge rate at S-79 during
2000 baseline simulation 164
5.44 The relationship between location of specific salinity value vs. river discharge at
S-79 165
5 45 The relationship between salinity at Fort Myers station vs. river discharge at S-79.
166
6.1 Tidal forcing and river discharges for 1996 simulations of Charlotte Harbor 169
6.2 Wind velocity for 1996 simulations of Charlotte Harbor 170
6.3 Air temperature for 1996 simulations of Charlotte Harbor 171
6.4 Locations of 1996 water quality measurement stations operated by EPA 176
6.5 Locations of 2000 water quality measurement stations operated by SFWMD and
SWFWMD 177
6.6 Light intensity at water surface for 1996 and 2000 simulations 178
6.7 Segments for Charlotte Harbor estuarine system 179
6.8 The scatter plots for water quality constituents during calibration period 189
6.9 Temporal water quality variations at CH002 station in 1996 194
6.10 Temporal water quality variations at CH004 station in 1996 195
6.1 1 Temporal water quality variations at CH005 station in 1996 196
6.12 Temporal water quality variations at CH006 station in 1996 197
xii
6.13 Temporal water quality variations at CH007 station in 1996 198
6.14 Temporal water quality variations at CH09B station in 1996 199
6.15 Temporal water quality variations at CH009 station in 1996 200
6.16 Temporal water quality variations at CH010 station in 1996 201
6.17 Temporal water quality variations at HB002 station in 1996 202
6.18 Temporal water quality variations at HB006 station in 1996 203
6.19 Temporal water quality variations at HB007 station in 1996 204
6.20 Temporal water quality variations at CH013 station in 1996 205
6.21 Simulated dissolved oxygen concentration in Charlotte Harbor estuarine system for
August 21, 1996 206
6.22 Simulated chlorophyll a concentration in Charlotte Harbor estuarine system for
August 21, 1996 207
6.23 Simulated dissolved ammonium nitrogen concentration in Charlotte Harbor
estuarine system for August 21, 1996 208
6.24 Simulated soluble organic nitrogen concentration in Charlotte Harbor estuarine
system for August 21, 1996 209
6.25 Simulated soluble reactive phosphorous concentration in Charlotte Harbor estuarine
system for August 21, 1996 210
6.26 Simulated soluble organic phosphorous concentration in Charlotte Harbor estuarine
system for August 21, 1996 211
6.27 The scatter plots for water quality constituents in 2000 213
6.28 Temporal water quality variations at CH002 station in 2000 218
6.29 Temporal water quality variations at CH004 station in 2000 219
6.30 Temporal water quality variations at CH005 station in 2000 220
6.31 Temporal water quality variations at CH006 station in 2000 221
xm
6.32 Temporal water quality variations at CH007 station in 2000 222
6.33 Temporal water quality variations at CH09B station in 2000 223
6.34 Temporal water quality variations at CH009 station in 2000 224
6.35 Temporal water quality variations at CH010 station in 2000 225
6.36 Temporal water quality variations at CES02 station in 2000 226
6.37 Temporal water quality variations at CES03 station in 2000 227
6.38 Temporal water quality variations at CES08 station in 2000 228
6.39 Temporal water quality variations at CHOI 3 station in 2000 229
6.40 The comparison between simulated dissolved oxygen concentration and the
possible causes hypoxia: river discharge, salinity, temperature, and re-aeration and
SOD fluxes at CH006 water quality measured station 230
6.41 Simulated longitudinal-vertical salinity and dissolved oxygen concentration along
the Peace River at 1 pm on June 18 (Julian Day 170), 2000 231
6.42 Simulated longitudinal-vertical salinity and dissolved oxygen concentration along
the Peace River at 1 pm on October 6 (Julian Day 280), 2000 232
6.43 Simulated near-surface chlorophyll a concentration in Charlotte Harbor estuarine
system for February 9, May 9, August 7, November 5, 2000 233
6.44 Simulated near-bottom dissolved oxygen concentration in Charlotte Harbor
estuarine system for February 9, May 9. August 7, November 5, 2000 234
6.45 Comparisons of simulated surface chlorophyll a concentration for three cases at
three selected stations: ST05 (Pine Island Sound), ST08 (San Carlos Bay), and
ST10 (Caloosahatchee River mouth) 237
6.46 Comparisons of simulated surface chlorophyll a concentration fields in San Carlos
Bay after 90 days of simulation for three cases 238
6.47 Comparisons of simulated surface dissolved ammonium nitrogen (NH4)
concentration fields in San Carlos Bay after 90 days of simulation for three cases.
239
6.48 The water quality species at CH004 water quality measured station before and after
100 % nitrogen load reduction from Peace River 247
xiv
6.49 The water quality species at CH006 water quality measured station before and after
100 % nitrogen load reduction from Peace River 248
6.50 The water quality species at CH004 water quality measured station before and after
100 % phosphorous load reduction from Peace River 249
6.51 The water quality species at CH006 water quality measured station before and after
100 % phosphorous load reduction from Peace River 250
6.52 The water quality species at CES02 water quality measured station before and after
100 % nitrogen load reduction from Caloosahatchee River 251
6.53 The water quality species at CES08 water quality measured station before and after
100 % nitrogen load reduction from Caloosahatchee River 252
6.54 The water quality species at CES02 water quality measured station before and after
100 % phosphorous load reduction from Caloosahatchee River 253
6.55 The water quality species at CES08 water quality measured station before and after
100 % phosphorous load reduction from Caloosahatchee River 254
6.56 Dissolved oxygen and Chlorophyll a concentrations at CH006 water quality
measured station before and after 100 % organic matter load reduction from Peace
River using DiToro's sediment flux model 255
6.57 Dissolved oxygen concentrations at CH004 and CH006 water quality measured
stations before and after 50 % SOD reduction and 75% SOD reduction 256
6.58 The comparison of hypoxia area at Upper Charlotte Harbor according to varying
SOD constant rate at 20°C 257
A.l Flow chart for main CH3D program 264
A.2 Flow chart for driving subroutine for the time stepping of the solution 266
A. 3 Flow chart for initializing sediment transport model 271
A.4 Flow chart for main sediment transport model 272
A.5 Flow chart for initializing water quality model 277
A.6 Flow chart for main water quality model 275
xv
A.7 Flow chart for computing temperature and light attenuation functions for water
quality model 277
D.l The vertical one-dimensional z-grid 292
E. 1 Nitrogen cycle 296
E.2 Phosphorous cycle 302
G.l Parallel speedup gained in performing the simulation on the SGI Origin platform.312
xvi
Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
MODELING THE CIRCULATION AND WATER QUALITY
IN CHARLOTTE HARBOR ESTUARINE SYSTEM, FLORIDA
By
Kijin Park
August 2004
Chair: Y. Peter Sheng
Department: Civil and Coastal Engineering
This study aims to develop an enhanced version of a three-dimensional curvilinear-
grid modeling system, CH3D-IMS, which include a 3-D hydrodynamics model, a 3-D
sediment transport model, and a 3-D water quality model, to simulate circulation and water
quality of the Charlotte Harbor Estuarine System and to provide quantitative assessment of
various management practices.
In the past decade, the upper Charlotte Harbor system has been suffering summer
hypoxia in bottom water. Field study indicated that hypoxia in the upper Charlotte Harbor
is related to a strong stratification caused by high freshwater flows and dissolved oxygen
fluxes at the air-sea and sediment-water interfaces. To simulate the hypoxia event, models
of oxygen balance and oxygen fluxes at the air-sea and sediment-water interfaces in previous
versions of CH3D-IMS are enhanced. The three dimensional temperature model and
physics-based light model in CH3D-IMS are also enhanced to enable better understanding
xvn
of the temporal and spatial variations of temperature and light and their effect on water
quality processes.
The hydrodynamics component of the integrated model CH3D-IMS for Charlotte
Harbor has been successfully calibrated, using hydrodynamic data gathered by the National
Oceanic and Atmospheric Administration (NOAA) and the United States Geological Survey
(USGS) in 1986 and 2000. This calibrated model was applied to assess the impact of the
removal of the Sanibel Causeway and the IntraCoastal Waterway on the circulation in the
San Carlos Bay area and to provide a quantitative evaluation of the minimum flow and level
(MFL) for Caloosahatchee River. The results show that the hydrologic alterations would not
noticeably affect the circulation, salinity, and water quality in San Carlos Bay except near the
mouth of Caloosahatchee River. The minimum flow required to produce a salinity of no
more than 10 ppt at Fort Myers is about 18 m3/s.
The water quality model of the CH3D-MS was calibrated with the systematic
calibration procedure and validated using hydrodynamic, sediment, and water quality data
provided by the USGS and the United States Environmental Protection Agency (USEPA)
in 1996, and by the USGS, the South Florida Water Management District (SFWMD) and the
Southwest Florida Water Management District (SWFWMD) in 2000. This validated model
was used to examine the temporal and spatial dynamics of factors which can affect hypoxia
in upper Charlotte Harbor, such as freshwater inflow, tidal variation, sediment oxygen
demand (SOD), water column oxygen consumption, and dissolved oxygen (DO) re-aeration.
The model results suggest that hypoxia in the upper Charlotte Harbor System is primarily
caused by a combination of vertical salinity stratification and SOD, while water quality also
affected by the oxygen re-aeration and water column oxygen consumption. This model was
xvin
applied to assess the effects of hydrologic alterations and to provide a preliminary evaluation
of pollutant load reduction goal (PLRG). Due to lack of detailed data and insufficient
understanding on the causes of SOD and how SOD is related to the loading, the present
model cannot simulate the complete effect of nutrient load reduction on the DO
concentration in the estuary. Further research and more complete data are needed
A systematic calibration procedure has been developed for a more efficient and more
objective calibration of the water quality model. A consistent framework for systematic
calibration is formulated, which include in the following steps: model parameterization,
selection of calibration parameters, and formulation of calibration criteria.
xix
CHAPTER 1
INTRODUCTION
Charlotte Harbor (Figure 1 . 1 ) is a shallow estuarine system in southwest Florida. The
estuary receives freshwater from the Caloosahatchee, Peace, and Myakka Rivers; is
connected to the Gulf of Mexico through the Boca Grande Pass, Gasparilla Pass, Captiva
Pass, Blind Pass, and San Carlos Bay; and provides water resources for several counties in
Southwest Florida. Charlotte Harbor is dominated by rivers that flow into the coastal area.
While most estuaries in Southwest Florida are influenced by the Gulf of Mexico, the
characteristics of the Charlotte Harbor system are strongly influenced by large rivers such as
the Caloosahatchee and Peace Rivers. Large fluctuations of river flow between wet and dry
seasons strongly affect the salinity and water characteristics in Charlotte Harbor (Estevez,
1998).
Population growth and development in the surrounding areas during the past few
decades have led to concerns over human impacts on the quality of the estuarine system.
Industrial and agriculture development also increase environment pollution. Growth and
development will cause an increased demand for fresh water and a corresponding increase
in urban, agricultural, and industrial waste. The inflow of freshwater is essential to the
integrity and health of the estuarine system. Increased freshwater withdrawal or diversion,
or increased wastewater discharges to the rivers and streams that flow into the estuary will
create environmental stress in the estuary (McPherson et al., 1996).
1
Myakka River
%> Deer Prairie Creek
\ Dirt Cl^i tr
Big Slough Canal
h& J Sarasota Co.
Nortlmortf*Z- •---
■y Char
Peace River
!\ V El Jpbean
i j« De_SotoC_o.
clia7iotteCo.*W'~
-. 11
\^Gaspar1lla\ "«*» i
w^ Sound ~ Charlotte jr
Gasparilla \h£ ,.) M Harbor It-
Pass -^ »iw/''fi £
V^Lower X
-rravfa Charlotte HarborV /
Boca Grande^ Bokeelia y.
Pass , -. ^^M^f%
CayoCostit.': S ^
S79
Caloosahatchee River. fJmQj <T'"
Charlotte Co.
Lee Co.
'u^ & ^ -\ Matlacha Pass J- * ~X
Captiva 'Pin_,-^i \ i*„ ■; --. Fort Myers
Pjjqq ^- I II IC -i .- j, ■■■■■ i, < _,
rass Island .. v &M r„„„ tJP ;'
■Island .\ \j?f>t Cape jgP /
v Sound i # a* C'or«/. >f ,>
RedfishV -T -.v aia^.^vv
'-■:, Sound-'^, %"' £* Cora/ ^"|f ,>
^ Cf V':\-. StlJ^ /^Whiskey Creek
f, PineSf 1^-^r^^
Captiva ,-, Islanq Ajlffr ._ *piVtr.. _ . f
Pass
Blind
Pass-
o
$iell PoiiU j
-■ .- „-*
Sanibel/ ./'' Causeway '%r^t
Island San Carlos Fort M s^fc ^
Bay &at./i <$ j
>P- •■'■>-> M
Estero Bay *5r Imperial River ■ *
:%"~ A
EXPLANATION
— — Subarea Divide
1
» Naples
Kilometers
Lb ^ ^ e
ho 0 10
20
_^__ i .
Figure 1.1 Charlotte Harbor estuarine system and its subarea boundaries
Water quality in the southern Charlotte Harbor estuarine system (the Caloosahatchee
Estuary, San Carlos Bay, and Pine Island Sound) appear to be influenced by freshwater
discharge from the Franklin Lock and Dam, also known as United States Army Corps of
Engineers Structure Number 79 (S-79). Waste load allocation studies conducted by the
Florida Department of Environmental Regulation (Degrove, 198 1 ; Degrove, and Nearhoof,
1987; Baker, 1990) concluded that the Caloosahatchee Estuary had reached its nutrient
loading limits as indicated by elevated chlorophyll-a and depressed dissolved oxygen (DO)
levels. Similarly, McPherson and Miller (1990) concluded that increased nitrogen loading
would result in undesirable increases in phytoplankton and benthic algae. In order to
effectively manage the loading of pollutants such as nutrients in a shallow estuary such as
Charlotte Harbor, and control eutrophication, it is necessary to have a quantitative
understanding of the transport and transformation processes of nutrients in system.
Increased nutrient loading from tributaries, the atmosphere, and bottom sediments
have been known to cause eutrophication in estuaries. Eutrophication varies both spatially
and temporally. Eutrophication does not necessarily occur in regions with high nutrient
loading because the circulation and sediment dynamics affect nutrient fate and transport. In
the Upper Charlotte Harbor where a pycnocline develops frequently during high river flow
conditions, hypoxia usually develops within two days of the formation of the pycnocline,
when coupled with significant sediment oxygen demand (SOD) (CDM, 1998). In fact,
hypoxia can occasionally extend all the way to the Boca Grande area. This strong linkage
between hydrodynamics and water quality dynamics suggests the importance of having a
comprehensive understanding of hydrodynamics.
Nutrient dynamics in estuaries are not only determined by biological and chemical
processes, but is also strongly affected by weather and climate (wind, air temperature, etc.),
hydrodynamics (wave, tide, current, turbulence mixing, etc.), and sediment transport
processes (resuspension, deposition, flocculation, etc.). In this study, integrated modeling
of the system is conducted to understand the complex water quality processes, which are
strongly linked to hydrodynamics and sediment processes. The primary objective of this
study is to use models and field data to produce a detailed characterization of
hydrodynamics, sediment, and water quality dynamics within the Charlotte Harbor estuarine
system.
The Chesapeake Bay model (Cerco and Cole, 1994) and Moreton Bay model
(McEwan and Garbric, 1998) are coupled hydrodynamics-water quality models. These
coupled models are more suitable for managing nutrient loads and predicting eutrophication-
related problems than uncoupled models. Because these models were not coupled with a
dynamic model for sediment transport, however, they could not accurately consider
sediment-process effects such as resuspension, deposition, flocculation, and settling on
nutrient dynamics in estuaries. Therefore, these loosely coupled models cannot account for
nutrient release by sediments in episodic events (Chen and Sheng, 1994)
Chen and Sheng (1994) developed a coupled hydrodynamic-sediment-water quality
model and applied it to Lake Okeechobee. Their 3-D model includes a hydrodynamic model,
a sediment transport model, and a water quality model with a nitrogen and a phosphorous
cycle on a rectangular grid. Measured nitrogen and phosphorous dynamics during episodic
events in 1993 and monthly sampling events in 1989 were accurately simulated with
absorption-desorption reactions and the exchange of nutrients between the sediment and
water column.
While Chen and Sheng (1994) developed a rectangular grid model, Yassuda and
Sheng (1996) developed an integrated model in curvilinear grids and applied it to Tampa
Bay. Their water quality model, based on the Chen and Sheng (1994) model, includes a
nitrogen cycle and an oxygen cycle, but not a phosphorous cycle. Zooplankton distribution
is allowed to influence the phytoplankton dynamics. Their integrated model also includes
a wave model, a light model, and a seagrass model. Although Yassuda and Sheng (1996)
simulated the observed hypoxia in Tampa Bay during 1991, their model did not include
dynamic fluxes of oxygen at the air-sea interface (re-aeration) and the sediment-water
interface (sediment oxygen demand). Their model could not simulate the daily fluctuation
and vertical stratification of DO. Although the phytoplankton growth rate in their model is
controlled by temperature and light, it is difficult to reproduce the vertical distribution of
phytoplankton because the temperature is not simulated. The light model in Yassuda and
Sheng (1996) has not been sufficiently validated with Tampa Bay data, thus limiting the
predictability of their model for vertical distribution of phytoplankton and dissolved oxygen.
Their model could not produce the low dissolved oxygen phenomena in upper Charlotte
Harbor because there are no dissolved oxygen fluxes at the air-water and sediment-water
interfaces, such as reaeration and sediment oxygen demand (SOD).
Sheng (2000) developed the framework for an integrated modeling system: CH3D-
EVIS (http://ch3d.coastal.ufl.edu). The integrated modeling system (Table 1.1) is based on
the curvilinear-grid hydrodynamic model CH3D (Sheng, 1987, 1989; Sheng et al., 2002) and
also includes a wave model, a sediment transport model (Sheng et al., 2002a), a water quality
model, (Sheng et al., 2001b), a light attenuation model (Sheng et al., 2002c, Christian and
Sheng, 2003), and a seagrass model (Sheng et al., 2002d). Sheng et al. (2002) applied the
CH3D-IMS to the Indian River Lagoon (IRL), Florida. The water quality model (Sheng et
al., 2001b) includes a nitrogen cycle, a phosphorous cycle, a phytoplankton cycle, and a DO
cycle. The CH3D-IMS was validated with comprehensive data collected from the IRL.
Table 1.1 Component Models of the CH3D-IMS
Component Model Model Name
Hydrodynamic Model CH3D
Flow Model
Salinity Transport Model
Wave Model SMB
Sediment Transport Model CH3D-SED3D
Water Quality Model CH3D-WQ3D
Dissolved Oxygen Model
Phytoplankton and zooplankton
Model
Nitrogen Model
Phosphorous Model
Light Attenuation Model CH3D-LA
Seagrass Model CH3D-SAV
The CH3D-IMS was able to successfully simulate the observed circulation, wave,
sediment transport, water quality, light attenuation, and seagrass biomass in IRL during 1998.
However, several aspects of the model were identified for further improvement. For
example, although a temperature model was developed during the later phase of the IRL
study, it was not sufficiently validated with data. The water quality model, while capable of
simulating the annual variation in DO, could not simulate the diurnal DO variation.
In this study, the CH3D-IMS is enhanced and used to study the circulation and water
quality dynamics in the Charlotte Harbor estuarine system. The validated model will be used
to determine minimum flow and level and pollutant load reduction goal by management
agencies. Specially, the CH3D-IMS used for this study includes the simulation of
temperature in addition to the flow and salinity simulations. An air-sea heat flux model is
implemented along with the temperature equation. The water quality model is enhanced to
include a re-aeration model and a sediment oxygen demand (SOD) model, such that hypoxia
and daily fluctuation and vertical stratification of DO can be simulated. Moreover, the same
basic light attenuation model developed for the IRL is used for this study.
Although the factors that cause hypoxia have not been positively identified, it is
generally accepted that hypoxia is related to strong stratification caused by high freshwater
inflow. The modified CH3D-IMS is used in this study to examine the major factors of the
observed hypoxia in 2000, including freshwater inflow, tidal variation, SOD, water column
oxygen consumption, and DO reaeration.
This study includes the following objectives:
• To validate the CH3D-IMS with extensive hydrodynamic and water quality data from
Charlotte Harbor,
To improve the simulation of dissolved oxygen processes, including the observed
daily dissolved oxygen fluctuation, and hypoxia in bottom water during high flow
event with vertical stratification,
• To validate the ability of temperature model and heat flux model to simulate seasonal
and spatial temperature distribution in Charlotte Harbor,
• To apply the physics-based light attenuation model to improve Charlotte Harbor
simulations, and
To demonstrate the feasibility of the validated modeling system for simulating the
effects of various anthropogenic impacts on the Charlotte Harbor estuarine system.
More specifically, the objectives of my research are to:
• Reproduce observed circulation and transport dynamics in 1986 and 2000,
• Reproduce observed water quality dynamics in 1996 and 2000,
• Investigate the effect of the Sanibel causeway islands and IntraCoastal Waterway on
the flow, salinity, and water quality distribution in San Carlos Bay and Pine Island
Sound,
Reproduce the main factors for bottom water hypoxia and habitat loss in Charlotte
Harbor estuarine system,
• Provide a preliminary determination of pollutant load reduction goal (PLRG) and
minimum flow and levels (MFL),
Perform sensitivity tests to quantify the influence of each model parameter and to
determine the most sensitive parameters and most plausible parameter set, and
• Establish a systematic calibration procedure.
To achieve these goals, the characteristics of Charlotte Harbor estuarine system are
presented in Chapter 2. Chapter 3 discusses the general hydrodynamic circulation and
transport of temperature, salinity, and sediment in the estuarine system. This chapter will
include initial and boundary conditions for the hydrodynamic and transport models. As an
air/sea interface boundary condition, a heat flux model is also presented in this chapter.
Chapter 4 describes the water quality processes and models of phytoplankton, zooplankton,
nitrogen, phosphorous, dissolved oxygen, and light attenuation processes in estuarine system.
The effects of temperature and light intensity, numerical solution technique, and model
parameters and calibration procedure are also discussed in this chapter. Chapter 5 presents
hydrodynamic field data and model simulations of hydrodynamics in Charlotte Harbor
estuarine system. These model simulations include model calibration and verification of
1986 and 2000 data, simulation of the effect of causeway islands and navigation channel, and
determination of minimum flow and levels (MFL). Chapter 6 presents water quality and
sediment field data and model simulations of circulation, sediment transport and water
quality processes in the study area. These model simulations include calibration and
verification of 1996 and 2000 data and determination of pollutant load reduction goal
(PLRG). Discussions and conclusions are presented in Chapter 7.
CHAPTER 2
CHARLOTTE HARBOR CHARACTERIZATION
Charlotte Harbor, a coastal-plain estuarine system is one of the largest estuarine
systems on the southwest florida coast and is an important part of the Gulf of Mexico
watershed. It is located on the southwest corner of the Florida peninsula, between 26° 20'
and 27° 10'N and 81° 40' and 82° 30' W. As shown in Figure 1.1, the Charlotte Harbor
estuarine system is sub-divided into Upper Charlotte Harbor, Lower Charlotte Harbor, Pine
Island Sound, Matlacha Pass, San Carlos Bay, Gasparilla Bay, Peace River, Myakka River
and Caloosahatchee River. The drainage area (Figure 2.1) consists of the Peace, Myakka,
and Caloosahatchee River Watersheds and the coastal area and islands that drain directly into
the harbor. The estuary has a surface area of 648 km2, a drainage area of a more than 10000
km2, and a average depth of 2. 1 m. The upper harbor has an average depth of 2.6 m, and the
lower harbor has an average depth of 1.6 m (Stoker, 1986). The estuary is separated from
the Gulf of Mexico by barrier islands and is connected to the Gulf through two major inlets:
Boca Grande and San Carlos; and through several smaller passes: Gasparilla, Captiva,
Redfish (McPherson et al., 1996).
In 1995, the Charlotte Harbor National Estuary Program (CHNEP) was established
by U.S. Environmental Protection Agency (USEPA) and the State of Florida. In its planning
documents, CHNEP identified three draft major problems: (1) hydrologic alterations, (2)
nutrient enrichment, and (3) habitat loss. In order to address these problems, it is useful to
10
have a comprehensive understanding of the climate, hydrodynamics, sediment properties, and
biological and chemical characteristics of the system.
27°45'
27°00' -
Figure 2.1 Drainage basins of Charlotte Harbor estuarine system (Stoker, 1992)
11
2.1 Climate
The climate of the Charlotte Harbor estuarine system is subtropical and humid. The
mean annually average temperature is 72 °F, with low mean of 60°F in December and
January and a high mean of 80 °F during the summer (McPherson et al., 1996).
Annual rainfall averages 132 cm, of which more than half occurs from June through
September, during local thundershowers and squalls. Rain during the fall, winter, and spring
seasons is usually the result of large frontal systems and tends to be more broadly distributed
than rain associated with local thunderstorms and squalls. The period from October through
February is characteristically dry, with November usually being the driest month. The
months of April and May also are characteristically dry. Low rainfall in April and May
coincides with high evaporation and generally results in the lowest streamflow, lake stage,
and ground-water levels of the year (Hammett, 1990).
The annual average wind speed is 3.9 m/s from the east. Four typical seasonal wind-
field patterns are shown in Figure 2.2. In the Winter months, the easterly trade winds
dominate the region south of latitude 27° N, while the westerlies dominate the area north of
latitude 29°N. Spring and Summer generally exhibit more southerly winds, and Fall is
characterized by easterly or northeasterly winds. Wind speed can exceed 10 m/s during the
passage of Winter storms or during Summer squalls, hurricanes and tornadoes (Wolfe and
Drew, 1990).
2.2 Hydrodynamics
The hydrodynamic behavior of the Charlotte Harbor estuarine system is affected by
physical characteristics (bathymetry and geometry) as well as weather, climate, and
oceanographic and hydrologic characteristics, such as wind, atmosphere heating and cooling,
12
evaporation and precipitation, tidal-stage oscillations, discharge through tidal inlets, tidal
velocity, and freshwater inflow.
Figure 2.2 Seasonal wind pattern in Florida (Echternacht, 1975)
13
2.2.1 Tidal Stage, Discharge at Inlet, and Tidal Circulation
Tides along the Gulf coast of West-Central Florida in the vicinity of Charlotte Harbor
have a range of 30 to 140 cm and are of the mixed type with both diurnal and semidiurnal
characteristics (Goodwin and Michaelis, 1976). Spring tides, which have the largest range,
sometimes have only a diurnal fluctuation, whereas neap tides, which have the smallest
range, approach semidiurnal conditions of two nearly equal high and low water levels per
day.
The boundary between the Charlotte Harbor estuarine system and the Gulf of Mexico
extends about 40 mi from Gasparilla Pass on the north to San Carlos Pass on the south.
Tidal characteristics in the Gulf of Mexico are nearly uniform from Gasparilla Island to the
western face of Sanibel Island, but are of a larger range off the southern shore of Sanibel
Island (Goodwin, 1996).
Circulation in the system is primarily driven by Gulf tides, entering the system
through San Carlos Pass, Boca Grande Pass, Captita Pass, Redfish Pass, Gasparilla Pass and
Blind Pass. San Carlos Pass has a maximum depth of 5.3 m and a width of 3.3 km; Boca
Grande Pass has a maximum depth of a 18.3 m and a width of 1.28 km. The discharge
through Boca Grande Pass is about twice the discharge through San Carlos Bay and three to
four times the discharge through Captiva and Redfish Passes (Goodwin, 1996). The
geometric narrowing that occurs at passes focuses tidal energy, resulting in high velocities
associated with the large volume of water moving through the pass. This tidal energy is
dispersed inside the harbor and influences the harbor and tidal rivers as much as 40 to 44 km
upstream from the Peace River mouth. There is about a 2-hour lag between tide phases at
Boca Grande and tide phase in the upper harbor near the mouth of the Peace River (Stoker,
14
1992). Tide and wind keep the water column well mixed in the southern part of the estuary.
In the northern part of the estuary, vertical stratification can develop during moderate to high
fresh water inflows and can persist for weeks after a high freshwater inflow event (Sheng,
1998).
2.2.2 Freshwater Inflow
The majority of the freshwater that enters Charlotte Harbor come from the Myakka
River, Peace River and Caloosahatchee River. Average flows are 17.8 m3/s, 56.9m3/s and
56.7 m3/s, respectively. Flows in the Myakka and Peace Rivers are largely unregulated,
while flow in the Caloosahatchee is controlled by operation of the Franklin Lock about 43
km upstream from the mouth. Discharge in the Peace and Myakka Rivers tend to peak in
August and September when rainfall totals are generally the greatest. Discharges are usually
lowest in April and May (Stoker, 1992). The Caloosahatchee River discharge does not
always correspond to rainfall patterns in the basin, since it is controlled by S-79.
Analyses of long-term streamflow trends in the Charlotte Harbor area have indicated
statistically significant decreases in streamflow at several gages in the Peace River basin
from 1931 to 1984 (Hammett, 1990). The long-term decrease in streamflow of the Peace
River is probably related to the increased use of ground water and subsequent decline of the
potentiometric surface of the upper Florida aquafer (Hammett, 1990). A sustained
significant reduction in streamflow could result in an increase of salinity in Upper Charlotte
Harbor, possibly approaching the Gulf of Mexico salinity (Sheng, 1998). The impact of
freshwater reduction is of major ecological and economic significance.
2.2.3 Salinity
As in any other typical estuarine system, Charlotte Harbor generally exhibits
15
significant horizontal gradients in salinity. The higher salinity values in the adjacent Gulf
of Mexico fluctuate around 36 ppt, whereas the lower salinity levels nearly zero in wet
season occur near the mouth of creeks and rivers. Seasonal changes in salinity occur
primarily in response to changes in freshwater inflow from the Peace, Myakka, and
Caloosahatchee River basins. Other sources of freshwater, including direct rainfall, runoff
form coastal areas, ground-water seepage, and domestic influence, have smaller and usually
more local effects on salinity in the estuary (McPherson et al., 1996).
Stoker (1992) described salinity characteristics in the system based on data collected
from June 1982 to May 1987. Salinity generally is the lowest during the wet season between
July and September, and is the highest during the dry season from January through March.
Salinity also varies daily in response to tidal fluctuation. Peak salinity is near the flood-tide
stage, and lowest salinity is near the ebb-tide stage.
Due to the shallow depth and because of significant vertical mixing, salinity is
generally not stratifed in the southern part of the estuary. Vertical salinity stratification in
the Upper Charlotte Harbor is a common seasonal occurrence (Environmental Quality
Laboratory, Inc., 1979). In high river inflow events, a stable vertical salinity gradient is
created which suppresses vertical mixing unless there are sufficient mixing by wind or tide.
2.3 Water Quality
Water quality refers to the condition of water (e.g., dissolved oxygen concentration,
chlorophyll concentration, light attenuation, etc.) relative to legal standard, social
expectations or ecological health. Overall, water quality in the Charlotte Harbor estuarine
system is fair or good, but some areas have poor water quality or declining trends (Estevez,
1998). The water quality of an estuary is strongly influenced by hydrodynamic processes
16
(e.g., circulation and flushing), and chemical and biological processes (e.g., ammonification,
mineralization, decomposition, algae uptake, excretion, and mortality, etc.) in the estuary and
basin.
Color levels and concentrations of nitrogen, phosphorous and chlorophyll-a in
Charlotte Harbor exhibit pronounced salinity-related gradients which extend from the head
of the estuary to its mouth (Morrison, 1997). The water quality in the Caloosahatchee system
is more degraded than the water quality in the Myakka or Peace systems. Oxygen depletion
is common upstream of Franklin Lock. Nutrient and chlorophyll levels are high, and algal
blooms occur regularly in the tidal river (Estevez, 1998).
2.3.1 Nutrients
Nutrient availability is a key factor in the regulation of primary productivity in
estuarine and coastal water (Ketchum, 1967). Increased nutrient loads related to the urban
development of coastal basins have been implicated in estuarine nutrient enrichment,
increased phytoplankton productivity, and incresed phytoplankton biomass (Jaworski, 1981).
and in declines of seagrass communities (Orth and Moore, 1983).
The distribution of nutrients in the system is mainly the result of nutrient input from
rivers, freshwater and tidal flushing, and recycling processes in the estuary (McPherson and
Miller, 1990). The major factor that influences estuarine nutrient distribution is freshwater
inflow from rivers, which contributes substantial nutrient loads and flushes nutrients
seaward. NOAA estimates that Charlotte Harbor receives about 2,500 tons of nitrogen as
total Kjeldahl nitrogen, or TKN, and 1,000 tons of phosphorous per year. Relative to its
dimensions and flushing characteristics, phosphorous loads are high, signifying a nitrogen
limited system (Estevez, 1986).
17
Molar ratios of dissolved inorganic nitrogen to dissolved inorganic phosphorous were
below the Redfield ratio of 16, averaging 5.7 at Caloosahatchee River during two sampling
periods, which are from 1985 to 1989 and from 1994 to 1995 (Doering and Chamberlain,
1997). By contrast, total nitrogen to total phosphorous ratios were above 16, averaging 35
in Caloosahatchee River. With these results, dissolved inorganic N:P ratios (<16) suggest
that nitrogen could limit phytoplankton productivity in agreement with McPherson and
Miller (1990), while total N:P ration (>16) indicated that phosphorous could limit
productivity (Doering, 1996). In fact, nutrient addition studies conducted by FDER found
N and P to be co-limiting (Degrove, 1981)
The distributions of phosphorous in the system were nearly conservative and a
function of river phosphorous concentration, flow, and physical mixing (McPherson and
Millor, 1990). A large amount of phosphorous from the watershed is carried by freshwater
discharge into the tidal reaches of the Peace River, with subsequent dilution by the lower-
nutrient seawater entering the estuary from the Gulf of Mexico. Concentrations of total
phosphorous averaged about 0.08 mg/L in Pine Island Sound, 0.15 mg/L in the tidal
Caloosahatchee River, and 0.62 mg/L in the tidal Peace River according to USGS nutrient
data from 1982 to 1989 (McPherson and Miller, 1990).
Most of the nitrogen in the rivers and estuary is organic nitrogen (McPherson and
Miller, 1990). Organic nitrogen concentrations decreased over the salinity gradient,
indicating river input as a source. The relatively low concentrations of inorganic nitrogen
could limit plant growth in the estuary (McPherson and Miller, 1990). Ratios of nitrogen and
phosphorous constituents at the Peace River, Myakka River, and Upper Charlotte Harbor
determined by USGS nutrient data during 1975 through 1990 are shown in Table 2.1 (Pribble
IS
etal., 1998)
Table 2.1 Ratios of nitrogen and phosphorous constituents (Pribble et al., 1998)
RATIO Peace River Myakka River Upper Harbor
NH3 : TN
0.0327
0.0349
0.0338
N03 : TN
0.4378
0.0505
0.2442
ON:TN
0.5614
0.9096
0.7355
P04 : TP
0.9384
0.9371
0.9361
OP:TP
0.0616
0.0683
0.0650
Concentrations of ammonia were highly variable along the salinity gradient and were
in about the same range as concentrations in the rivers (McPherson and Miller, 1990).
Ammonia concentrations increased in the deeper water of Charlotte Harbor during summer
(Fraser, 1986). Ammonia enrichment probably was related to density stratification and to
low concentrations of dissolved oxygen in bottom waters (McPherson et al, 1996).
Concentrations of nitrate and nitrite nitrogen were nonconservative and decreased sharply
along the salinity gradient (McPherson and Miller, 1990). The sharp decline of the nitrate
and nitrite nitrogen in the low salinity regions indicates a substantial removal of nitrogen
from the water column due to biological uptake.
2.3.2 Dissolved Oxygen
Dissolved oxygen is critical for survival of plants and animals in fresh and salt water,
and a major constituent of interest in water quality study. Dissolved oxygen concentrations
in the near surface water of the system ranged from about to 6 to 8 mg/L during daylight
samplingin 1982 -84 (Stoker, 1986). Dissolved oxygen concentrations of near bottom water
of the estuary generally are lower than near surface concentrations.
Bottom water hypoxia (dissolved oxygen <2 mg/L) in Charlotte Harbor has been
reported periodically by the Environmental Quality Laboratory (EQL) since 1975 (Heyl.
19
1996). Dissolved oxygen concentrations were measured monthly from 1976 to 1984 in the
river mouth of Peace River near PuntaGorda (CH-006) (Fraser, 1986). The average monthly
near-surface concentrations declined from 8.5 to 6.7 mg/L from January to July and then
began to rise (Figure 2.3). Near-bottom average monthly concentrations at this area were
highest in February, declined slowly through May, and then declined more rapidly until July
(Fraser, 1986). The hypoxia conditions during summer are attributed to strong stratification,
which cause restricted reaeration, and to SOD. After breakup of the stratification, the
concentration increased from October to December.
F M
M J
A S O N
Figure 2.3 Average monthly concentration of dissolved oxygen in upper Charlotte
Harbor, site CH-006, 1976-84 (Fraser, 1986)
2.3.3 Phytoplankton
Phytoplankton is an important component of water quality processes and a major
primary producer in coastal and estuarine waters. The temporal and spatial variability of
20
phytoplankton productivity and biomass in Charlotte Harbor have been investigated by
Environmental Quality Laboratory, Inc (EQL) (1987). Phytoplankton productivity and
biomass (as chlorophyll _a) in the system are relatively low most of time. Productivity
ranged from 5 to 343 (mgC/m3)/h and averaged 59 (mgC/m3)/h from 1985 to 1986
(McPherson et al., 1990). Chlorophylljx concentrations ranged from 1 to 46 mg/m3 and
averaged 8.5 mg/m3. Both productivity and biomass were greater during summer near the
mouth of tidal rivers which has middle range salinity of 6 to 12 ppt (McPherson et al., 1990).
Phytoplankton productivity and biomass in the system are affected by freshwater
inflow that lowers salinity, increases nutrient availability, and reduces light penetration in the
water column. The nutrient rich colored water is diluted by seawater at middle range salinity
of 10 to 20 ppt, so that availability of light increases and sufficient nutrient concentrations
remain available from runoff, to stimulate productivity and growth of phytoplankton in these
areas (McPherson et al., 1990). Of the major nutrients for phytoplankton productivity,
inorganic nitrogen is in lowest supply and most critical in limiting phytoplankton
productivity and growth in the system (Fraser and Wilcox, 1981).
The composition of the phytoplankton in the system varied with location and season
(McPherson et al., 1990). Diatoms were dominant in 55 % of 289 phytoplankton samples
collected in the system between 1983 and 1984, cryptophytes in 35 %, cyanophytes (blue-
green algae) in about 6 %, dinoflagellates in about 4 %, and other classes in 1 % (McPherson
et al., 1996).
2.4 Sediment
Sediment quality data from Charlotte Harbor are scarce. Organic carbon, nitrogen
and phosphorous data exist in only two studies. Hwang (1966) obtained data in 1965 from
21
119 stations throughout the study area. The FDEP Coastal Sediment Contaminant Survey
obtained carbon, nitrogen phosphorous, and metal data in the sediment column from 33
stations during 1985 through 1989, in all areas except the Gulf of Mexico (Sloane, 1994).
Mean carbon/nitrogen ratios in the six different areas of Charlotte Harbor ranged from 8.9
at Lower Charlotte Harbor to 16.4 at Gasparilla Sound (Schropp, 1998). Despite data
limitation, the available data indicate that Charlotte Harbor is relatively free of sediment
contamination in comparison with Tampa Bay and Biscayne Bay, where sediment
contaminations are widespread and are present in higher concentrations (Schropp, 1998).
2.5 Light Environment
The amount of photosynthetically active radiation (PAR) in natural water is
fundamentally important in determining the growth and vigor of aquatic plants (McPherson
et al., 1996). Light is reflected, absorbed, and refracted by dissolved and suspended
substances in the water column, and by water itself. The controlling factors are chlorophyll-
a, dissolved substances, and non-algal particulate matter (Christian and Sheng, 2003).
Dissolved and suspended matter are the major causes of light attenuation in the
system: phytoplankton and chlorophyll-a are generally minor causes of attenuation
(McPherson and Miller, 1990). On average, non-chlorophyll suspended matter accounted
for 72% of light attenuation, dissolved organic matter (color) accounted for 21%,
phytoplankton chlorophyll for 4%, and water for the remaining 3% (McPherson et al., 1996).
Water color can cause light attenuation at tidal rivers because the source of the water color
is terrestrial dissolved organic matters. Dissolved organic matter has little effect on light
attenuation in much of the southern part of the estuary, where suspended matter is the major
cause of light attenuation (McPherson and Miller, 1996). The source of suspended matter
22
includes the bottom of the estuary, which consisted of very fine to fine sand (Hwang, 1966)
and organic detrital -material (McPherson and Miller, 1990), and the major rivers.
CHAPTER 3
HYDRODYNAMICS AND SEDIMENT TRANSPORT MODEL
Biological processes in an estuarine system are strongly affected by hydrodynamics
and sediment transport processes. Therefore, as the first step to study nutrient cycling in an
estuary, hydrodynamics and sediment transport processes must be investigated and
understood.
The three-dimensional hydrodynamics model CH3D (Sheng et al., 2001) and the
associated sediment transport model CH3D-SED3D (Sheng et al., 2000a) are used for
numerical simulations in this study. The model framework has been improved and modified
from earlier versions in order to develop an integrated model that couples hydrodynamics,
sediment and water quality dynamics. The code of the enhanced integrated model is
optimized to achieve more efficient coupling and more systematic structure. The detail
structure of enhanced model is explained in Appendix A as flow chart. The application of
the circulation and transport model to produce a detailed characterization of hydrodynamics
within system is the first step in the development of the integrated model of the system.
The CH3D model for Tampa Bay and Indian River Lagoon did not include the
simulation of temperature distribution. In this study, the temperature field is simulated by
solving the temperature equation with an air-sea interface boundary condition. Temperature
is an important factor for baroclinic circulation and water quality processes, in which almost
all reaction parameters in the nutrient cycle are a functions of temperature. Therefore,
23
24
improved simulation of temperature is expected to produce more accurate baroclinc and
water quality simulations.
3.1 Governing Equation
Hydrodynamics and sediment transport processes in estuaries are complicated, three-
dimensional and time-dependent. Mathematical descriptions of these processes generally
require simplifying assumptions.
3.1.1 Hydrodynamic Model
The governing equations that describe the velocity and surface elevation fields in
shallow water are derived from the Navier-Stokes equations. In general, four simplifying
approximations are applied: 1) the flow is incompressible, which results in a simplified
continuity equation, 2) the horizontal scale is much larger than the vertical scale such that
the hydrostatic pressure distribution is valid, 3) the Boussinesq approximation can be used
to simplify the treatment of the baroclinic terms, 4) the eddy-viscosity concept, which
assumes that the turbulent Reynolds stresses are the product of mean velocity gradients and
"eddy viscosity", can be employed. With the above assumptions, the continuity equation and
x- and y- momentum equations have the following form (Sheng, 1983):
du dv dw _
— + — + — = 0 (3.1)
dx dy oz
+
dz
du duu duv duw _ dC,
dt dx dy dz dx H \dx2 dy
dv dvu dvv dvw dC, ( d2v 32v| dfA oV
fd2u aV
J
d ( du
^
(3.2)
UVU UVV UVYV l/L, V V V r , w \ A v /TT\
dt dx dy dz ' dy {dx2 dy
J
dz\ dz
where u(x,y,z,t), v(x,y,z,t), w(x,y,z,t) are the velocity components in the
horizontal x- and y-directions, and vertical z-direction; t is time; g(x, y, t) is the free surface
25
elevation; g is the gravitational acceleration; and AH and \ are the horizontal and vertical
turbulent eddy coefficients, respectively.
In Cartesian coordinates, the conservation of salt and temperature can be written as
dS duS dvS dwS d
— + + + = —
dt dx dy dz dx
dT duT dvT dwT d
DH —
ox
d'
+ —
dy
dy
dz
+•
+ ■
■ + ■
dt dx dy dz
Kt
dT_
dx
+ ■
< dT
dy[ dy
+ ■
dz
DvTz
v V dz
(3.4)
(3.5)
where S is salinity; T is temperature; DH and KH are the horizontal turbulent eddy
diffusivity coefficients for salinity and temperature, respectively; and Dv and Kv are the
vertical turbulent eddy diffusivity coefficients for salinity and temperature, respectively.
Since the length scales of horizontal motion in estuarine systems are much greater
than those of vertical motion, it is common to treat vertical turbulence and horizontal
turbulence separately. In shallow estuaries, the effect of the horizontal eddy viscosities on
circulation are much smaller than the effect of the vertical eddy viscosity, although the
horizontal eddy viscosity is typically 2-3 orders of magnitude larger than the vertical eddy
viscosity.
Vertical turbulent mixing is an important process, which can significantly affect
circulation and transport in an estuary. Since turbulence is a property of the flow instead of
the fluid, it is essential to use a robust turbulence model to parameterize the vertical turbulent
mixing. In this study, the vertical eddy coefficients ( Av , Dv and Kv ) are computed from
a simplified second-order closure model developed by Sheng and Chiu (1986), and Sheng
and Villaret (1989).
26
Various form forms of the equation of state can be used. The present model uses the
equation given by (Eckart, 1958)
P
P =
(a + 0.698P)
P = 5890 + 387/ -0.357r2 + 3S (3.6)
a = 1779.5 + 1 1.257 -0.04757/2- (3.8 + 0.017)5
where T is in °C , S is in ppt and p is in g I cm .
The complete details of model equations in the curvilinear boundary-fitted and sigma
coordinates have been derived (Sheng, 1989), and are presented in Appendix B.
3.1.2 Sediment Transport Model
The suspended sediment model includes the advection-diffusion processes, which are
computed by the hydrodynamics model, as well as such processes as erosion, deposition,
flocculation, settling, consolidation, and entrainment (Sheng, 1986; Metha, 1986).
The governing equation that represents the transport of suspended sediments is given
by:
dc due dvc d(w + w)c d
— + + ^ +
dt dx dy dz dx \ H dx ) dy\ dy
B*
+ ■
'B^
K dz
c)z
(3.7)
where c is the suspended sediment concentration, w5 is the settling velocity of suspended
sediment particles (negative downward), BH is the horizontal turbulent eddy diffusivity, and Bv
is the vertical turbulent eddy diffusivity.
Four simplifying approximations are implied in the above equation: 1) the concept
of eddy diffusivity is valid for the turbulent mixing of suspended sediments, 2) the suspended
sediment dynamics are represented by the concentrations of two particle size groups (Sheng
et al., 2002a), 3) the suspended sediment concentrations are sufficiently small that all
27
particles follow turbulent eddy motions, 4) the SSC is sufficiently low that non-Newtonian
behavior can be neglected.
In this study, the determination of settling, flocculation, deposition, erosion,
flocculation, and consolidation processes are based on the previous work of Sheng and Lick
(1979), Sheng(1986), Metha (1989), Sheng et al. (1990), Chen and Sheng (1994) and Sun
and Sheng (2001).
3.2 Boundary and Initial Conditions
Obtaining solutions for Equations (3.1) Through (3.7) requires the specification of
appropriate boundary and initial conditions.
3.2.1 Boundary Conditions
The boundary conditions at the free surface (a=0) in a non-dimensional, vertically
stretched, boundary-fitted coordinate system are:
du H „
For Hydrodynamics: \
dcr Ey
*" 3.7 E, "
For Salinity: = 0 (3.8)
dcr
dT H Pr,
v
For Temperature: — — = qT
dcr E
v
Dv fci ^
For Sediment: w.c. + —^—^ = 0
" ' H dcr
where Ev is a vertical Ekman number and Prv is a vertical Prandtl number; The Cartesian
wind stress, t", is calculated using
Tw =pC,v Ju2 +v2 (3.9)
y fa as w v w w v '
28
where pa is the air density (0.0012 g/cm3), uw and vw are the components of wind speed
measured at some height above the sea level. Cds, the drag coefficient, is given as a function
of the wind speed measured at 10 meters above the water surface by (Garrat, 1977).
Cds = 0.001(0.75 + 0.067 Ws) (3.10)
Boundary conditions, if specified in a Cartesian coordinate system, such as wind
stress, must be transformed before being used in the boundary-fitted equations. For example,
the surface stress in the transformed system is given by
3£ 3£
dx dy
drj drj
r<-ar-+*r<
r,„=-^,+— r,v (3-u)
ox oy
The second surface boundary condition is the kinematic free surface boundary
condition which states
BC d£ dC
w = -?- + u-^ + v^f- (3.12)
dt ox oy
The boundary conditions at the bottom in a non-dimensional, vertically stretched (o
= -1), boundary-fitted coordinate system are
For Hydrodynamics:
\ -£- = —T^ =-^HrZrCd [guu2b + 2gl2ubvb + g22v;~\ub
da Ev "' A,
A ^L = JLt-=£j-HZC
da Ev \r
guu; + 2gnubvb + g22v2b~\vb
For Salinity: ^ = ° (313)
dcr
dT n
For Temperature: — — = U
da
Dv dc
For Sediment: w.ci + — -— -1- = A _ Et
si ' H da ' '
29
where ub and vh are the contravarient velocity components at the first grid point above the
bottom. Ej is the erosion rate and D, is the deposition rate for sediment group i. Cd is drag
coefficient which is a function of the size of bottom roughness elements, z0, and the height
at which uh is measured, so long as z, is within the constant flux layer above the bottom. The
drag coefficient is given by (Sheng, 1983)
cd =
K
(3.14)
ln(z,/z0)
where k=0.4 is the von Karman constant, z0=k/30 and ks is the bottom roughness.
Along the shoreline where river inflow may occur, the conditions are generally
u = u(x,y,(J,t)
v = v(x,y,cr,t)
w = 0
(3.15)
S = S(x,y,cr,t)
T = T(x,y,cr,t)
c, =ci(x,y,a,t)
Contravarient velocity components provide lateral boundary conditions similar to
those in Cartesian systems (x,y). Along solid boundaries, the normal velocity component
must be zero to satisfy the no-slip condition. In addition, normal derivatives of salinity,
temperature and suspended sediment concentration are assumed to be zero. When flow is
specified at a boundary, the normal velocity component is prescribed. Tidal boundary
conditions are specified using water level, (, directly When tidal boundary conditions are
given in terms of C the normal velocity component is assumed to be of zero slope, while
tangential velocity component may be either zero, of zero slope, or computed from the
momentum equations. During an ebb tide, the concentrations of salinity, temperature, and
suspended sediment flowing out are calculated using a 1-D advection equation while during
30
a flood tide the offshore concentrations are generally prescribed as either fixed or time
varying.
3.2.2 Initial Conditions
To initiate a simulation, the initial spatial distribution of (,, u, v, w, S, T, and c, must
be specified. When these values are unknown, "zero" initial fields can be used. When these
values are known at a limited number of locations, an initial field can be generated by spatial
interpolation. In the principle, the interpolated field should satisfy the conservation equation
for a particular variable. For practical simulations, a "spin-up" period is required to damp
out transients caused by the initial condition, which does not satisfied conservation. The
length of a spin up period is variable and depends on such factors as basin size, flushing
time, and current velocity.
3.3 Heat Flux at the Air-Sea Interface
The ocean receives energy through the air-sea interface by exchange of momentum
and heat. The temperature of ocean waters varies from place to place and from time to time.
Such variations are indications of heat transfer by currents, absorption of solar energy, and
loss by evaporation. The size and character of the temperature variations depends on the net
rate of heat flow into or out of water body (Pickard and Emery, 1990). The transfer of heat
across the air-sea interface determines the distribution of temperature in the ocean as a
surface boundary condition for the temperature equation, as follow:
dT
PoKv — = qT (3-16)
az
where qT is the net heat flux across the surface water.
To estimate the net heat flux at the air-sea interface, it is necessary to consider the
following processes: net heat flux across the surface, qT\ heating by incoming solar radiation
31
(insolation), qs ; warming by conductive heat exchange across the surface (sensible heat
flux), qh ; cooling by net outgoing long wave radiation, qh ; heat loss as water evaporated
(latent heat flux), q, ; heat flux from precipitation. Other source of heat flow, such as that
from the earth's interior, change of kinetic energy of waves into heat at the surf; heat from
chemical or nuclear reactions, are small and can be neglected. The net heat exchange at the
surface can be divided into four terms:
qT = <2s-<ib-<ik-<ii (3-17)
Direct measurement of these fluxes is the best way to provide the net heat flux at the
air-sea interface. However, directly measured air-sea fluxes are only available at very few
stations to allow calculation of air-sea interface over a large area. Instead, directly measured
air-sea fluxes are used for developing, calibrating, and verifying the parametric formulae for
estimating the fluxes from the primary variables including wind speed, air temperature, water
temperature, and cloud cover (Taylor et al., 2000). These required parametric formulae are
then used to compute heat fluxes over a large area.
3.3.1 Short- Wave Solar Radiation
The main source of heat flux through the air-sea interface is short-wave solar
radiation, gv (incoming solar radiation), received either directly or by reflection and scattering
from clouds and the atmosphere. The incoming solar radiation is based on the empirical
formula of Reed (1977):
qs = S sin y ( A + fisin y)(\ - 0.62n - 0.0019r)(l - a) (3.18)
where S is solar constant =1353 w/m2
Y is solar elevation angle
n is cloud cover = 0-1
32
a is the albedo = 0.06
A and B are empirically determined coefficients for each category of reported total
cloud amount and the solar elevation
The solar elevation angle was computed to the nearest 0. 1 ° from date and time for the
latitude and longitude of the study area, using the following equations (Miller and
McPherson, 1995):
360
<p = (d-l)
365.242
J = 12 + 0.1236sin^-0.0043cos^ + 0.1538sin2^ + 0.0608cos2^
<t = 279.9348 + p + 1.9148 sin p-0.0795cos^ + 0.00199 sin 2^ + O.OO16cos20
Y = \5(t-S)-A (3.19)
k = arcsin (0.39785077 x sin a)
sin /3 - sin /sin k + cos y cos /f cos Y
where (p is the angular fraction of the year, in degree; d is Julian date; 6 is true solar noon,
in hours; T is the solar hour angle, in degrees; T is the Greenwich Mean Time, in hours; X
is the longitude, in degrees; o is an estimate of the true longitude of the sun, in degrees; K is
the solar declination, in degrees; y is the latitude, in degrees; and P is the solar elevation
angle, in degrees.
3.3.2 Long-Wave Solar Radiation
The back radiation term, qh, is the net amount of energy lost by the sea as long-wave
radiation. The amount of long wave flux is dependent on surface water temperature,
atmospheric temperature, humidity and cloud cover (Clark et al, 1974)
qb = eoT* (0.39 - 0.05e05 ) (l - An2 ) + 4eaT* (Ts - Ta ) (3.20)
where e is the emittance of the sea surface = 0.98
o is Stefan-Boltzman constant = 5.673 x 10"8 W/m2
33
e is the water vapor pressure
Ta and Ts are the air and sea temperatures in K.
A is a cloud cover coefficient which varies with latitude
The Antoine constants (http://www.owlnet.rice.edu/~ceng301/31.html) give the vapor
pressure as a function of temperature so that the approximate value for water vapor pressure
is:
( 3985.44 A
16.5362-
e = exp
Ts -38.997
where Ts is the surface water temperature (°K).
(3.21)
A theoretical calculation of the mean values of the coefficient X for different latitudes
has been made by M. E. Berliand (1952). In this calculation, the mean frequency of clouds
of different layers at each latitude is taken into account. The values obtained for the
coefficient X are given in Table 3.1 (Budyko, 1974). The reduction of values of this
coefficient in low latitudes is explained mainly by a higher mean altitude of clouds in these
regions.
Table 3.1 Mean latitudinal values of the coefficient X
*
75°
70°
65°
60°
55°
50°
45°
40°
X
0.82
0.80
0.78
0.76
0.74
0.72
0.70
0.68
d>
35°
30°
25°
20°
15°
10°
5°
0°
X
0.65
0.63
0.61
0.59
0.57
0.55
0.52
0.50
3.3.3 Sensible and Latent Heat Fluxes
The most significant part in terms of heat transfer from the sea to the atmosphere is
latent heat flux. The rate of heat loss is equal to the rate of vaporization times the latent heat
of evaporation. Sensible heat flux is due to temperature gradient in the air above the sea.
The rate of loss or gain of heat is proportional to the temperature gradient, heat conductivity
34
and the specific heat of air at constant pressure. Sensible and latent heat fluxes are estimated
by using the bulk aerodynamic equations (Mellor, 1996):
qk=pcpCHUw(Ts-Ta) (3.22)
qi=pCELUl0(hs-ha) (3.23)
where p is the density of air
CH and CE are transfer coefficients for sensible and latent heat respectively
UI0 is the wind speed at 10 m height above surface
Ts and Ta are the sea surface and air temperatures
cp is the specific heat of air, 1.0048 x 103 J/kg°K
L is the latent heat of evaporation, 2.4 x 106 J/kg
hs and ha are specific humidities at the surface and at a 10 m height above the surface
The air density is determined from the ideal gas equation in the form,
p = pJ(RTa ) , where Pa is air pressure and R is universal gas constant, 287.04 J/kg °K.
The traditional estimates of CH and CE over the ocean tend to support a fairly constant value
over a wide range of wind speed. Smith (1989) recommended a constant "consensus" value
CE = (1.2 ± 0.1)xl0"3 for winds between 4 and 14 m/s. DeCosmo et al. (1996) also suggested
a near constant value with CE = (1.12 ± 0.24)xl0 3 for winds up to 18 m/s. For the Stanton
number, CH, Friehe and Schmitt (1976) obtained slightly different values for unstable and
stable conditions, 0.97 xlO"3 and 0.86 xlO'3 respectively. Smith (1989) suggested CH = 1.0
x 10 3. CE- 1.2 xlO"3 and CH = 1.0 x 10"3 were used for this study.
The formulae used for air specific heat, cp , and latent heat for evaporation, L, have
been taken from Stull (1988),
cp =1004.67 + 0.84^ (3.24)
35
L = 106 [2.501 -0.00237(7; -273.16)] (3.25)
An empirical relation (based on the Clausius-Clapeyron theory) provides the
• i ir.(0.7859+O.O3477r)/(l+O.0O4l2r) , » „-„_-„!
saturation, water vapor partial pressure, es = 10v mb. A general
relation between specific humidity and partial pressure is q - (Q.622e/P)/(l - 0.31Se/P)
, where P is the atmospheric pressure (Mellor, 1996). Specific humidities at surface and at
a 10 m height above the surface, hs and ha, can be calculated using these empirical formulae
with Ts and Ta respectively.
CHAPTER 4
WATER QUALITY MODEL
In this chapter, a water quality model for simulating water quality processes occurring
in the Charlotte Harbor estuarine system in both the water and sediment columns is
presented. This is an extension of an earlier model developed to simulate water quality for
Lake Okeechobee (Sheng et al., 1993; Chen and Sheng, 1994), Tampa Bay (Yassuda and
Sheng, 1996) and the Indian River Lagoon (Sheng et al., 2001 and 2002). These models
include the effect of sediment transport on nutrient dynamics through the explicit use of a
sediment transport model and the incorporation of a nutrient resuspension flux.
The water quality model incorporates on the interactions between oxygen balance,
nutrient dynamics, light attenuation, temperature, salinity, phytoplankton and zooplankton
dynamics. To develop the water quality model, the mass conservation principle can be
applied to each water quality parameter, as it relates to phytoplankton and zooplankton
dynamics, nitrogen and phosphorous cycles, and oxygen balance.
The nitrogen and phosphorous cycles in an estuarine system, are modeled through a
series of first order kinetics. Nutrient concentrations in the estuarine system are constantly
changing in time and space due to loading from rivers, exchanges with the ocean, seasonal
climatic changes, biogeochemical transformations, hydrodynamics, and sediment dynamics.
Nitrogen species include ammonia nitrogen (NH3), soluble ammonium nitrogen (NH4),
nitrate and nitrite nitrogen (N03), particulate ammonium nitrogen (PIN), soluble organic
36
37
nitrogen (SON), particulate organic nitrogen (PON), phytoplankton nitrogen (PhyN), and
zooplankton nitrogen (ZooN). Similar to the nitrogen species, phosphorous species include
soluble reactive phosphorous (SRP), particulate inorganic phosphorous (PIP), soluble organic
phosphorous (SOP), particulate organic phosphorous (POP), phytoplankton phosphorous
(PhyP), and zooplankton phosphorous (ZooP).
Phytoplankton kinetics are the central part of this water quality model, since the
primary water quality issue in the estuarine system is eutrophication (Boler et al., 1991).
Phytoplankton population is a complex variable to measure in the field. However, the lack
of data on each specific species prevented a more detail characterization, the entire
phytoplankton community is represented in this study by a single state variable and
quantified as carbonaceous biomass. Chlorophyll _a concentrations, for comparison with
observations, are obtained through division of computed carbonaceous biomass by the
carbon-to-chlorophyll_a ratio.
The oxygen balance couples dissolved oxygen to other state variables. Reaeration
through the air-sea interface, and phytoplankton production during photosynthesis are the
main source for oxygen. Oxidation, nitrification, respiration, mortality and SOD reduce
oxygen in the system. Oxidation of organic matter and carbonaceous material, respiration by
zooplankton and phytoplankton, and oxygen consumption during the nitrification process are
collectively grouped into the CBOD (Carbonaceous-Biogeochamical Oxygen Demand)
variable, which is a sink for dissolved oxygen (Ambrose et al., 1994).
The methods of coupling with hydrodynamic and sediment transport models, the
simulated parameters, the assumptions, the chemical/biological processes of the CH3D water
quality model (CH3D-WQ3D) was compared with those for existing water quality models,
38
specially Water Quality Analysis ans Simulation Program (WASP), the integrated
Compartment water quality model developed by the US Army Corps (CE-QUAL-ICM) in
Appendix C.
Temperature, salinity, and light are important parameters that effect the rate of
biogeochemical reactions. Most of the transformation processes in the nutrient cycle are
affected by temperature. Salinity also influences DO saturation concentration and is used in
the determination of kinetics constants that differ in saline and fresh water. Light intensity
affects the photosynthesis process and thus algae growth rate.
The water quality model is enhanced to include a reaeration model and a sediment
oxygen demand (SOD) model and newly coupled with temperature model and physics-based
light attenuation model. Applying enhanced water quality model and a more accurate light
model and temperature model could improve the vertical distribution and daily fluctuation
of both phytoplankton and dissolved oxygen. Furthermore, the bottom water hypoxia at
upper Charlotte Harbor which is caused by SOD and vertical stratification can be reproduced
and analyzed in this study.
4.1 Mathematical Formulae
The water quality equations are derived from a Eulerian approach, using a control
volume formation. In this method, the time rate of the concentration of any substance within
this control volume is the net result of (i) concentration fluxes through the sides of the
control volume, and (ii) production and sink terms inside the control volume. The
conservation equation for each of the water quality parameters is given by:
M + V.(fl«) = V-[DV(fl«)] + Q,
(i) («) (Hi) (iv)
(4.1)
39
where (I) is the evolution term (rate of change of concentration in the control volume), (ii)
is the advection term (fluxes into/out of the control volume due to the advection of the flow
field), (iii) is the diffusion term (fluxes into/out of the control volume due to turbulent
diffusion of the flow field), and (iv) is the sink and source term, representing the kinetics and
transformations due to sorption/desorption, oxidation, excretion, decay, growth, and bio-
degradation. In the finite difference solution of the water quality model, the advection and
horizontal diffusion terms are treated explicitly, whereas the vertical diffusion and
biogeochemical transformations are treated implicitly. The detail descriptions of numerical
solution technique are described in Appendix D. The water quality equations in the
curvilinear non-orthogonal boundary fitted system (£, X], o) are given by:
dt
HSCV da
3 < a* ^
V
'V -s
da
-R,
d(H(D)(pt
da
/?„
**\*Z
—(yfg~0Hu<pi) + — (Jg~0Hv(pi)
+
^CHyJSo
^
goH8
+ -
^CH V #0
drj
8oHS
drj
21 d<P,
+ Jg0Hgl
+ JgoHg'
drj
(4.2)
J]
+ Q,
where (pi represents any water quality parameter, -^g^ x% me Jacobian of horizontal
transformation, (g ,g ,g ) are the metric coefficients of coordinate transformation,
and Q. represent biogeochemical processes. Equation (4.2) is in dimensionless form and the
dimensionless constants are defined in Appendix B.
In the following sections, the biogeochemical processes controlling the sink/source
term of Equation (4.2) will be discussed in detail for nutrient dynamics, zooplankton and
40
phytoplankton dynamics, and oxygen balance in system.
4.2 Phytoplankton Dynamics
The overall water quality in the system is markedly influenced by the dynamics of
zooplankton and phytoplankton communities (Boleretal., 1991). Phytoplankton dynamics
and nutrient dynamics are closely linked, since nutrient uptake during phytoplankton growth
is the main process to remove dissolved nutrients from the water, and phytoplankton and
zooplankton respiration and mortality are major components of nutrient recycling. The
diurnal variation of dissolved oxygen is related with photosynthetic oxygen production
during the day and oxygen consumption due to phytoplankton respiration during the night.
4.2.1 Modeling Approach
Phytoplankton kinetics are represented by growth, respiration, non-predator-mortality,
grazing by zooplankton, and a settling term. The phytoplankton sources and sinks in the
conservation equation can be written as:
dPhvC ( d s\
^1= Ma-K^-Kas+-WSaigae -PhyC-^ZooC (4.3)
at \ az J
Where PhyC is phytoplankton biomass, expressed as carbon (gCm3); /Ja is phytoplankton
growth rate (1/d); Kas is respiration rate (1/d); K^ is non-predator mortality (1/d); WSalgae is
the phytoplankton settling velocity(m/d); ZooC is zooplankton biomass (gCm3); and /uz is
zooplankton growth rate (1/d).
Phytoplankton growth is determined by the intensity of light, by the availability of
nutrients, and by the ambient temperature. Light limitation is formulated according to the
photo inhibition relationship (Steel, 1965). The quantity of the growth limitation factor for
nutrients is related with a half-saturation constant. The half-saturation constant refers to the
concentration of the nutrient at which the growth rate is one half its maximum value. This
41
results in a hyperbolic growth curve. An exponential increasing function is applied for
temperature limitation..
The minimum formation approach has been used to combine the limiting factor of
light and each limiting nutrient. The minimum formation is based on "Liebig's law of the
minimum" which states that the factor in shortest supply will control the growth of algae
(Bowie et al., 1980)
Va=(Ma)aaK-f(T)'M)'f(N,P)
(4.4)
= (/0™C2°-min
— exp
L
Is J
NH4 + N03 SRP
Hn+NH4 + N03 Hp+SRP
where OuJmax is the phytoplankton maximum growth rate (1/day); d is temperature
adjustment coefficient; Tis temperature (°C); /is the light intensity, calculated by the light
attenuation model; Is is the optimum light intensity for algae growth; Hn is half saturation
concentration for nitrogen uptake (gN — 3); Hp is half saturation concentration for
phosphorous uptake (gP — 3); NH4 is ammonium concentration (gN — 3); N03 is nitrate
concentration (gN ~~ 3); and SRP is soluble reactive phosphorous concentration (gP — 3).
Respiration and mortality are considered to be an exponentially increasing functions
of temperature:
T T (4.5)
where (KaJTr and (KM)Tr are respiration and mortality rate at Tr (1/day); and Tr is reference
temperature of respiration and mortality.
For phytoplankton, literature values of algae settling velocity, which account for the
limited vertical motion of these organisms will be used.
Zooplankton are included in water quality models primarily because of their effects
42
on algae and nutrients. Phytoplankton and zooplankton dynamics are closely tied through
predator-prey interaction. Phytoplankton dynamics are of major concern in this study while
no attempt is made to investigate zooplankton dynamics due to lack of zooplankton data.
Zooplankton is only considered as the predators of phytoplankton, utilizing their available
biomass as food supply. Zooplankton kinetics, influenced by growth, respiration and
mortality, are represented in a source and sink term as (Bowie, 1985):
l^-fa-K.-K.yZooC (4.6)
at
where //, is zooplankton growth rate (1/day); Kas is respiration rate (1/day); and K^ is
mortality (1/d)
Zooplankton growth is represented by a temperature-dependent maximum growth
rate, which is limited by phytoplankton availability:
K '~ ' Hphr+PhyC-Trsph:1
where (/uz)m.M is the zooplankton maximum growth rate (1/day); 6\s temperature adjustment
coefficient; Hphy is half saturation concentration for phytoplankton uptake (gC — 3); and Trsphy
is threshold phytoplankton concentration for zooplankton uptake (|ig/l).
4.2.2 Relationship between Phytoplankton and Nutrients
Phytoplankton biomass is quantified in units of carbon. In order to express the effects
of phytoplankton on nitrogen and phosphorous, the ratio of nitrogen-to-carbon and
phosphorous-to-carbon in phytoplankton biomass must be specified. Global mean values of
these ratios are well known (Redfield et al., 1966). The amounts of nitrogen and
phosphorous incorporated in algae biomass is quantified through a stoichiometric ratio.
Thus, total nitrogen and total phosphorous in the model are expressed as:
43
TotN = NH. + NH, + NO, + SON + PON + PIN + Anc ■ PhyC + Anc ■ ZooC
4 J (4.8)
TotP = SRP + SOP + POP + PIP + Ape ■ PhyC + Ape ■ ZooC
where TotN is total nitrogen (gN ~3); NH4 is dissolved ammonium nitrogen (gN ~3); NH3
is ammonia nitrogen (gN ~3); N03 is nitrate and nitrite nitrogen (gN ~3); SON is soluble
organic nitrogen (gNm~3); PON is particulate organic nitrogen (gN _3); PEN is particulate
inorganic nitrogen (gN ~3); Anc is Algae nitrogen-to-carbon ratio (gN/gC) ; TotP is total
phosphorous (gP ~3); SRP is soluble reactive phosphorous (dissolved phosphate) (gPrn 3);
SOP is soluble organic phosphorous (gP-3); POP is particulate organic phosphorous (gP-3);
PIP is particulate inorganic phosphorous (gP _3); and Ape is algae phosphorous-to carbon
ratio (gP/gC).
The connection between the carbon, nitrogen and phosphorous cycle is shown in
Figure 4.1. Phytoplankton uptakes dissolved ammonium nitrogen, nitrate and nitrite
nitrogen, and soluble reactive phosphorous during production and releases dissolved
ammonium nitrogen, soluble reactive phosphorous and organic nitrogen, organic
phosphorous during respiration and mortality processes. Zooplankton has similar kinetic
processes as phytoplankton. The measured phytoplankton as algae mass per volume, was
converted to phytoplankton carbon with a algae to carbon ratio. The amounts of nitrogen and
phosphorous from phytoplankton can be converted with a nitrogen to carbon ratio and a
phosphorous carbon ratio, respectively.
4.3 Nutrient Dynamics
Nutrients are essential elements for life processes of aquatic organisms. Nutrients
of concern include carbon, nitrogen, phosphorous, silica and sulfur. Among these nutrients,
the first three elements are utilized most heavily by zooplankton and phytoplankton. Since
carbon is usually available in excess, nitrogen and phosphorous are the major nutrients
44
regulating the ecological balance in an estuarine system. Nutrients are important in water
quality modeling for several reasons. For example, nutrient dynamics are critical
components of eutrophication models since nutrient availability is usually the main factor
controlling algae bloom. Algae growth is typically limited by either phosphorous or nitrogen
(Bowie et al., 1980). Details on the nutrient dynamics including all the equations used by
the water quality model to calculate the nitrogen and phosphorous, can be found in
Appendix-E.
Excretion
NH3
Volatilization
I Sorption/
Desorption w i r 1 1
Mortality
Ammonification
*■
SON
Sorption/
Desorption
NITROGEN
A
Nitrification
NH4
N03
PhyN=ar ,'PhyC
CARBON
i,
Excretion
Uptake
Mortality
Mortality
»- PON
Mortality
Uptake
Phytoplankton (PhyC)
Uptake
PhyP^ 'PhyC
Uptake
PHOSPHOROUS
Excretion
SRP
n i 4
PIP
Sorption/
Desorption
Zooplankton (ZooC)
Mortality
•H POP r*
Mortality
Mortality
Mineralization
SOP
Sorption/
Desorption
Mortality
Excretion
Figure 4.1 The connection between nitrogen, phosphorous and carbon cycle
45
4.4 Oxygen Balance
Dissolved oxygen (DO) refers to the volume of oxygen contained in water. Five state
variables participate in the dissolved oxygen balance: phytoplankton carbon (PhyC),
ammonia (NH4), nitrate (N03), carbonaceous biochemical oxygen demand (CBOD), and
dissolved oxygen (DO). A summary is illustrated in Figure 4.2. The methodology for the
analysis of dissolved oxygen dynamics in natural water, particularly in streams, rivers, and
estuaries is reasonably well-developed (O'Connor and Thomann, 1972). Dissolved oxygen
evolution depends on the balance between production from photosynthesis, consuming from
respiration and mortality, and exchanges with the atmosphere and sediment.
The main physical mechanisms influencing DO concentration are horizontal and
vertical dispersion and diffusion. Vertical diffusion occurs across the air-sea interface as a
function of wind, waves, currents, and DO saturation rate. Laboratory experiments show that
the bottom shear stress controls the dissolved oxygen diffusive layer thickness and the flux
at the sediment-water interface (Steinberger andHondzo, 1999). Slow oxygen diffusion rates
and high oxygen demand by sediment results in a thin aerobic layer. Two distinct sediment
zones are created in the sediment column: an aerobic layer and an anaerobic layer. The
thickness of the aerobic soil zone is influenced by oxygen concentration in the overlying
water column, and concentration of the reduced compounds in the anaerobic soil zone. The
model dissolved oxygen cycle includes the following processes
1 ) Reaeration
2) Carbonaceous oxygen demand (CBOD)
3) Nitrification
4) Sediment oxygen demand (SOD)
46
5) Photosynthesis and respiration
Carbonaceous Oxygen Demand
Zooplankton
Oxidation
Mortality
K^'ZOOC
Haunt + DO
CBOD
CBOD
1
Mortality
K^'PHYC
c
1
1
^
f
r
Phytoplankton
dz
i Diffusion
CBOD
Oxidation
Denitrification
DO
»™» +DO
-CBOD K,
H ,»;)
«„,,+ DO
-NO 3
Dissolved Oxygen
Air
Oxidation
KD — CBOD
"HiM + DO
Reaeration K^DOg-DO)
Phytoplankton
DO
Photosynthesis & Respiration
,->•
Nitrification
6f^_22_AW4
14 mHxrT+DO
1(1.3-0.
3P.)^-(^+^.)l*PWC»«,
Sediment Oxygen Demand
KO2+D0
Water column
* Diffusion
DO
Nitrification
DO
14 " H^+DO
NH4
Aerobic Layer
Figure 4.2 DO and CBOD cycles
Reaeration
Reaeration is the process of oxygen change between the atmosphere and sea surface.
Typically, dissolved oxygen diffuses into surface waters because dissolved oxygen levels in
most natural waters are below saturation. However, when water is super-saturated as a result
of photosynthesis, dissolved oxygen returns the atmosphere. Dissolved oxygen saturation
in seawater is determined as function of temperature and salinity (APHA, 1985)
47
1.575701xl05 6.642308xl07 1.2438x10'° 8.621949x10"
In DO. =-139.34411 + — - + ~3 =5
r
(4.9)
Sa I , 1.9428x10
3.1929x10- + •
, 1.9428x10 3.8673xl03^
1.80655
T2
where DOs is equilibrium oxygen concentration, mg/1, at standard pressure
T is temperature, °K, °K = °C + 273.150
Sa is salinity, ppt
The reaeration process is modeled as the flux of dissolved oxygen across the water
surface:
V^- = KAEAs(DOI-DO) (4-10)
at
where V and As are volume and surface area of the water body
In case where the air-sea interface is not constricted, the volume is V = As ■ Az . The
equation for reaeration can be expressed as
^. = ^AL(D0 -DO) (4.11)
dt Az V v
where DO is dissolved oxygen concentration (mg/1); KAE is reaeration coefficient (m/day).
Many empirical formulas have been suggested for estimating reaeration rate
coefficient specially in the river. Bowie et al. (1985) have reviewed thirty-one reaeration
formulas, and have tried to evaluate the performance of the each formulas. Most formulae
have been developed based on hydraulic parameters, most often depth and velocity. This
review of stream reaeration has shown that no one formula is best under all conditions, and
depending on the data set used, the range of the reaeration coefficients in the data set, and
error measurement selected, the best formula may change. Among these formulas, the most
48
common method of simulating reaeration in rivers is the O'Connor-Dobbins formula. This
method has the widest applicability being appropriate for moderate to deep streams with
moderate low velocities. With approximately 2.09xl0"5 cm2/s diffusivity of oxygen in
natural waters, the O'Connor-Dobbins formula can be expressed as
jjO.5
Ku=3.93— (4.12)
For standing water, such as lake, impoundments, and wide estuaries, wind becomes
the predominant factor in causing reaeration. The oxygen-transfer coefficient itself can be
estimated as a function of wind speed by a number of formulas. Chapra (1997) compared
four common wind-dependent reaeration formulas: Broeckeret al. (1978), Banks andHerrera
(1977), O'Connor (1983), and Wanninkhof et al. (1991). The comparison shown in Figure
4.3 show all these methods except Broeckeret al.'s have similar reaeration coefficients when
wind speed is less than 5 m/s. When wind speed is greater than 5 m/s, Bank and Herrera's
formula produce the middle range of reaeration coefficient among these three formulas.
This formula uses various wind dependencies to attempt to characterize the difference
regimes that result at air-water interface as wind velocity increase (Banks 1975; Banks and
Hen-era, 1977).
Kl=0.72SU°w5-03llUw + 0mi2Ul (4.13)
Since estuary gas transfer can be affected by both water and wind velocity, effort to
determine reaeration in estuaries combines elements of current and wind-driven approaches.
Thomann and Fitzpatrick (1982) combined the two approaches for estuaries affected by both
tidal velocity and wind,
KAE = 3.93J^ + 0.728£/° 5- 0.3 17£/„ -0.0372£/H2 (4.14)
49
where U0 is depth averaged velocity (m/s); H is a depth (m); Uw is wind speed (m/s).
For the Charlotte Harbor estuarine system, The aeration coefficient is assumed to be
proportional to the water velocity, depth, and wind speed following Thomman and
Fitzpatrick (1982).
10r-
9 -
8 -
7 -
n
? 5
* 4
3
2
1
0
Broeckeretal. (1978) ,'
Wanninkhofetal(1991).
Banks and
Herrera(1977)
4 6
Uw (m/s)
Figure 4.3 Comparison of wind-dependent reaeration formulas.
Carbonaceous Oxygen Demand
The use of carbonaceous oxygen demand (CBOD) as a measure of the oxygen-
demanding processes simplified modeling efforts by aggregating their potential efforts
(Ambrose et al., 1994). Oxidation organic matter, nitrification, non-predatory mortality and
respiration by zooplankton and phytoplankton are nitrogenous-carbonaceous-oxygen-
demand, collectively combined into the state variable CBOD.
50
The kinetic pathway of CBOD is represented in the source term of the equation as
(Ambrose et al., 1994):
For water column:
?-CBOD = -*-
dt dz
IcBOD = -f [wsCB0D • (1 - fdCB0D ) ■ CBOD] - KD ■ °° ■ CBOD
at dz tl r-o^-r UU
(4.15)
--■ — ■KDN ^ NO,+ Aoc(Kar PhyC + KaZooC)
4 14 Hno3+DO V
For sediment column:
IcBOD = +^-[wsCBOD ■ (1 - fdCB0D ) ■ CBOD] - KD ■ °° ■ CBOD
(4.16)
--■ — ■KDN ^ NO,
4 14 Hnu3 + DO
where fdCB0D corresponds to the fraction of the dissolved CBOD; wsCBOD is the settling
velocity for the particulate fraction of CBOD ; KD is oxidation coefficient which is a
temperature function; HCB0D is a half saturation constant for denitrification; KDN is
denitrification constant; Hm3 is a half saturation rate for denitrification; and Aoc is oxygen-
carbon ratio (g02/gC).
The consumption of oxygen, as a function of water column CBOD decay, can be
expressed as CBOD oxidation;
— DO = KD — CBOD (4.17)
dt HCB0D + DO
Note that non-oxidative processes such as settling, denitrification, and mortality do
not contribute to dissolved oxygen depletion, and are not included in the expression.
Nitrification
The transformation of reduced forms of nitrogen to more oxidized forms
(nitrification) consumes oxygen. Although nitrification is also a nutrient transformation, this
section addresses oxygen consumption. First order kinetics is the most popular approach for
51
simulating nitrification in natural system:
±DO—£k„— ^— -NH4 (4.18)
dt 14 m Hnil+DO
where KNN is nitrification rate which is a function of temperature; and Hnil is the half-
saturation constant for the bacteria growth.
Sediment Oxygen Demand
Sediment oxygen demand is a dissolved oxygen flux at water/sediment interface due
to the oxidation of organic matter in bottom sediments. The particulate organic matters are
from a source outside the system such as wastewater particulate or leaf litter materials, and
generated inside system as occurs with plant growth in highly productive environments.
According to accumulate these particulate organic matter in the bottom sediment, the
sediment oxygen demand will increase due to oxidation of the accumulated organic matter
at aerobic sediment layer. Figure 4.4 shows the mechanism of SOD flux with particulate
organic matters in the sediment column. Particulate organic matter (POM) is delivered to
the sediments by settling. Within the anaerobic sediment layer, the organic carbon, sulfate,
nitrate undergo reduction reactions to yield dissolved methane, sulfide, dissolved ammonium
nitrogen. These reduced species (CH4, H2S, NH4) diffuses upward to the aerobic layer
where these are oxidized. During these oxidation reactions, SOD is generated. Any residual
reduced species that are not oxidized in the aerobic layer is diffused back into the water
column where additional oxygen is consumed by oxidation. Developing an SOD model is
quite complicated task since many aerobic and anaerobic reactions in the sediments are
involved and many chemical species are included in mass balance equation. Therefore, these
redox chemistry with POM are usually treated as a composite characteristics of the particular
system. Recently, techniques have been developed for investigating these factors.
52
DiToro et al. (1990) developed a model of the SOD process in a mechanistic fashion
using the square-root relationship of SOD to sediment oxygen carbon content. Using similar
analysis as applied to carbon, they also evaluate the effect of nitrification on SOD. In this
model, carbon and nitrogen diagenesis are assumed to occur at uniform rates in a
homogeneous layer of the sediment of constant depth (active layer). The sediment oxygen
demand and sediment fluxes are calculated by the concentrations of particulate organic
carbonaceous material and of particulate organic nitrogenous material in this active layer.
The more detail description of sediment flux model developed by DiToro and Fitzpatrick
(1993) is presented in Appendix F. Their framework has been applied to the Chesapeake
Bay (Cerco and Cole, 1994; DiToro and Fitzpatrick, 1993) with sediment flux model which
include ammonia and nitrate flux and the sulfide, oxygen, phosphorous and silica flux.
A calculation of the detailed redox chemistry of the sediment interstitial water is
required for a detail understanding of the situation for each reduced species and chemical
parameters for redox reaction such as pH, Eh in the system. In the absence of data for CH4,
H2S, Iron, Methane, and C02, and detail understanding of redox chemistry of the system,
applying this model could create more uncertainty than the simple empirical formula based
on the measurement technique.
The process of oxygen demand in the sediment/water interface is usually referred to
as sediment oxygen demand (SOD) because of the typical mode of measurement: enclosing
the sediments in the chamber and measuring the change in the dissolved oxygen
concentration at several time increments. The major factors affecting SOD are: temperature,
oxygen concentration at the sediment water interface, organic and physical characteristics of
the sediment, and current velocity over the sediments (Bowie, 1985).
Water Column
SL
53
02
Aerobic Layer
Oxidation
CH4, NH4, H2S
Anaerobic Layer
Diffusion
Reduction
^ CH4, NH4, H2S
Production of
CH4, NH4, H2S
Figure 4.4 The relationship between POM flux and SOD flux related in the oxidation and
reduction of organic matter in sediment column.
54
Typical values of SOD are listed in Table 4.1 (Chapra, 1997). In general, values from
about 1 to 10 g02/m2-day are considered indicative of enriched sediments. According to this
table, SOD at a mud bottom is higher than SOD at a sandy bottom. In Charlotte Harbor
estuarine system, the sediment type near upper Charlotte Harbor is finer and includes more
clay and mud than those for the other study area. The finer sediment which has clay, silt, and
mud would have higher SOD as more organically rich sediments. From this assumption,
SOD can be applied as a function of sediment bottom type (sediment size).
Table 4.1 Average Values of Oxygen Uptake Rates of Bottom (Chapra, 1997)
SOD rate at 20°C (g02/m2-day )
Bottom type and Location
Average Value
Range
7
4
2-10
1.5
1-2
1.5
1-2
0.5
0.2-1
0.07
0.05-1
Sphaerolitus (10 g-dry wt — 2)
Municipal sewage sludge:
Outfall vicinity
Downstream of outfall
Estuarine mud
sandy bottom
Mineral soil
The effect of temperature and sediment type on SOD can be represented by
Sb(T) = Sb,20'ST '&
7-20
(4.19)
where SB20 is an areal SOD rate at 20 °C (g02/nr-day ) which is user defined value; ST is
a fractional coefficient for sediment type. (When sediment type <2, ST=1, when sediment
type >3, ST=0.5) ; 0 is temperature coefficient. Zison et al. (1978) have reported a range of
1.04 to 1.13 for 0. A value 1.065 is commonly employed and is used in this study.
Oxygen is another factor that affects sediment oxygen demand. Sediment oxygen
consumption is reduced as oxygen concentration in the overlying water decrease. Lam et al.
(1984) use a Michealis-Menten relationship to represent the dependence, by a saturation
relationship,
55
SB(DO)= D°SB(T) (4.20)
KSOD+UU
where KSOD is half saturation rate for SOD. Lam et al. (1984) have suggested a value for KS0D
of 1.4 mg/1.
The decay of substrate is assumed to balance continued settling resulting in a steady-
state sediment concentration of oxygen demand substrate. According to this assumption, the
kinetic equation for sediment oxygen demand is (Ambrose et al., 1994):
dDO__SOD (421)
dt H
where H is water depth (m); and SOD is sediment oxygen demand (as measured), g02/m2-
day. SOD can be calculated as a function of temperature, dissolved oxygen at the water-
sediment interface, and sediment bottom type based on characteristics of SOD measurement.
Exchange of material between the water column and benthic sediment is an important
component of the eutrophication process. Sediment oxygen demand may comprise a
substantial fraction of total system oxygen consumption (Cerco and Cole, 1995).
Oxygen consumption in the sediments depends upon water-column temperature and
oxygen availability, and sediment type. As temperature increases, respiration in the sediment
increases. Sediment oxygen consumption is reduced as oxygen concentration in the
overlying water decreases. Therefore, the kinetic equation for sediment oxygen consumption
(SOC) in sediment column can be represented as (Cerco and Cole, 1995):
dDOsed SOC _ 1 s QT^ DO,
dt H H B-2° KS0D+1
ML - -^1 = - J- . 5 . flr-20 . _ f^Mfc (422)
where SB20 is a function of sediment bottom type (from table 4.1).
The processes that create sediment oxygen demand are little affected by the
56
concentration of oxygen in the overlying water. When oxygen is unavailable to fulfill
sediment oxygen demand, the demand is exported to the water column. The exported
demand may be in the form of reduced iron, manganese, methane, or sulfide, which are
represented in the model as sediment oxygen demand and provides a function which
computes additional release as oxygen consumption in the sediment is restrained (Cerco and
Cole, 1995).
d£O_SOD=_l_T.20t_K^_
dt H H Bao KSOD + DO
The relationship between sediment oxygen consumption and sediment oxygen
demand is represented in Figure 4.5. The SOD is negligible when DO much higher than
KSOD. When dissolved oxygen is absent from the water column, the maximum oxygen
demand is released to the water as sediment oxygen demand.
57
SOC (Sediment Oxygen Consumption) at sediment column
SOD (Sediment Oxygen Demand) at water column
where SB (T= 20°C) = 2 g/m2/day
KS0D = 2 g/m3
X
2-0.5
■1
■1.5
-2
Dissolved Oxygen (g/m )
Figure 4.5 Effect of dissolved oxygen on sediment oxygen consumption and SOD release
58
Photosynthesis and Respiration
The photosynthesis and respiration of phytoplankton can add and deplete significant
quantities of oxygen from natural systems. The produced oxygen concentration by
photosynthesis depends on the form of the nitrogen species accessed for phytoplankton
growth. One mole carbon dioxide can produce one mole oxygen when ammonium is the
nitrogen source, while one mole carbon dioxide produces 1.3 moles oxygen when nitrate is
the nitrogen source, according to Morel's equation (1983)
106Ca +\6NH: + H.PO; + 106ff2O -» protoplasm + l06O2 + 15H+
(4.24)
\06CO2+\6NO~ +H2PO; +\22H20 + UH^ -> protoplasm + 13802
The simple representation of the respiration process can be used to determine how
much oxygen would be consumed in the decomposition of a unit mass of organic carbon,
6C02 + 6H20 ^ C6Hl206 + 602 (4.25)
The dissolved oxygen-to-carbon ratio in respiration can be calculate from this
equation.
Aoc = (^- = 2.67gO/gC (4.26)
oc 6(12)
The equation that describes photosynthesis and respiration on dissolved oxygen is:
^ = [(1.3 - 0.3 • PH)jUa ~ K„ - KM ] • Aoc ■ PhyC (4.27)
where Pn is nitrogen preference coefficient; //a is phytoplankton growth rate, which is a
function of the intensity of light, the availability of nutrients, and the ambient temperature;
and Kas and K^ are respiration and mortality, which are functions of temperature.
The mass balance equation for dissolved oxygen is written by combining all oxygen
transformation processes.
59
For water column:
^ = +^{D0 -DO)-KD 52 CBOD
dl teK ' ' HCBm + DO
14 m HNIT+DO Az
+[(l3-03.Pn)jua-K(U-Kax}Aoc-PhyC
which include reaeration, oxidation by CBOD, nitrification, sediment oxygen demand, and
photosynthesis and respiration terms.
Fore sediment column:
dDO DO „ _ 64 „ DO XTr, SOC
= -KD CBOD KNN NH4 (4.29)
& HCBOD+DO 14 NN HNIT + DO Az
which include oxidation by CBOD, nitrification, and sediment oxygen consumption.
4.5 Effects of Temperature and Light Intensity on Water Quality Processes
Temperature and light intensity are the most important parameters for transformation
processes. To achieve a better understanding of water quality processes, it is necessary to
improve the spatial and temporal variation of these parameters.
4.5.1 Temperature
In the nutrient cycle, almost all the reaction parameters are affected by temperature,
such as zooplankton and phytoplankton growth, respiration and mortality, nitrification,
denitrification, NH3 stability, mineralization, oxidation, sediment oxygen demand, and
sorption/desorption reactions. The effect of temperature on reaction rates can be explained
by the Van't Hoff-Arrhenius equation, as follows:
^^1 = ^L (4.30)
dt RT2
where K is reaction rate at temperature T, AH is the amount of heat required to bring the
molecules of the reactant to the energy state required for the reaction, and R is universal gas
60
constant.
Integrating Equation (4.44) from temperature T, to T2 gives:
K2
— - = exp
AH
RT{T2
(ra-rt)
(4.31)
where K, and K2 are reaction rate at temperature T, and T2 , respectively. The temperature
adjustment function ( 6 = exp
A//
RTJ2
) is almost constant at the temperature range of
interest (0° ~ 30°C), ranging from 1.01 to 1.2. This equation can be rearranged into a more
useful form as:
K(T) = K(Tref)-0{T~T"/>
(4.32)
where K(T) is the reaction coefficient at temperature T, such as /ua (phytoplankton growth
rate), K^ (phytoplankton respiration rate), Km (phytoplankton mortality rate),//, (zooplankton
growth rate), Ka (zooplankton respiration rate), K^ (zooplankton mortality rate), KM
(ammonia instability), Kom (ammonification rate), KNN (nitrification rate), dun
(sorption/desorption rate for organic nitrogen), dm (sorption/desorption rate for inorganic
nitrogen), K0PM (mineralization rate), dop (sorption/desorption rate for organic phosphorous),
dap (sorption/desorption rate for inorganic phosphorous), KD (oxidation rate),or SOD
(sediment oxygen demand). Each rate term has a unique temperature adjustment function.
Most models, which use exponential temperature functions, assume a reference
temperature of 20 °C (Chen and Orlob, 1975; Thomann and Frizpatrick, 1982). Eppley
(1972) showed that an exponential relationship describes the envelope curve of growth rate
versus temperature data. The determination was made with a large number of studies, with
61
many different species.
4.5.1 Light intensity
Light intensity affects the photosynthesis process and thus the phytoplankton growth
rate. The effects of light intensity on nutrient cycling is often modeled by a light intensity
limiting function as follow (Steele, 1974)
/(/) = y-exp
1-1
(4.33)
where / is the light intensity, and Z, is the optimum light intensity for phytoplankton growth
. According to the Lambert-Beer equation, the light intensity over the water depth is:
I(z) = I0-exp[-Kd(PAR)-z] (4.34)
where I(z) is the light intensity at depth z, and I0 is the light intensity at the water surface.
KJPAR) is a function of suspended sediment concentration, algae concentration, and color.
This value was calculated by the light attenuation model, which will be discussed in the next
section.
4.6 Light Attenuation Model
One of the most important variables controlling phytoplankton photosynthesis is
"photosynthetically active radiation" (PAR), or light, in the range of wavelengths from 400-
700nm, which provides the predominant source of energy for autotrophic organisms (Day
et al., 1989). Absorption and scattering of light by water and dissolved and suspended
matter determine the quantity and spectral quality of light at a given depth (Jerlov, 1976;
Prieur and Sathyendranath, 1981), which in tern affect the photosynthesis of aquatic plants.
One way to develop a light attenuation model is to find a simple regression
relationship between Kd (PAR), calculated from light measurements, and water quality
62
measurements collected at the same instant as the Kd (PAR) vales. This type of empirical
model is a simple way to relate light attenuation to water quality, at a certain time. This
method was used by Mcpherson and Miller (1994) in Tampa Bay and Charlotte Harbor. A
physics-based light attenuation model was developed as part of the CH3D-EVIS (Sheng et al.
2001c, Christian and Sheng, 2003) and successfully applied to the Indian River Lagoon. This
light model is adopted for the Charlotte Harbor study.
In the light model, the vertical light attenuation coefficient (Kd) is a function of solar
zenith angle (yu0), scattering (b) and total absorption (a,) (Kirk, 1984)
1 r , , N -|l/2
Mo J (4.35)
and G(p0) = grfi0-g2
with g, =0.473 and g2=0.218 determined for the mid point of euphotic zone (Kirk, 1984).
The scattering coefficient, b, is determined only using particles since scattering due
to particles is much greater than scattering due to water (Gallegos, 1994). Scattering can be
described as a function of turbidity (Morel and Gentili, 1991) as follow:
b(A) = (550/ A) ■ [Turbidity] (4.36)
Total absorption (a,) is partitioned into attributes of water (aw), phytoplankton (aph),
dissolved color (ad(), and detritus {ad).
a, = «»• + aPh + adc + ad (4.37)
The absorption of water (a J can be determined from literature values (Smith and
Baker, 1981) with 1 nm linearly interpolated from 5 nm found in the literature. Chlorophyll-
specific absorption (aph) is calculated for the model from the linear relationship between the
maximum absorption and analytical chlorophyll_a concentration (Dixon and Gary, 1999):
63
anh C^) = \aoh ) ' formalized _ Spectra(A)
MAX (4.38)
fa . J =0.0209 -(Chlorophyll a, corrected)
where (aph)MAX is the maximum absorption of chlorophyll-a. To calculate normalized_spectra
(X), individual spectra are normalized to the maximum absorption (437 - 440 nm), and
averaged for all samples for an overall normalized spectra.
The absorption by dissolved color, adc, for each wavelength in the visible spectrum
can be found using a negative exponential function (Bricaud et al., 1981)
adlU) = 8uo -«p[-** (A -440)] (4.39)
where g440 is the absorption by dissolved color at 440 nm and sdc is spectral slope. Dixon
and Gary (1999) calculated empirical absorption at 440 nm and spectral slope as a function
of color (in PCU) at Charlotte Harbor in the form:
g44() = 0.0667 ■ f color] , n = 129, r = 0.9329
(4.40)
sJc = 0.00003 • [color] - 0.0178, n = 129, r = 0.5 1 1 1
Absorption due to organic and mineral detritus is represented as a function of
turbidity (Gallegos, 1994):
ad = Gd (A) • [Turbidity] (4.41)
where turbidity is in NTUs and od(A) is the wavelength specific absorption cross section of
turbidity as calculated in:
<?d U) = <*bi + ^oo • exP [~sd (A ~ 40°)] (4-42)
in which abl is the longwave absorption cross section, a400 is the maximum detritus
absorption at 400 nm, and sd is exponential slope.
The total absorption and scattering are used in Equation (4.49) to calculate the
64
vertical attenuation coefficient, KJA), for each wavelength depending on color,
chlorophylls , and turbidity. This KJA) value and incident irradiation EJA) can be used in
the equation for calculating irradiance EZ(A) at the reference depth, zr:
Ez(A) = E0(A)-exp[-Kd(A)-zr] (4.43)
The spectrum of incident sunlight data from table F-200 in Weast (1977) is used for
incident spectral information, EJA), as in Gallegos's work (1994) . These data are shown
in Table 4.2.
Table 4.2 The spectrum of incident sun
X (nm)
400
405
410
415
420
425
430
435
440
445
450
455
460
465
470
475
480
485
490
495
500
X (nm)
4.780
5.568
505
6.003
510
6.052
515
6.135
520
6.017
525
5.893
530
5.940
535
6.659
540
7.152
545
7.548
550
7.826
555
7.947
560
7.963
565
7.990
570
8.119
575
8.324
580
8.014
585
7.990
590
8.113
595
8.119
600
ight data (Gallegos, 1994)
X (nm)
8.108
505
8.026
510
7.894
515
7.970
520
8.130
525
8.163
530
8.133
535
8.051
540
7.993
545
7.993
550
7.982
555
7.937
560
8.055
565
8.160
570
8.265
575
8.318
580
8.375
585
8.387
590
8.369
595
8.359
600
8.332
8.340
8.323
8.305
8.289
8.271
8.267
8.263
8.239
8.213
8.207
8.201
8.180
8.157
8.136
8.114
8.106
8.089
8.052
8.013
EJA) and E,(A) are integrated over the visible spectrum to get PAR0 for the incident
PAR and PARZ for the calculated PAR at the reference depth. The spectrally sensitive
65
attenuation coefficient Kd{PAR) is calculated from these integrated values by using the
rearranged form of the Lambert-Beer equation:
~ 1 <
Kd(PAR) = —\n
Z.
PARZ
V PARo j
(4.44)
This Kj(PAR) will be used in the model to calculate light levels throughout the
water column as a function of measured incident light intensity. Dixon and Gray (1999)
calibrated the light attenuation model with measured data and applied the model to determine
the light requirements for seagrasses of the Charlotte Harbor estuarine system. The results
of the optical model are very good, with mean percentage agreement between modeled and
observed Ktl of 1 10%. Using water quality data from all stations as inputs, Dixon and Gray
(1999) determined that chlorophyll, color, and turbidity account for 4%, 66%, and 31% of
water column light attenuation, respectively. The maximum annual average chlorophyll
contribution is 6%, with an individual maximum of 18% during a phytoplankton bloom.
Color dominated water column attenuation, ranging from 40% to 78%. Stations in the lower
Harbor and southern sites showed increased attenuation due to turbidity, compared to the
upper Harbor site (Dixon and Gary, 1999). For the southern sites, a much larger portion of
light attenuation is produced by turbidity, up to 55% for the station near Captiva pass.
To provide a light attenuation component which can be coupled with water quality,
hydrodynamic, and sediment model components in the Charlotte Harbor estuarine system.
A stand alone light model needs to be developed and calibrated with measured data. The
stand alone light model has been calibrated by finding best fit between simulated and
measured light attenuation with various coefficient sets.
For calibration of the light model, aM, a400, sd, and s are allowed to vary within
66
certain ranges. The predicted kJPAR) values for the stand alone light model are compared
to the corresponding kJPAR) values from data provided by the SFWMD. If the predicted
values do not match data well, then attempts are made to try to find model coefficients to
produce better fit. In this case, the model is run and the coefficients are varied between
maximum and minimum literature values specified in Table4.3 (Christian and Sheng, 2003).
More than one thousand runs of light model simulations were conducted with this data set.
The RMS errors were calculated for each run to test model performance of each set of light
model coefficients. The root mean square error is an indication of the average discrepancy
between observations and model results. In addition to root mean square error, model
calibration was assessed via plots of model output and observation. Scatter plots of model
output and observed data provide an indication of overall model performance. The best
RMS errors for each of the tests is 0.61 1 1 — ' with best fit coefficients shown in Table 4.4
Dixon and Gray (1999) found the coefficients in Table 4.5 provided the best fit to all the data
they examined. They applied a400 and sy as a function of color. The RMS error with this
coefficient is 0.65 11m"1. Even though, they reduced adjustable coefficients with relationship
of a400 andsy with color, the RMS error is greater than that for adjusting all four coefficients.
In this study, the best fit coefficients were used for simulating light attenuation. Figure 4.6
contains calibration period scattering plots with best fit of coefficient sets. The location of
circles indicate the correlation between model predictions and observed data. A perfect
match between model and observed data is indicated by the diagonal line on each graph.
Circle above the line shows over prediction, while circles below the line indicate that under
prediction.
67
Table 4.3 Coefficient ranges for use in stand along light model.
Coefficients
Minimum value
o,.
o_
41)11
0.004
0.255
0.011
0.011
Maximum value
0.090
0.600
0.019
0.019
Table 4.4 Best fit light model coefficients for Charlotte Harbor estuarine system.
Coefficient
<Ju
aA
401)
Value
0.01 m-'NTTJ-1
0.59 m'NTU''
0.014 nm1
0.013 nm'1
Table 4.5 Dixon and Gray's model coefficients for the Charlotte Harbor estuarine system.
Coefficient
Value
°u
0.064 m'NTU-1
°400
0.0014xcolor+0.0731 m'NTU'1
*i
0.01125 nm1
Sv
0.00003xcolor+0.0178 nm '
68
T3
0
■o
cr
<
Q.
5 -
4 -
3 -
2 -
1 -
0
• PARPS model
•
*
— •
0
_ 0
0
0
_ 0
0
0
- 0
^m 0
0
0
— 0
0
0
0
0
0
0
_ 0
•
0
0
•
— ■ 0
0
•
• /
0
_ #
0
0
0
• • •+' •
•*''• ••• •
• %T\0 • #
•
_ *
0
'**' I I I I J 1 1 I I I I I I 1 I I I I 1 I I I I 1 I I I I
2 3 4
KpAR observed
Figure 4.6 The scatter plots for Kd(PAR) during calibration period with best fit coefficients.
69
The water quality and sediment model outputs needed for the light model are
turbidity in NTUs, chlorophyll_a concentration in Hg/L, and color in Pt. Units. To calculate
solar angle, this model needs the latitude and simulation day and time. Because the sediment
model simulates the total suspended solids (TSS), it is necessary to develop a regression
between TSS (mg/L) and turbidity (NTU) as follow:
Turbidity = 0.2677 x TSS + 0.9665 for 1996 data
Turbidity = -0.0 1 53 x TSS + 2.602 for 2000 data
The stand alone light model was tested, calibrated, and then coupled with the models
of hydrodynamics, sediment and water quality.
4.7 Model Parameters and Calibration Procedures
Reaction terms described in the previous section contain many model parameters
which must be determined before using the model. The determination of model parameters
is generally very difficult because they depend on many physical and biochemical factors,
such as the location of the estuary, temperature and tidal variation, and point or non-point
source loadings of nutrients and other chemical materials. In practice, parameters are
selected from a range of feasible values, tested in the model, and adjusted until an optimal
agreement between simulated and measured values is obtained. Two ways to determine
feasible ranges of model parameters are field observation and laboratory experimentation.
When field observations and laboratory experimentation are not available, the feasible ranges
are obtained from literature or previous modeling studies. A list of the important kinetic
parameters and their literature values are given in Table 4.6 ~ 4.11.
70
Table 4.6 Temperature adjustment functions for water quality parameters
Paramete
r
description
Unit
literature
value
Reference
(»*)
temperature function for
phytoplankton growth
-
1.01 ~ 1.2
1.09
1.08
Di Toro et al. (1980)
Pribble et al.(1997)
Sheng etal. (2001)
K»)
temperature function for
phytoplankton
respiration and mortality
-
1.045
1.05
1.08
Ambrose (1991)
Pribble etal.(1997)
Sheng et al. (2001)
(»«)
temperature function for
zooplankton growth,
respiration and mortality
-
1.01 - 1.2
1.04
Di Toro etal. (1980)
Sheng et al. (2001)
\"oNM )
temperature function for
ammonification
-
1.0- 1.04
1.07
1.02
Bowie et al. (1985)
Pribble etal.(1997)
Sheng et al. (2001)
ifim)
temperature function for
nitrification
-
1.02-1.08
1.08
1.08
Bowie et al. (1985)
Pribble et al.(1997)
Sheng et al. (2001)
(<V)
temperature function for
denitrification
-
1.02- 1.09
1.04
1.045
Bowie etal. (1985)
Pribble et al.(1997)
Sheng et al. (2001)
M
temperature function for
ammonia instability
-
1.08
Sheng etal. (2001)
(0OPM )
temperature function for
mineralization
-
1.08
Sheng etal. (2001)
{&s/d)
temperature function for
sorpti on/desorpti on
-
1.08
Sheng etal. (2001)
M
temperature function for
oxidation
-
1.02-1.15
1.08
Bowie et al. (1985)
Sheng etal. (2001)
(8 SOD )
temperature function for
sediment oxygen
demand
-
1.045
1.08
Bowie et al. (1985)
Sheng et al. (2001)
71
Table 4.7 Water quality parameters related to conversion rate
Paramete
description
Unit
literature
Reference
r
value
Ajc
Phytoplankton Carbon
go2/gc
2.67
Ambrose(1991)
/ Oxygen rate
2.67
2.67
Cerco and Thomas
(1995)
Shengetal. (2001)
ChlaC
Phytoplankton Carbon
gC/gChl
10- 112
Bowie et al. (1985)
/ Chlorophyll_a rate
a
100
60
50
Pribble et al.(1997)
Cerco and Thomas
(1995)
Shengetal. (2001)
A:/V
Phytoplankton Carbon
gN/gC
0.05 -0.43
Jorgensen (1976)
/ Nitrogen rate
0.15
0.167
0.15
Pribble et al.(1997)
Cerco and Thomas
(1995)
Sheng et al. (2001)
\p
Phytoplankton Carbon
gP/gC
0.005-0.03
Jorgensen (1976)
/ Phosphorous rate
0.027
0.025
Cerco and Thomas
(1995)
Shengetal. (2001)
Table 4.8 Water quality parameters related to phytoplankton and zooplankton
Paramete
r
description
Unit
literature
value
Reference
ta>L
maximum phytoplankton
growth rate
1/day
0.2-8
2.25-2.5
1.06-2.68
Bowie et al. (1985)
Cerco and Thomas
(1995)
Shengetal. (2001)
Hn
Nitrogen half saturation
rate for phytoplankton
uptake
ng/i
1.5-400
0.5
1
10
Bowie etal. (1985)
Pribble et al.(1997)
Cerco and Thomas
(1995)
Sheng et al. (2001)
HP
Phosphorous half
saturation rate for
phytoplankton uptake
Mfl
1. - 105
1
1
2-4
Bowie etal. (1985)
Pribble et al.( 1997)
Cerco and Thomas
(1995)
Shengetal. (2001)
72
opt
optimum light intensity
for phytoplankton
growth
ly/day
225-600
300
Canale et al. (1976)
Sheng et al. (2001)
KM
phytoplankton
respiration rate
1/day
0.02- 0.24
0.03- 0.09
0.03- 0.05
Jorgenson (1976)
Cerco and Thomas
(1995)
Sheng etal. (2001)
Kus
phytoplankton non-
predator mortality
1/day
0.01-0.22
0.03- 0.09
0.02- 0.06
Jorgenson (1976)
Cerco and Thomas
(1995)
Sheng etal. (2001)
wsP*y
phytoplankton settling
rate
m/da
y
0. - 3.
0. - 0.25
0.05-0.1
Bowie etal. (1985)
Cerco and Thomas
(1995)
Sheng et al. (2001)
HPky
phytoplankton half
saturation rate for
zooplankton uptake
ng/i
200-2000
800-1200
Bowie et al. (1985)
Sheng et al. (2001)
TrSP»y
phytoplankton threshold
for zooplankton uptake
ran
1-200
200
Bowie et al. (1985)
Sheng et al. (2001)
V * /max
maximum zooplankton
growth rate
1/day
0.15-0.5
0.18-0.2
Bowie et al. (1985)
Sheng et al. (2001)
K„
zooplankton respiration
rate
1/day
0.003-0.07
5
0.01
Bowie etal. (1985)
Sheng et al. (2001)
Ka
zooplankton non-
predator mortality
1/day
0.001-0.36
0.015-0.05
5
Jorgensen (1976)
Sheng et al. (2001)
Table 4.9 Water quality parameters in the nitrogen dynamics
Paramete
r
description
Unit
literature
value
Reference
K-onm
Ammonification rate
1/day
0.001-0.4
0.1
0.015
0.01
Bowie et al. (1985)
Pribble et al. (1997)
Cerco and Thomas
(1995)
Sheng etal. (2001)
73
KNN
nitrification rate
1/day
0.004-0.11
0.08
0.07
0.01-0.02
Bowie et al. (1985)
Pribble et al.(1997)
Cerco and Thomas
(1995)
Shengetal. (2001)
Hnil
DO saturation rate for
nitrification
mg/1
0.1-2.0
2
2
Ambrose (1994)
Pribble et al.(1997)
Sheng et al. (2001)
KDN
denitrification rate
1/day
0.02- 1.0
0.09
0.09
Bowie et al. (1985)
Pribble et al.(1997)
Sheng et al. (2001)
Hnoi
DO saturation rate for
denitrification
mg/1
0.- 1.0
0.1
Bowie et al. (1985)
Shengetal. (2001)
Pm
partition coefficient of
PON/SON
-
l.E-5
1.E-6-9E-6
Simon (1989)
Sheng et al. (2001)
Pan
partition coefficient of
PIN/NH4
5.E-6-1.E-
5
3.E-5-4.E-
3
Simon (1989)
Sheng et al. (2001)
don
sorption/desorption rate
for SON/PON
1/day
0.02
0.08
0.01-0.02
Bowie etal. (1985)
Cerco and Thomas
(1995)
Shengetal. (2001)
dan
sorption/desorption rate
for PIN/NH4
1/day
0.01
Shengetal. (2001)
^■PDN
preference partition
coefficient of mortality
for SON / PON
-
0.5
0.5
Cerco and Thomas
(1995)
Sheng etal. (2001)
Table 4.10 Water quality parameters in
the phos]
Dhorous dynamics
Paramete
r
description
Unit
literature
value
Reference
"~OPM
mineralization rate
1/day
0.001 - 0.6
2.27
0.1
0.1
Bowie et al. (1985)
Pribble et al. (1997)
Cerco and Thomas
(1995)
Sheng et al. (2001)
Pop
partition coefficient of
POP/SOP
-
8.E-6-1.E-
4
Shengetal. (2001)
74
%
partition coefficient of
PIP/SRP
-
1.E-6-6E-
4
Sheng et al. (2001)
<p
sorption/desorption rate
for SOP/POP
1/day
0.08
0.01
Cerco and Thomas
(1995)
Sheng etal. (2001)
dap
sorption/desorption rate
for PIP/SRP
1/day
0.01
Sheng etal. (2001)
PpDP
preference partition
coefficient of mortality
for SOP / POP
-
0.5
0.5
Cerco and Thomas
(1995)
Sheng etal. (2001)
Table 4.1 1 Water quality parameters in
the oxygen balance
Paramete
r
description
Unit
literature
value
Reference
KD
oxidation rate
1/day
0.02 - 0.6
0.05
Bowie et al. (1985)
Sheng et al. (2001)
" CBOD
DO half saturation rate
for oxidation
mg/1
1.5-400
0.5
1
10
Bowie et al. (1985)
Pribbleetal.(1997)
Cerco and Thomas
(1995)
Sheng et al. (2001)
J®CBOD
partition coefficient of
particular/dissolved
CBOD
-
0.5
0.3-0.5
Bowie et al. (1985)
Sheng et al. (2001)
SOD
Sediment oxygen
demand
g02/m
2-day
0.02-10.
0.0 - 10.7
Thomann (1972)
Bowie et al. (1985)
" SOD
DO half saturation rate
for sediment oxygen
demand
mg/1
0.01-0.22
0.03- 0.09
0.02- 0.06
Jorgenson (1979)
Cerco and Thomas
(1995)
Sheng et al. (2001)
KAE
reaeration rate
1/day
0. - 3.
0. - 0.25
0.05-0.1
Bowie etal. (1985)
Cerco and Thomas
(1995)
Sheng etal. (2001)
75
Model calibration is the first stage testing or tuning of the model to a field data not
used in the original construction of the model. Such tuning is to include consistent and
rational set of theoretically defensible parameters and inputs (Thomann, 1992). Proper
calibration of the water quality model requires having accurate representation of the inflow
and loads of nutrients into the water body and selecting appropriate model parameters.
The reaction equation shown in previous section is a function of its concentration and
the water quality parameters connected to it by the indicated processes. Within each reaction
equation, there are numerous kinetic parameters and additional parameters. Water quality
model in CH3D-EVIS consists of 13 state variable equations with over 40 interrelated
parameters. The interactions of water quality model parameters and the constituent equations
as shown in Table 4.12 clearly indicate the complexity of the calibrating these types of
models. The column of Table 4.12 show that each modeled constituent equation contains
between 2 and 15 different water quality parameters. According to each row, each parameter
can be found in up to 10 different constituent equations. Therefore, changing one parameter
to improve the calibration of one constituent will simultaneously affect many other
constituents. Traditionally, calibration of water quality models has been performed manually
using a trial-and-error parameter adjustment procedure. The process of manual calibration
depending on the number of model parameters and the degree of parameter interaction. With
complexity of water quality model, the traditional process is a very tedious and time
consuming task. It is necessary to apply more systematic and efficient calibration procedure
for reducing calibration time and effort.
Based on the cascading effect of adjusting interrelated parameters, the efficient
calibration of the water quality model should begin with the parameters that affect the more
76
constituents and the more sensitive parameters. First of all, each water quality parameter
can be ranked with these sensitivity and relativity as parameterization. According to this
order, high ranked parameters will be adjusted to reproduce major pattern of all constituents,
and then lower ranked parameters will be calibrated for detail characteristics of local
constituents. This procedure will reduce numerous calibration efforts and increase efficiency
and effectiveness.
The calibration procedure involves optimization of numerical measures (objective
functions) that compare observations of the state of the system with corresponding simulated
values. The most commonly used objective function adopted in calibration is the root mean
square errors between the observed and simulated model response. The root mean square
error is an indication of the average discrepancy between observations and model results.
It is computed as follow:
T(o-p)2
RMS = J^- (4.45)
n
where RMS is root mean square error
O is observation
P is model prediction
n is number of observation
In addition with root mean square error, model calibration was assessed via plots of
model output and observation with correlation coefficient R2. Scatter plots of model output
and observed data provide an indication of overall model performance. The correlation
coefficient is defined by
77
ssl C£0P-nOP):
R* =
(4.46)
SS^-SS^ (£02-n.02).(£P2-n-P2)
SSxx=Z(Ol-d)2=YJ02-n.02
SS^ZiP.-Pf-TP'-n-P2
SSv^iOt-OM-P^J^O-P-nO-P
The process of model calibration is illustrated in Figure 4.7. In systematic calibration
procedure, parameters are adjusted according to order of sensitivity and relativity according
to parameterization for optimization of certain criteria (objective functions) that measure the
goodness-of-fit of the simulation model. The process is repeated until a specified stopping
criterion is satisfied.
Formation of a proper framework for systematic calibration involves the following
key elements:
• Sensitivity analysis
• Model parameterization and choice of calibration parameters
• Specification of calibration criteria
The best way to calibrate water quality parameters is the automatic calibration, in
which parameters are adjusted automatically according to a specific search scheme for
optimization of certain calibration criteria. The process is repeated until a specified stopping
criterion is satisfied. In this study, automatic calibration procedures have been developed
and tested, using a Gauss-Newton method.
To test the automatic calibration procedure of the CH3D water quality model, a 1996
baseline run was adopted as representative of the true and valid field condition. All
parameters from baseline run were considered representative of conditions that could
hypothetically exist in the field. A small number of these parameters were perturbed slightly,
and the automatic calibration procedure was employed. If the procedure is correct, these
78
perturbed parameters should converge on pre-perturbed values. The results of the test show
that the changed parameters did converge on pre-perturbed baseline values, with the Gauss-
Newton method. The method is therefore considered valid and accurate.
In real case, a trial run was conducted by using all the measured data to find best fit
parameters. The calibration procedure continuously failed as calibration parameters were
automatically adjusted outside the upper and lower bounds. The accuracy of the model
calibration generated by this procedure, relies heavily on the quality and quantity of field
observation, the model structure, and the nature of the system. Limited data, and
uncertainties in water quality processes caused the procedure to fail to generated a valid
calibration condition. To apply an automatic calibration procedure to this, or other water
quality models, further investigation is required. These investigations should be focus on the
optimization algorithm and refinement of the water quality processes.
79
Table 4.12 The relationship
> between water quality parameters and model constituents
CA CZ PIN NH4 N03 SON PON PIP SRP SOP POP DO CBOD
#of
Eqns
AGRM
HALFN
HALFP
KAEX
KAS
WAS
HALFA
TRESHA
ZGRM
KZEX
KZS
SONM
NITR
DENR
PCON
DRON
KPDN
PCAN
DRAN
SOPM
PCOP
DROP
KPDP
PCIP
DRIP
SODM
FD5
AKD
AOC
CHLAC
CAN
ACP
TOPT
OPTL
AKNIT
AKDEN
AKBOD
AKSOD
AKAIR
1 i ;
! l
1 I
l i 1 ;
5
1
1
5
7
1
2
2
2
3
6
2
3
2
2
2
2
2
2
2
2
2
2
2
2
1
2
2
10
2
6
5
1
1
3
2
2
1
1
i i
1"! i
i i l
l !
l
l l
l ;
i ■ ' :
; l ! l
l
:
: :
: i : :
i I i i
i i ;
1 ; i !
!
: i i :
i l
i
i i i
...........J
l i
1 :
1
L.1...L. ...L
l
l
i i
i
l
l
i
1 i 1
!
1 1
1 i
1 i
1
i 1
1
l i
1 1
1
l
:
1
l
1
l
:
l
1
i
l
1
:
"T'"?
1
1
l
i
l
l
1
j- f j- 1
l 1 !
l i 1 1
1 1 !
i 11 ii
11 = 1
i
111
1 j
l j
•1 i ;
! l ;
t "T" "I
t t m
i i
1 ]
! ! I ! ! i i i ! i'T"
#of
Parameters
15 82 10 7872887 14 7
103
80
INITIAL VALUE for WQ parameter
I
Execute WQ model
I
Adjust WQ river boundary Condition
I
Execute WQ model
change i parameter for sensitivity test
I
Execute WQ model
Sensitivity analysis :
check RMS differneces between base run and each sensitivity test
I
parameterization :
make order with sensitivity result and relativity table
1
Adjust i ranked parameter
Execute WQ model
Calculate statistics : scatter plot, RMS, R2
Plot time series for each species with measured datat
Figure 4.7 Systematic calibration procedure
CHAPTER 5
APPLICATION OF CIRCULATION AND TRANSPORT MODEL
Circulation in Charlotte Harbor is driven by tide, wind, and density gradient. The
numerical model of circulation and transport described in Chapter 3 was applied to the
Charlotte Harbor estuarine system, a large shallow estuary in southwest Florida. Model
applications include 1) the simulation of three-dimensional circulation during summer 1986
2)one year simulation of flow, salinity, temperature, and sediment transport in 2000 and 3)
simulation of the impact of removal of Sanibel Causeway on the circulation in San Carlos
Bay and Pine Island Sound
A major purpose of the first two simulations is the calibration and validation of the
Charlotte Harbor circulation and transport model using field data obtained by USGS, NOAA,
SWFWMD, and SFWMD in Summer 1986 and January to December in 2000. During the
calibration process, a few model coefficients and/or boundary/initial conditions are adjusted
to produce more accurate overall simulation of observed data collected in July 1986 by
USGS (Hammet, 1992; Stoker, 1992; Goodwin, 1996) covered the entire Charlotte Harbor
estuarine system and include water level, horizontal currents and salinity. Hence July 1986
data were used for short-term calibration. To supplement the short-term calibration of July
1986, the long-term model calibration was conducted using one year water level and salinity
data in Caloosahatchee River during 2000.
81
82
5.1 A High-Resolution Curvilinear Grid for Charlotte Harbor Estuarine System
To procedure a successful numerical simulation of the Charlotte Harbor estuarine
system, it is necessary to design a numerical grid which represents the dominant geographic/
bathymetric features with sufficient spatial resolution. Since CH3D use a boundary-fitted
curvilinear grid in the horizontal directions, it is possible to align the grid lines to coincide
with shorelines, causeways, and bridges. The Charlotte Harbor estuarine system
encompasses about 735 km2 on Florida's southwest coast. The model domain includes all
the sub-basins (Upper Charlotte Harbor, Lower Charlotte Harbor, Pine Island Sound,
Matlatcha Pass, and San Carlos Bay), the major tributaries (Peace, Myakka, and
Caloosahatchee up to the Franklin Dam), Estero Bay, and some offshore water. The model
should be run with a spatial grid sufficiently fine to resolve the Sanibel Causeway. The
eastern boundary includes many of the major rivers with specified flow rates, while the
western boundaries and the western portion of north and south boundaries are open tidal
boundaries where tidal elevations are prescribed. The interior of the model domain is
represented with a boundary-fitted grid which is non-orthogonal but as orthogonal as
possible, i.e. the grid analysis are usually between 60° and 120°.
The final boundary fitted grid used for the three-dimensional curvilinear-grid model
(CH3D) of Charlotte Harbor estuarine system was generated using a grid generation program
originally developed by Thompson et al. (1985) and further enhanced for this study. This
grid (Figure 5.1) contains 92 x 129 horizontal cells and eight vertical layers, with a total of
1 1 648 horizontal grid cells which include 5367 water cells and 628 1 land cells. Grid spacing
varies from 40 to 2876 meters (average 598 meters).
The bathymetry of the Charlotte Harbor estuarine system and the nearshore region
83
of the Gulf of Mexico is derived from data obtained from the Geophysical Data System of
National Geophysical Data Center. Bathymetry of navigation channels in San Carlos Bay
and the vicinity of Sanibel Causeway are based on the recent survey data provided by Lee
County in December 1 999. Bathymetry for Peace River and the upper Charlotte Harbor area
was collected by SWFWMD and that for Caloosahatchee River was collected by SFWMD.
All bathymetric data were converted to NAVD88 to unify vertical datum level (Appendix
H). Charlotte Harbor grid bathymetry (Figure 5.2) was developed using all these bathymetric
data. While an inverse distance interpolation followed by simple smoothing scheme was the
primary method for determining bathymetry in the study area, several areas were added and
the bathymetry further adjusted after the interpolation and smoothing was performed to
ensure the proper passage of flow through a navigation channels.
84
C
o
z
\
J?
?
A V
o
h
^^fc^ffiE^
o
Sv»T-
o
AVC*^
in
\yvow**'
Y ,---
r-
co
v"
CM
j i^^H
O
o
o
o
^m \s&&
LO
CD
SgS&SRi s^^W
^■rstLrf/fiv
C\J
w
§11 \1
o
illlllll
WELug; \ L
o
?Crt>, Vv^H|
o
■~''s$5r>SJtit. 4$K2>
LO
—
C\J
en
SSsc$v---"
CM
v' '' '~^*^$x?xX&
il
o$x38BKv\S^^^^S^Stfcjj*t-->
v^^sSSSKS^^^^ww^
o
o
o
s/?c^>otjo?v^0cfilS^^B^
o
o
CO
liPH
CM
lilt
i i
1
1 1 _J_S^
r 1 i i i i 1 i i i i
350000
375000 400000
Easting (m)
425000
Figure 5.1 Boundary-fitted grid (92 x 129) used for numerical simulation for Charlotte
Harbor estuarine system.
85
£ H>£J
r
J I I L
J i i i i
350000
375000 400000
Easting (m)
425000
2000
1900
1800
1700
1600
1500
1400
1300
1200
1100
1000
900
800
700
600
500
400
300
200
100
0
Figure 5.2 Bathymetry in boundary-fitted grid for Charlotte Harbor estuarine system (92
x 129).
86
5.2 Forcing Mechanism and Boundary Conditions
The forcing mechanisms for the hydrodynamic model include tide, wind, and density
gradients. To incorporate these mechanisms in the model simulations, tides and oceanic
salinity are specified along the offshore boundary, wind is specified at the air-sea interface,
and river discharges are specified at the river boundaries.
The boundary between the Charlotte Harbor estuarine system and the Gulf of Mexico
extends about 64 km from Gasparilla Pass on the north to San Carlos Pass on the south. One
of the most important features is the similarity between the tidal curves at Venice and Cayo
Costa which are located at northern end and near Boca Grande Pass (Figure 1.1). The
phasing of the Gulf tides is nearly synchronous along the boundary of the estuarine system
between Venice and Naples, which are located at the northern and southern ends.
For the 2000 simulation, tidal forcing along the Gulf of Mexico boundary was
prescribed with water level data measured at the Naples station by the NOAA CO-OPS. For
the 1986 simulation, water level data at Venice measured by USGS were used to provide the
tidal boundary condition, since there was no data from the Naples station in this period. To
unify the datum level with the bathymetry, these water level data were converted to NAVD88
datum level.
Discharge and runoff boundary conditions were imposed using the daily measured
discharge data described in Table 4.1. Because no flow data were available for the Estero
Bay and Caloosahatchee River except at S-79 in 1986, the discharges were assumed to be
one half and a quarter of the Myakka River discharges, respectively.
The surface wind boundary condition is supplied using hourly wind magnitude and
direction data collected at three stations, which are the National Data Buoy Center C-MAN
87
stations at Venice (27.07 N, 82.45 W ), NOAA CO-OPS stations at Fort Myers (26° 36' N,
81° 52' W), and Naples (26.13 N, 81.807 W ). For the 2000 simulation, hourly wind data
from all three stations are available, while only the Venice and Fort Myers stations are
available for 1986 simulation. The hourly wind magnitude and direction are converted into
x-(EastAVest) and y- (North/South) velocity components using Garrat's formula (Garrat,
1977). These values are then interpolated to the entire computational grid, with a weighting
function inversely proportional to the square of distances. Figures 5.3 shows the water level
at Venice station for tidal boundary condition and the river flows at river boundaries in 1986.
Figure 5.4 shows wind speed and direction at Venice and Fort Myers station in 1986. Figure
5.5 shows the water level at Naples station for tidal boundaries and the river flows at river
boundaries in 2000. Figure 5.6 shows wind speed and direction at Venice, Fort Myers and
Naples in 2000.
Table 5.1 Descriptions of 1986 and 2000 river boundary conditions for Charlotte Harbor
wluu""v "J"
Station
Name
Agency
Latitude
Longitude
drainage
perio
degree W
degree N
area:mi2
d
Myakka River
02298830
near Sarasota
USGS
27°14'25"
82°18'50"
229.0
daily
02299120
at Deer Prairie Slough
USGS
27°08'06"
82°15'24"
-
daily
02299410
at Big Slough Canal
USGS
27°11'35"
82°08'40"
36.5
daily
Peace River
02296750
at Arcade
USGS
27°13'19"
81 °52'34"
1 ,376.0
daily
02297100
at Joshua Creek
USGS
27°09'59"
81°52'47"
132.0
daily
02297310
at Horse Creek
USGS
27°11'57"
81°59'19"
218.0
daily
02298202
Shell Creek
USGS
26°59'04"
81 °56'09"
373.0
daily
in Caloosahatchee River
02292900
S79 spillway
USGS
26°43'25"
81°41'55"
-
daily
02293214
Cape Coral
USGS
26°38'10"
81 °5548"
-
daily
02293230
Whiskey Creek
USGS
26°34'29"
81 °53'29"
"
daily
in Estero Bay
02291500
Imperial
USGS
26°20'07"
81°44'59"
65.0
daily
02291524
Spring Creek
USGS
26°21'42"
81 °47'27"
-
daily
02291580
Estero River
USGS
26°26'30"
81°47'45"
-
daily
02291673
Mullock & Henry Creek
USGS
26°30'19"
81°51'00"
~
daily
88
Baroclinic simulations require realistic initial salinity and temperature fields in order
to minimize the effects of the initial conditions on the solution, and to minimize the "spin-
up" time. Salinity along the offshore boundary is assumed to be approximately constant at
36.5 ppt. Salinity at the river boundaries was set to 0 ppt. The near bottom salinity data
measured at the nine USGS stations in 1986 were used to determine the initial salinity at the
corresponding grid cells. These salinity data were interpolated over the entire computational
grid, with a weighting function inversely proportional to the square of the distance to the
three nearest stations. Initial temperature field for 2000 simulation was interpolated using
water temperature data at Fort Myers and Naples stations measured by NOAA CO-OPS and
Venice stations measured by National Data Buoy Center (NDBC). Air temperature data at
the same locations were used to calculate heat flux for air-sea interface (Figure 5.7). These
air temperature data were also interpolated over the study area.
89
160
1 80 200
Julian Day
220
300 r
Caloosahatchee River
Peace River
Shell Creek
Myakka River
160
180 200
Julian Day
220
Figure 5.3 Tidal forcing and river discharges for 1986 simulation of Charlotte Harbor
circulation.
at Venice
90
I ;
' I I
I i
150 160 170 180 190 200
Julian Day
210
220
230
at Fort Myers
j_
150 160 170 180 190 200
Julian Day
210
220
230
Figure 5.4 Wind velocity for 1986 simulation of Charlotte Harbor circulation.
91
Julian Day
200 r
100
Caloosahatchee River
Peace River
Shell Creek
Myakka River
200
Julian Day
300
Figure 5.5 Tidal forcing and river discharges for 2000 simulation of Charlotte Harbor
circulation
Wind speed and direction at Venice
92
36850
36875
36900
36925
36950
Wind speed and direction at Fort Myers
36850
36875
36900
36925
36950
Wind speed and direction at Naples
36660
36720
36680 36700
Julian Day since 1 900
Figure 5.6 Wind velocity for 2000 simulation of Charlotte Harbor circulation.
93
at Venice
100
200
Julian Day
300
at Fort Myers
200
Julian Day
100 200 300
Julian Day
Figure 5.7 Air temperature for 2000 simulation of Charlotte Harbor circulation.
94
5.3 Simulations for 1986 Hydrodynamics
During the calibration process, a few model coefficients and/or boundary and initial
conditions were adjusted to allow accurate simulation of observed data. As shown in Table
5.2, measured data of tidal stage, current, and salinity at several locations throughout the
Charlotte Harbor were provided by USGS (Goodwin, 1992). Figure 5.8 shows the locations
of tidal stages and Figure 5.9 shows the locations of current velocity and salinity
measurement. Water elevation data at USGS station 4 (Venice) were compared with those
at USGS stations 5 (Cayo Costa) and 8 (Gulf of Mexico at Fort Myers) which are located in
the Gulf of Mexico (Figure 1.1). The tidal elevation and phase at stations 4 and 5 are almost
identical. Tidal range at Fort Myers beach (station 8) is larger than those at stations. The
reason for the larger tidal range at Fort Myers is probably due to the abrupt change in the
direction of shoreline at the southern end of Sanibel Island, rather than different offshore
tides. Therefore, water level data at Venice station were used as tidal boundary conditions
along all open boundaries. All specific conductance measurements were converted to salinity
concentrations in parts per thousand using the conversion equation by Miller (Goodwin.
1992). These data were used for comparison with model simulations of water level, velocity
and salinity.
Before baroclinic simulations are performed, it is necessary to generate an initial
salinity field because, in general, salinity is much slower to adjust to initial transients than
water level or currents. The initial salinity field for the 1986 simulation is generated by
"spinning up" a prescribed salinity field for a sufficiently long period. The salinity values
at the end of the simulation are then used the initial values for the 1986 simulations. This
process is discussed below in more detail.
95
First, a salinity field is created by linearly interpolating measured salinity onto the
Charlotte Harbor grid. The bottom salinity data measured by USGS during 1986 are used
to create this salinity field. Next, a salinity assimilation term is added to the salinity transport
equation in the model. This term forces the salinity in a selected cell to approach a specific
value, in this case, the initial value. The six USGS measured bottom salinity values shown
in Figure 5.6 are chosen to be assimilated into the simulation. These values were chosen
because they were measured closest to the July 9 starting date of the 1986 simulations. The
assimilation term , which appears on the right side of the salinity equation, takes on the
following form (Sheng and David, 2002):
T
where S"iJk is the simulated salinity at the n-th time level, (SA)iJk is the value to be
assimilated, i.e., the initial salinity values, and T is the assimilation period, which is set to
30 days for Charlotte Harbor simulation. It can be seen, when S"ijk is not equal to (SA)iJk a
force is created to drive S"iJk, toward the value of (SA)ijk. Because of the addition of the
salinity assimilation term, the salinity at the six USGS sites at the end of a simulation will
be very close to their measured values at the beginning of the simulation.
A 30-day spin-up simulation was performed from May 28 to June 27 during a dry
season with all barotrophic forcing mechanisms (tides, river discharges, wind) to allow water
level, velocity and salinity field to reach a dynamic steady-state throughout the computational
domain. Using the last time step surface elevation, velocity, and salinity of the spin-up
simulation as the initial condition, the Charlotte Harbor circulation and salinity transport
from June 27 to July 30 was simulated with tidal forcing, wind field and river discharges as
boundary condition.
96
Site
(I ,J) Location in
Duration
Data Type
Number
Computational
domain
1
(29,106)
6/15-8/31,1986
Tidal Stage
2
(40,102)
7/15-8/31,1986
Tidal Stage
3
(23, 81)
6/15-8/31, 1986
Tidal Stage
4
Outside, at Venice
6/30-8/31,1986
Tidal Stage
5
(12, 81)
6/30-8/31, 1986
Tidal Stage
6
(14, 62)
7/10-8/31,1986
Tidal Stage
7
(28, 47)
8/14-8/31,1986
Tidal Stage
8
(33, 29)
6/26 - 8/20, 1986
Tidal Stage
SI-1
(33,100)
7/9-8/ 6, 1986
Velocity & Salinity
SI-2
(30, 90)
11 9-8/ 8, 1986
Velocity & Salinity
SI-3
(35, 86)
7/9-8/ 6, 1986
Velocity & Salinity
SI-4
(11, 96)
7/10-7/15, 1986
Velocity & Salinity
SI-6
(18, 76)
7/10 - 7/23, 1986
Velocity & Salinity
SI-7
(17, 65)
7/10 - 7/23, 1986
Velocity & Salinity
SI-8
(20, 51)
7/16 - 7/20, 1986
Velocity & Salinity
97
o
o
o
o I—
o
o
CO
o
o
o
m
(j>
CM
£ 8
w o
£ CM
o
o
o
o
o
o
a>
CM
_
J_
350000
• Tidal Measurement
Stations (1 to 8)
0 Discharge Measurement
Stations (9 to 15)
ite
r
4
100
Y>Ur
7 ^#
J I . L
J_
J L
J I I L
J__l L
J_
375000
400000
Easting (m)
425000
450000
Figure 5.8 Locations of 1986 water level and discharge measurement stations of USGS
(Goodwin, 1992).
98
o
o
o
o
o
o
CO
w o
U) o
.b en
.C CM
t
o
z
o
o
o
to
CM
en
CM
o
o
o
o
o
en
c\j
J L
±
Velocity and Salinity
Measurement Stations
(SI-1 toSI-10)
SI-3 /
s|-9\ %¥%&*
V
SI-10
J I l_L
J_
J I L
_L
J I I L
-L
J__l I L
350000 375000 400000 425000 450000
Easting (m)
Figure 5.9 Locations of 1986 velocity and salinity measurement stations of USGS
(Goodwin, 1992).
99
5.3.1 Sensitivity and Calibration Simulations
Several simulations were performed in order to test the consistency and sensitivity
of numerical grid, bottom roughness, horizontal diffusion coefficient, and bathymetry. To
quantify the model sensitivity to variations of these model parameters, time series statistics
are presented in terms of the root-mean-square error (RMS) and normalize RMS error, define
as the ratio between the RMS error and the observed range. The normalized RMS error gives
a more meaningful indication on model's ability to reproduce the tidal signal and salinity at
each station. The RMS error is calculated according to:
E =
■ , N -11/2
,r / j \ simulated data )
TV i
(5.2)
where N is the total number of measured data, Ssimulated is the model result, and Sdata is
measured data at each station.
To determine which non-tidal boundary and forcing conditions are most important
to Charlotte Harbor estuarine system modeling, a series of 1986 simulations were performed
with a different boundary and forcing conditions removed in each run (Table 5.3). These
simulations show that baroclinc forcing is the most important factor in simulating water
level, flow, and salinity at overall available measured stations.
Table 5.3 The effect of removing selected boundary conditions on the accuracy of simulated
water level, velocity and salinity in July 1986. Values shown are average RMS differences
vs. baseline simulation at all data stations.
Boundary and Forcing
Water level
U Velocity
V Velocity
Salinity (ppt)
Condition Removed
(cm)
(cnr/s)
(cnr/s)
Baroclinic Forcing
2.35
3.17
2.05
3.08
Wind
1.82
2.64
2.08
1.03
River Discharge
0.57
1.15
0.84
1.05
As shown in Equation 3.28, the bottom drag coefficient is defined as a function of
100
the bottom roughness, z0. The default value of bottom roughness, z0, used for Charlotte
Harbor simulations is chosen to be a constant 0.3 cm over the entire domain based on
running with the several bottom roughness values which range between 0.01 and 1 (Table
5.4). Because of the numerous different bottom types in the estuary, a constant value of
bottom roughness may not be appropriate. To determine whether a spatially varying bottom
roughness or a constant bottom roughness produces better simulated circulation and
transport, a series of 1998 simulation was performed for the Indian River Lagoon (Davis and
Sheng, 200 1 ). The results show that simulations with the spatially varying bottom roughness
have slightly smaller errors in water level but slight larger errors in salinity and flow. Since
the varying bottom roughness does not improve the simulated circulation and transport
significantly, the constant bottom roughness of 0.3 cm is used for all simulations in this
study.
Table 5.4 The effects of varying bottom roughness, z0, on the accuracy of simulated water
level velocity, and salinity in July 1986. Values shown are average RMS errors at all data
stations.
Bottom roughness (cm)
Water level
(cm)
Velocity
(cm:/s)
Salinity (ppt)
0.1
8.11
7.87
1.71
0.3
8.12
7.88
1.71
1
8.11
7.87
1.71
As shown in Equations 3.2 and 3.3, the sub-grid scale motion is estimated with a
horizontal diffusion coefficient, AH. A default value of 10,000 cm2/s was used for Charlotte
Harbor simulations after performing several simulations using different coefficient values
(Table 5.5). The results show little difference in water level and salinity.
Besides adjusting boundary conditions and model coefficients, it is also important to
use more accurate grid and bathymetry. To improve the accuracy of bathymetry in Charlotte
101
Harbor, bathymetric surveys were performed in Caloosahatchee River and San Carlos Bay,
and upper Charlotte Harbor by SFWMD and SWFWMD, respectively.
Table 5.5 The effect of varying horizontal diffusion, AH, on the accuracy of simulated water
level, velocity and salinity in July 1986. Values shown are average RMS errors at all data
Horizontal diffusion
Coefficient (cm2/s)
Water level
(cm)
Velocity
(cm2/s)
Salinity (ppt)
5000
8.12
7.86
1.75
10000
8.12
7.88
1.71
20000
8.07
7.88
1.64
Overall, the model is able to simulate the surface elevation and salinity within 10%
normalized error with maximum value. This provides validation that the hydrodynamic
model reproduces the basic circulation of the Charlotte Harbor estuarine system.
To supplement this short term calibration of July 1986, long term model calibration
was conducted using one year water level and salinity data in Caloosahatchee River during
2000. This 2000 model calibration include several test simulations to investigate effects of
bottom roughness, salinity advection schemes, and grid resolution and bathymetry.
5.3.2 Results of the 1986 simulation
For the 1986 simulation, the 92x129 grid and updated bathymetry were used along
with boundary conditions and model parameters described in Table 5.6. Water level, current
velocity, and salinity are compared both qualitatively and quantitatively with measured data,
where available.
Water Level
Calculated RMS errors between simulated and measured water level during the July
1986 simulation are shown in Table 5.7. The normalized RMS errors are less than 10% at
all water level stations, demonstrating the model's ability to accurately reproduce surface
102
elevation in the system. The highest errors at stations 1 and 2 can be attributed to the
relatively coarse horizontal grid near the river where the gages were located.
Table 5.6 A summary of boundary conditions and model parameters used in the 1986
simulation. .
Boundary Condition or Model Parameter Value
Tidal Forcing Measured
Wind speed and direction Measured at 2 stations
Fresh Water Discharge Estimated (for Estero Bay) and Measured
Bottom Roughness Constant (0.3 cm)
Horizontal Diffusion Constant (10000 cm2/s)
Horizontal grid 92 x 129
Vertical Layers 8
Table 5.7 Calculated RMS errors between simulated and measured water level in July 1986.
Station Number RMS error (cm) Range (cm) % RMS error
1
9.59
104.24
9.20
2
9.61
104.55
9.19
3
6.61
90.83
7.28
5
9.55
117.05
8.16
6
7.56
107.90
7.00
7
5.78
144.48
4.00
Average
8.12
111.51
7.47
Figure 5.10 shows the comparison between measured and simulated water level at
six Charlotte Harbor stations. The result shows a good agreement between data and model
results, in both amplitude and phase. From these figures, it can be seen that tidal range
decreases about 15% from Boca Grande Pass (site 5) to Peace and Myakka Rivers (sites 1
and 2). Also, the tidal wave takes about two to three hours to reach these rivers.
Figure 5.11 shows the spectra of water level (data and model results) for the same
103
stations. The spectral curves reveal major energy bands centered around two tidal
frequencies: the diurnal band (between 0.8 to 1.2 cycle per day) and a semi-diurnal band (1.8
to 2.2 cycles per day). These two bands combined to represent the tides propagating from
Gulf of Mexico into the estuarine system. The secondary energy bands, representing
nonlinear interactions between diurnal and semi-diurnal tides with complex geometries and
bathymetric features inside the estuary, are represented by the relatively small third and
fourth diurnal peaks. Comparisons show that the model is able to capture both the
amplitudes and phases well.
Velocity
The simulated and measured horizontal velocities at selected sites are shown in
Figure 5.12. The angle of the vector shows the azimuth of the flow vector at a specific time,
and the length of the vector represents the magnitude of the flow. It is important to point out
that the observed velocity is measured at a discrete point in space and is subjected to many
local influences that are not well represented in the model. These influences include physical
features such as small channels, depression shoals, and mounds that are not resolvable at the
model scale, but often influence the distribution of velocity at that location. Because of local
influences, discrete measurement points are not consistently reliable indicators of the general
velocity characteristics (Goodwin, 1996). The model results are spatially averaged values
over a grid cell, hence are not expected to agree exactly with discrete data.
Site SI-1 is located in northern Charlotte Harbor in an area of converging river flow
and having a complex bathymetry that creates a complex lateral distribution of velocity. The
simulated velocity at SI-1 agrees well with measured data. However, some simulated
velocity magnitudes were 50% smaller than measured magnitudes and there was some
104
deviation in the azimuth of more than 30 degrees. Since the current meter was located near
the surface at this station, this station could be influenced by freshwater inflow, wind and any
other effects as indicated by Goodwin (1996).
At Si-2 and Si-3, there is good agreement between simulated and measured velocities.
In the Pine Island Sound, site-7 is located in the middle of several islands, and the observed
data were strongly affected by local features that cannot be resolved at the scale modeled.
However, agreement between simulated and measured data for this site is quite good.
Calculated RMS errors between simulated and measured current velocity are shown
in Table 5.8. The average normalized RMS error is less than 20%. The RMS error is quite
reasonable in comparison to the range of velocity, considering the fact that many local
influences that are not well represented in the model.
Table 5.8 Calculated RMS errors between simulated and measured current velocity for 1986
Station
RMS error (cm/s)
Range (cm/s)
% RMS error
Number
U
4.38
14.34
SI-1
V
5.97
30.54
19.56
u
11.70
21.20
SI-2
V
13.14
55.19
23.82
V
7.16
18.72
SI-3
V
6.78
38.25
17.72
u
3.06
8.79
SI-6
V
10.82
34.79
31.10
Average
7.88
19.40
Salinity
Figure 5.13 shows the near-bottom simulated and measured salinity at USGS stations
105
SI-1,2,3,6,7, and 8 inside of Charlotte Harbor. The difference between simulated and
measured salinity does not exceed 2 ppt except at station SI-7.
Calculated RMS errors between simulated and measured salinity during July 1986
simulation are shown in Table 5.9. The results show the model's ability to simulate salinity
within 10% accuracy. The highest error is at station SI-7. The measured salinity shows very
strong daily fluctuation which is five times greater than simulated values. This location is
near the Captiva pass which is possibly affected by local features such as small channels,
depressions, shoals, and mounds as mentioned before.
Table 5.9 Calculated RMS errors between simulated and measured salinity for 1986
simulation .
Station Number RMS error (ppt) Maximum (ppt) % RMS error
SI-1
1.23
24.01
5.15
SI-2
1.47
25.42
5.77
SI-3
1.51
27.05
5.60
SI-6
0.61
33.83
1.81
SI-7
3.52
33.82
10.42
SI-8
1.90
25.38
7.50
Average
1.71
26.55
6.04
Flow Patterns
Typical averaged flow patterns during one tidal cycle, based on the water level at
open boundary, for August, 6 on 1986 are shown in figure 5.14. Easterly flow on the western
Florida shelf is dominant during the flood tide, whereas northerly and westerly shelf flows
are dominant during the ebb tide. Two major passes in the Charlotte Harbor estuarine system
are Boca Grande Pass and San Carlos Pass. The large flood and ebb flows through the Boca
Grande Pass affects the tidal prism in upper and lower Charlotte Harbor and the flow through
106
San Carlos Pass satisfies the tidal prism in the Caloosahatchee River, San Carlos Bay, the
lower part of Matacha Pass and lower extremity of Pine Island. The flow through Gasparilla,
Captiva, and Redfish passes have effects that appear to be limited to the local area. In Pine
Island Sound, the flow is very low during flood and ebb tide periods. However, there is some
water transport through Pine Island Sound during high slack and low slack tide periods.
During near high slack tide, there is northerly flow from San Carlos Bay, while there is
southerly flow during near low slack tide.
Residual Flow and Salinity Patterns
Figure 5.15 show the 29-day residual circulation during July 2 to July 30, 1986. In
the upper Charlotte Harbor, there are opposite direction surface residual flows which are the
landward surface residual flow in relatively deep channel along the right side channel and
seaward surface residual flow along the left side shoreline. In the bottom residual flow, the
landward flow is dominant across the channel.
In the Pine Island Sound and Matlacha Pass, the northward residual flow is dominant
in the surface, while there is a week southward flow in the bottom. The surface flow come
from upper Charlotte Harbor, Pine Island Sound, and Matlacha Pass create a strong seaward
surface residual flow at the lower Charlotte Harbor.
The offshore residual flow in the Gulf of Mexico is northward. The strong outflows
from the Boca Grande Pass and San Carlos Pass create the clockwise gyres near the surface
in the Gulf of Mexico.
The residual salinity has a uniform vertical and horizontal distribution in the lower
Charlotte Harbor, while the salinity has horizontal inclination across the channel and
vertically stratification in the upper Charlotte Harbor. In upper Charlotte Harbor, the surface
107
salinity along the right side shoreline has greater value than that along the left side shoreline
because of the residual surface flow pattern. The residual salinity field shows a vertical
stratification of 2 ppt at the Peace River mouth since the strong river discharge and mixed
tide from Gulf of Mexico create two layered flow (seaward flow at surface and landward
flow at bottom) in this area. A two layered flow and salinity stratification is often presented
at the Charlotte Harbor estuarine system. The significant vertical stratification can suppress
vertical mixing and lead to the formation of pycnocline and hypoxia in the upper Charlotte
Harbor (Stoker, 1992; CDM, 1998). Therefore, understanding of the density-driven
circulation is critical to the management of Charlotte Harbor estuarine system.
108
100
100 r
■100
190
Myakka River at El Jobean (site 1) July, 1986
Julian Day
Peace River at Punta Gorda (site 2) July, 1 986
200
210
Julian Day
Charlotte Harbor at Bokeelia (site 3) July, 1 986
IUU
?
o
50
i
<D
|
0
i-
CO
-50
'.
210
Julian Day
Figure 5.10 Comparison between simulated and measured water level in July 1986.
109
Gulf of Mexicoat Cayo Costa near Boca Grande (site 5) July, 1 986
100 r
-100
100 r
Julian Day
Pine Island Sound near Captiva (site 6) July, 1 986
200.
Julian Day
B. Fort Myers Beach (site 8) July, 1 986
210
190
Figure 5.10 continued
ulian Day
110
Site 1
Site 2
>-
a ,
T105
to'
■D
c
fto1
o
a
12 3 4
Frequency (cycle/day)
Site 6
Simulated
Measured
Frequency (cycle/day)
||Q
o
Q.
710-
5iry
f
C103
0)
TJ
|io2
w
|10°
o
CL
2 3 4
Frequency (cycle/day)
Site 5
Frequency (cycle/day)
Site 8
Simulated
Measured
12 3 4
Frequency (cycle/day)
Figure 5.1 1 Comparison between simulated and measure spectra of water level in July
1986.
Simulated Velocity
111
Measured Velocity
m161910(SI-1), July 1986
o
<v
o
o
I I I I I I
J l_
10 11 12 13
14
15 16
o
o
o
W^w
y/Mi
WYV
LP
-i I I L.
9 10 11 12 13 14 15 16
Figure 5.12 Comparison between simulated and measured velocity data in July 1986.
112
Measured Velocity
m809110(SI-2), JULY 1986
o
I
o
o
Simulated Velocity
o
o
Figure 5.12 continued.
Measured Velocity
m810110(SI-3), July 1986
o
111
i
o
I l_J I ^_l l_-l 1 L_J-
J
10
11
12
13
14
15
16
Simulated Velocity
o
0)
I
o
o
_J I I I I 1 1 L-
10 11
12
13
14
15
16
Figure 5.12 continued.
113
114
Measured Velocity
S
E
o
o
m149614(SI-6), July 1986
i ' i i i i i i i i i i — i — i — i — i — i — i — i — i — i — i
9 10 11 12 13 14 15 16
Simulated Velocity
1
\V\
-
i« M Ill
ftfMi
i i i i i i
■ .... i . . i i i i i i i i
9 10 11 12 13 14 15 16
Figure 5.12 continued.
115
SI-1 station, July 1986
SI-2 station, July 1986
35
30
25
P
L
L
£
=15
s
10
5
35
30
25
*■»
D.
320
= 15
10
5
9
simulated salinity
measured salinity
f90 195 . .. 200
Julian Day
205
SI-3 station, July 1986
;
H^^r.^y^r^^X'yW^''%-A
-
-
simulated salinity
measured salinity
. i
90 195 . ,. 200
Julian Day
205
35
35
30
25 h
Q.
S20
£15
s
10
5 -
9
90
simulated salinity
measured salinity
1 95 200 205
Julian Day
S 1-6 station, July 1986
simulated salinity
measured salinity
195 , .. 200
Julian Day
205
35
30
25
Q.
S20
£
£15
10
5
9
SI-7 station, July 1986
1 i ill ;i."iiV:c
i nihil!
\m\i
>j ys.-^
90
simulated salinity
measured salinity
195 . . . 200
Julain Day
205
35
SI-8 station, July 1986
30 -
25
C
a.
81
10
simulated salinity
measured salinity
90
195 , .. 200
Julian Day
Figure 5.13 Comparison between simulated and measured salinity in July 1986.
205
116
l Vw
Low Slack Tide
8/06/86 10:00
1 m/sec
Flood Tide
8/06/86 13:00
!»>»-•".,
/WW-'.'lf"':-*
j^.^
High Slack Tide
8/06/86 16:00
••• '-••'•-•-'■■••'••••••• '•■y/4/>y^y^k
••.••.•'.'•.'/.v:.;;^:$^
Ebb Tide
8/06/86 19:00
I^r<
'•.\\VWSavV
Figure 5.14 Typical flow pattern of Charlotte Harbor estuarine system during one tidal
cycle for August 6, 1986.
117
I
&
X.
V; v
<A fe*d
Residual Flow
at Surface
1 0 cm/sec
Residual Flow
at bottom
1 0 cm/sec
\x, \t h
Residual Salinity
at Surface
rca*
Residual Salinity
at bottom
,4r*
Figure 5.15 The 29-day residual flow and salinity for Charlotte Harbor estuarine system
during July 2 to July 30, 1986
118
5.4 Simulations for 2000 Hydrodynamics
1986 hydrodynamic simulation of Charlotte Harbor estuarine system was conducted
for 30 days to calibrate the major model coefficients and inputs. The field data did not cover
the Caloosahatchee River which is a very important segment of the study area. To
supplement 1986 calibration simulation, long-term model calibration was conducted using
one year water level and salinity data in the Caloosahatchee River during 2000. Figure 5.16
shows the salinity measurement stations in Caloosahatchee River. Water levels were
measured at Fort Myers and Shell Point. The validated CH3D model was then used to assess
the effects of Sanibel Causeway and the navigation channel in San Carlos Bay, and to
quantify the relationship between freshwater inflow and spatial and temporal salinity
distribution in the Caloosahatchee River.
5.4.1 Sensitivity and Calibration Simulations
Several simulations were performed in order to test the effects of grid configuration,
bottom roughness, bathymetry, and salinity advection scheme on salinity distribution in the
Caloosahatchee River. To study how the horizontal grid resolution affects simulated
circulation and transport within the estuarine system, two 2000 simulations are performed
using two different grids. Figure 5.17 shows the comparison of the coarse grid (71x92) and
the fine grid (92x 1 29). The fine grid has about two times finer resolution than the coarse grid
in San Carlos Bay and near the Caloosahatchee River mouth. Table 5.10 shows that the two
different horizontal grids affected the salinity less significantly than the water level. The
geometry and bathymetry at the river mouth play a very important role for water circulation
inside the river. The coarse grid tends to produce more error in water level simulation than
fine grid because the coarse grid does not accurately to represent the islands and navigation
119
channels in the San Carlos Bay and Caloosahatchee River mouth. The fine grid (92x129)
is used for all simulations for Charlotte Harbor estuarine system.
o
o
o
o
CD
CD
CM
T
BR31r ^#^
§ityf Fort Myers
JKTSF***
1
J L
h.-B-K^ I I I
400000
410000 420000
Easting (m)
S79
-L
430000
Figure 5.16 Locations of the available 2000 water level and salinity measured stations at
Caloosahatchee River operated by SFWMD.
120
depth
(cm)
■i 1000
900
800
700
600
500
400
300
200
100
0
t
depth
(cm)
1000
900
800
700
600
500
400
300
200
100
0
•A>
Figure 5.17 The comparison of the coarse grid (71x92) and the fine grid (92x129) for
Charlotte Harbor estuarine system.
121
Table 5.10 The effect of horizontal grid resolution, on the accuracy of simulated water level
and salinity. Values shown are average RMS errors for 2000 calibration at all available
stations. Values shown in parenthesis are % RMS error normalized by maximum values.
Variables Station 71x92 92 x 129
S79
S
B
1.69(4.62)
2.54(6.93)
1.46(3.98)
1.84(5.03)
Salinity
(Ppt)
BR31
S
B
1.17(3.20)
1.89(5.16)
1.55(4.22)
2.06(5.61)
Fort Myers
S
B
1.55(4.23)
2.07(5.63)
1.85(5.03)
2.63(7.18)
Shell Point
S
B
3.78(10.29)
3.37(9.17)
3.76(10.25)
3.53(9.64)
Sanibel
S
B
1.85(5.05)
2.05(5.59)
1.94(5.28)
2.00(5.45)
average
2.20(5.99)
2.26(6.17)
Water level
Fort Myers
9.08(5.44)
5.51(3.30)
(cm)
Shell Point
6.92(4.83)
4.25(2.96)
average
8.00(5.14)
4.58(3.13)
Vertical salinity stratification in the Upper Charlotte Harbor and Caloosahatchee
River is a common seasonal occurrence (Environmental Quality Laboratory, Inc., 1979). In
high river inflow events, a stable vertical salinity gradient is created which suppresses
vertical mixing unless there are sufficient mixing by wind or tide. Therefore, the vertical
grid resolution is an important factor to reproduce vertical salinity distribution which is a one
of major cause effects of hypoxia in upper Charlotte Harbor. To study the effect of varying
the number of vertical layers used by CH3D model, 2000 simulations were performed with
4 and 8 vertical layers (Table 5.11). Since a one year simulation of hydrodynamics, sediment
transport, and water quality took several days, the simulation with more than 8 vertical layers
which required the additional computational time, is not feasible for this study. While the
simulated water level RMS error changes little, the averaged simulated salinity RMS error
122
improves about 1 ppt with 8 layer simulation. The largest improvement in simulated salinity
was achieved at Shell Point station with 3 ppt using 8 vertical layers. To reproduce vertical
stratification for salinity and water quality simulation, eight layer simulations are deemed
appropriate for the Charlotte Harbor estuarine system.
Table 5.1 1 The effect of vertical grid resolution, on the accuracy of simulated water level
and salinity. Values shown are average RMS errors for 2000 calibration at all available
stations. Values shown in parenthesis are % RMS error normalized by maximum values.
8 layers
Variables
Station
4 layers
S79
S
B
1.85(5.08)
2.09(5.73)
1.46(3.98)
1.84(5.03)
Salinity
(Ppt)
BR31
S
B
1.86(5.11)
2.37(6.51)
1.55(4.22)
2.06(5.61)
Fort Myers
S
B
2.34(6.43)
2.30(6.31)
1.85(5.03)
2.63(7.18)
Shell Point
S
B
6.82(18.70)
5.31(14.55)
3.76(10.25)
3.53(9.64)
Sanibel
S
B
2.48(6.80)
4.02(11.02)
1.94(5.28)
2.00(5.45)
average
3.14(8.63)
2.26(6.17)
Water level
Fort Myers
4.50(2.70)
5.51(3.30)
(cm)
Shell Point
4.42(3.08)
4.25(2.96)
average
4.46(2.89)
4.58(3.13)
In Caloosahatchee River, water level and salinity are very sensitive with the varying
bottom roughness while the varying bottom roughness does not affect the simulated
circulation and transport significantly in the estuary in July 1986. However, several 2000
simulations were performed using different coefficient value to determined the effect of
varying the bottom roughness on the accuracy of the simulation circulation and transport.
(Table 5.12). With increasing bottom roughness, accuracy of water level is improved while
that for salinity is worse. A constant bottom roughness of 0.3 cm is used for all simulations.
123
Table 5.12 The effect of varying bottom roughness, z0, on the accuracy of simulated water
level and salinity in 2000. Values shown are average RMS errors at all data stations.
Bottom roughness (cm) Water level (cm) Salinity (ppt)
0.2 4.94 2.28
0.3 4.87 2.29
04 4J38 231
Model sensitivity study was conducted to investigate the relative accuracy of the
salinity advection schemes including upwind, QUICKEST, and Ultimate QUICKEST
methods. Time series comparisons of measured and simulated surface salinity using three
advection schemes at three stations, Shell Point, Fort Myers, andBR31, are shown in Figures
5.18 to 5.20. The salinity comparison at Shell point does not show much difference for all
three schemes, while at Fort Myers and BR31, the simulated salinity with upwind scheme
is much higher than those for the other advection schemes and measured salinity. The
advection scheme plays a very important role for salinity simulation at cells inside river
where there is usually very strong salinity gradient from the river mouth at Shell Point to the
upstream. Lower order advection scheme such as upwind scheme tends to produce more
error in salinity simulation because of its inherently high numerical diffusion. A series of
vertical-longitudinal salinity profiles along the axis of the river during slack water is shown
in Figure 5.21 to compare salinity distribution along the river simulated by various advection
schemes. The salinity distribution obtained with the QUICKEST method shows similar
pattern with that for Ultimate QUICKEST method. The upstream salinity near BR3 1 shows
much higher value with upwind scheme because salinity is quickly diffused from river mouth
to the upstream. With the upwind scheme, it is difficult to reproduce the vertical salinity
stratification which often occurs during high river flow period, while the other advection
schemes produced the stratification.
124
Ultimate QUICKEST Scheme
200 250
QUICKEST Scheme
D.
a.
>.
4-1
c
200
Upwind Scheme
35 r
250
simulated salinity
measured salinity
300
300
200
250
300
350
350
350
Julian Day
Figure 5.18 A comparison between simulated and measured salinity at Shell Point using
Ultimate QUICKEST, QUICKEST, and upwind advection schemes.
125
Ultimate QUICKEST Scheme
35
30
a- 25
Q.
S 20
i 15
5
w 10
5
0
200 250
QUICKEST Scheme
35
30
S? 25
Q.
3 20
I 15
CO
w 10
5
0
200
Upwind Scheme
250
^A^^
-
simulated salinity
measured salinity
300
300
350
350
200
250
300
350
Julian Day
Figure 5.19 A comparison between simulated and measured salinity at Fort Myers using
Ultimate QUICKEST, QUICKEST, and upwind advection schemes.
126
Ultimate QUICKEST Scheme
20
-, 15
a
S
.IT 10
c
TO
200 250
QUICKEST Scheme
20
^ 15
a
a.
£• 10
c
nj
0
200
Upwind Scheme
250
~ 15 -
Q.
a
£* 10
200
250
simulated salinity
measured salinity
300
300
350
350
300
350
Julian Day
Figure 5.20 A comparison between simulated and measured salinity at BR31 using
Ultimate QUICKEST, QUICKEST, and upwind advection schemes.
127
Current Time : 9/ 7/2000 1 9:00
100 Shell Fort
Point Myers
-40000
BR31
-30000 -20000 -10000
Distance froma S79 (cm)
Salinity
S79 <PP*>
100 Shell
Point
Fort
Myers
BR31
-40000
-30000 -20000 -10000
Distance from S79 (cm)
Salinity
S79 (ppt)
BR31
-40000
10000
Salinity
S79 (Ppt)
-30000 -20000
Distance from S79 (cm)
Figure 5.21 Simulated longitudinal-vertical salinity along the Caloosahatchee River at
slack water before flood on September 7, 2000.
128
Table 5.13 shows the effect of salinity advection scheme on the accuracy of simulated
water level and salinity. The results show slight difference between water level and salinity
simulated with the Ultimate QUICKEST scheme and the QUICKEST scheme, while the
simulated salinity RMS errors with upwind scheme is much greater. Maximum
improvement in simulated salinity is achieved when the QUICKEST scheme is used because
most of the model parameters and boundary conditions are calibrated using the QUICKEST
scheme. However, because of the higher cost of model simulations using the Ultimate
QUICKEST scheme, coupled with only a marginal improvement in the simulated results, the
QUICKEST scheme is deemed appropriate for this study.
Table 5.13 The effect of varying salinity advection scheme on the accuracy of simulated
water level and salinity in 2000. Values shown are average RMS errors at all data stations.
Advection scheme Water level (cm) Salinity (ppt)
2.35
2.34
3/73
Beside adjusting boundary condition and model coefficients, it is also interesting to
see how varying the model grid and bathymetry affects simulated circulation and transport
within the estuary. Because flow through the Caloosahatchee River mouth is crucial to the
water level and salinity inside the river basin, it is useful to perform a numerical simulation
to study how the flow through the river mouth is affected by its cross-sectional area. As was
discussed in the previous section, the Charlotte Harbor grid bathymetry was developed by
interpolating measured bathymetry onto the entire grid followed by a simple smoothing
scheme. Although the bathymetry in upper Charlotte Harbor and inside Caloosahatchee
River were updated by measured data by SWFWMD and SFWMD, the resolution of
bathymetry in San Carlos Bay area, including the Caloosahatchee River mouth, is very coarse
Ultimate QUICKEST
5.56
QUICKEST
5.63
Upwind
5.36
129
(about 300 m resolution). To study how the accuracy of grid bathymetry affects simulated
the circulation and transport, several 2000 simulations were performed using grid systems
with modified bathymetry data. The first simulation used a modified grid bathymetry which
had a minimum depth of 2.5 m (1.5 m for baseline simulation). This simulation was
performed to determine how important the "shallowness" of the river is to the circulation and
transport. The second simulation used a bathymetry that included an artificially dredged
channel through out the entire river. The navigation channel included in the grid bathymetry
at a depth of 3.5 m (NAVD88). Since the grid system is too coarse to adequately resolve the
navigation channel, forcing a channel into the grid system resulted in over estimation of the
cross-sectional areas within the river. Table 5.14 compares the results of the year-long
simulations performed using the original and two modified grid bathymetries. While forcing
an deeper navigation channel into the grid system had little effect, imposing a 2.5 m
minimum cell depth slightly improved simulated salinity and greatly worsened water level
through the Caloosahatchee River. Since neither modified grid system improved the 2000
simulation, the standard bathymetry is used for all subsequent simulations.
Table 5. 14 The effect of modifying bathymetry on the accuracy of simulated water level and
salinity in 2000. Values shown are average RMS errors at all data stations.
Simulation description
Water level (cm)
Salinity (ppt)
Standard bathymetry
4.88
2.29
Minimum depth of 2.5 m
6.14
1.90
Deeper navigation
channel
4.90
2.32
Overall, the model is able to simulate water level and salinity within 10 % normalized
RMS error with the maximum measured values. This indicates that the hydrodynamic model
reproduces the basic circulation and salinity transport of the Caloosahatchee River so that
130
this model can be used to simulate water quality processes.
5.4.1 Results of the 2000 Simulation
This section summarizes the calibration of the long term Charlotte Harbor circulation
and transport model using the field data collected in January to December 2000. In addition,
the mathematical model of the temperature transport with heat flux model as surface
boundary condition was applied to improve the simulation of circulation in the study area.
Based on the results, it is apparent that the model accurately simulated the observed water
level, salinity, and temperature distribution in the Caloosahatchee River. Table 5.15
summarizes the boundary conditions and model parameters for 2000 simulation.
Table 5.15 A summary of boundary conditions and model parameters used in 2000
simulation.
Boundary Condition or Model Parameter Value
Tidal Forcing Measured at Naples
Wind speed and direction Measured at 3 stations
Fresh Water Discharge Estimated for Estero Bay and Measured at
Peace, Myakka, and Caloosahatchee Rivers
Bottom Roughness Constant (0.3 cm)
Horizontal Diffusion Constant (10000 cm2/s)
Horizontal grid 92 x 129
Vertical Layers 8
Water Level
When measured water level at an open sea is used as a tidal boundary condition, it
is necessary to unify datum level with bathymetry and water level measured in the estuary.
For the 2000 simulation of Charlotte Harbor, the datum level for all measured water levels
and the bathymetry were converted to NAVD88. Water level at the Naples station, which
is the tidal boundary condition, was leveled to MLW by NOAA CO-OPS and converted to
131
NAVD88 according to tidal bench mark at Naples (26° 7.8'N, 81° 48.4W).
Calculated RMS errors between simulated and measured water level for 2000
simulations are shown in Table 5.16. The normalized RMS errors are less than 3% at all
available water level stations, demonstrating the model's ability to accurately reproduce
surface elevation in the system.
Table 5.16 Calculated RMS errors between simulated and measured water level for 2000
simulation
Station Name
RMS
! error
(cm)
Max
imum Range
(cm)
% RMS error
Fort Myers
5.51
166.80
3.30
Shell Point
4.25
143.26
2.96
Average
4.88
155.03
3.13
Figure 5.22 shows a year long and 20 days' comparison between simulated and
measured water level at Shell point and Fort Myers. Although the normalized RMS errors
are very small, the model constantly overestimated at Fort Myers and underestimated at Shell
Point station. A probable sources of these differences could be the bathymetry, grid
resolution, bottom roughness, and open boundary condition. In the Caloosahatchee River,
there is a very narrow navigation channel which the current grid system could not resolved.
The bottom roughness is a very sensitive parameter, specially in the river environment, as
shown in previous calibration. This study uses a constant bottom roughness of 0.3 value, but
there is no direct measurement of the bottom roughness for this estuarine system. This might
cause some error in model simulation. In this study, CH3D does not include calculation of
flooding/drying cell. Therefore, this study uses a minimum depth of 1.5 m which could not
resolve any shallow water below this depth. Once again, it has been demonstrated that model
is sensitive to the bathymetry and grid resolution, and the model accuracy will improve if
132
more accurate bathymetry and grid are used.
Salinity and Temperature
The circulation in Charlotte Harbor estuarine system is driven primarily by the mixed
(diurnal and semi-diurnal) tides from the Gulf of Mexico, as well as by wind and density
gradient. During periods of high freshwater inflow from the rivers, significant vertical
salinity stratification can be found in the Upper Charlotte Harbor and Caloosahatchee River.
A two layered flow and salinity structure characteristic of the "classic estuarine circulation"
is often present (Sheng, 1998). Therefore, understanding of vertical salinity distribution is
a very important part of hydrodynamic and transport simulations. The salinity is measured
at two locations in the water column referred to as "upper" and "lower". The vertical
positions of the salinity measurements taken by SFWMD are given in Table 5.17. For this
study, eight vertical layers are used by the CH3D model. The simulated salinity values at
eight vertical layers are interpolated vertically to allow comparison of salinity at the exact
location of the measured stations.
Table 5.17 presents the maximum and minimum measured salinity, the RMS error,
and the normalized RMS error (with respect to maximum salinity) for all stations during
2000. The results show the model's ability to simulate salinity is within 7% error except at
the Shell Point station. The Shell Point station is located at the Caloosahatchee River mouth,
which is very sensitive to bathymetry and grid resolution. As mentioned before, to improve
salinity at this point, finer grid resolution and bathymetry data will be needed. It should be
pointed out that % RMS error is not a good measure of model accuracy. At S79 and BR3 1,
although the % RMS error is low, the actual error is quite significant.
Figure 5.23 to 5.27 show the near-bottom and near-surface simulated and measured
133
salinity at SFWMD stations in Caloosahatchee River. The results show reasonable
agreement with measured data for both wet season and dry season except at S-79 and BR3 1 .
S-79 and BR31 stations are directly affected by the river boundary condition because these
two stations are located within a few grid cells from S-79. The water level and discharge at
the Lock and S-79 is controlled by 8 tainter gates and 2 sector gates whereas the river
boundary condition in CH3D is specified by the averaged flow rate. At these two stations,
there is only one grid cell across the relatively narrow width. Therefore, the current boundary
condition and grid resolution at S-79 and BR31 stations could not resolve the detail
characteristics of circulation and salinity transport. Furthermore, the bathymetry in this area
was produced from only a few cross sectional bathymetry data provided by SFWMD.
Table 5.17 Calculated RMS errors between simulated and measured salinity for 2000
simulation
Station
Name
Mean
depth
(cm)
Layers
Location
from
bottom
(cm)
RMS
error
(PPt)
Minimum
Salinity
(PPt)
Maximum
Salinity
(PPt)
%
RMS
error
S-79
591
upper
298
1.49
0.3
13.85
4.07
lower
103
1.88
0.3
14.55
5.11
BR31
678
upper
386
1.60
0.3
13.96
4.36
lower
251
2.11
0.3
15.51
5.76
Fort Myers
261
upper
227
1.84
0.3
18.66
5.01
lower
101
2.56
0.3
20.83
6.98
Shell Point
304
upper
209
3.82
2.4
34.50
10.41
lower
93
3.57
3.1
36.13
9.72
Sanibel
254
upper
135
1.96
11.2
36.69
5.34
lower
19
2.05
14.3
36.45
5.59
Average
2.29
6.23
Seasonal salinity patterns occurred in response to the variation in volume of
134
freshwater inflow. The highest measured salinity occurred during an extended period of low
flow in December 2000. During the wet season, the fresh water from S-79 reached into San
Carlos Bay. The salinity in San Carlos Bay was directly affected by the variation of river
discharge from S-79 station.
Figure 5.28 shows the simulated and measured temperature near the surface at the
Fort Myers station. The measured and simulated temperature show significant annual
variations in the Charlotte Harbor estuarine system. Water temperature ranges from an
average of about 30 °C during the summer to about 15 °C in December and January. The
daily fluctuation of water temperature is about 1 to 3 °C. The simulation result shows very
good agreement with measured data over seasonal and daily scales. The RMS error is less
than 2.5 °C at the Fort Myers station which is about 7% RMS error of the maximum
temperature of 32.20 °C.
Flow Patterns
Typical flow pattern (vertically averaged flow) in San Carlos Bay during flood tide
and ebb tide on August, 7 2000 are shown in figure 5.29. The large flood and ebb flows
through San Carlos Pass satisfy the tidal prism in the Caloosahatchee River, San Carlos Bay,
the lower part of Matacha Pass and lower extremity of Pine Island Sound. Due to Sanibel
Causeway and three navigation channels, the flood tide is separated into three major flows
along the navigation channel. The flow through the right side of causeway moves toward the
Caloosahatchee River , the flow through left side of causeway moves to Pine Island Sound,
and flood tide across the center channel of causeway follows navigation channel toward the
Matlacha Pass. During ebb tide, the flows are relatively weak and follow the opposite
directions. Along the East-West direction navigation channel, there are two direction flows:
135
the eastward flow during flood tide and the westward flow during ebb tide.
Residual Flow and Salinity Patterns
Figures 5.30 and 5.3 1 show the one year residual flow and salinity in 2000. In the San
Carlos Bay, the westward residual flow is dominant in the surface and bottom, while there
is a relatively week northward flow along the left and right side of shoreline. The strong
outflows from the left side of San Carlos Pass create the anti-clockwise gyres near the surface
at the outside of Sanibel Causeway. This anti-clockwise gyre is associated with the
clockwise gyre in the Gulf of Mexico which is shown in Figure 5.15 for the 1986 simulation.
The residual salinity has a relatively uniform vertical and horizontal distribution in
the San Carlos Bay, while the salinity has a strong vertical stratification at the
Caloosahatchee River mouth due to the strong river discharge and mixed tide from Gulf of
Mexico. Therefore, the density-driven circulation is a very important factor for
understanding the circulation near such large rivers as Peace and Caloosahatchee.
136
at Fort Myers
100
80 85
at Shell Point
3
85
200
300
simulated
measured
95
100
>
•
1 ,'S^fi»*^ $2
t r ff
t
k
30
200
300
simulated
measured
95
JuliarrDay
Figure 5.22 Comparison between simulated and measured water level for 2000
simulation.
100
137
atS79
30 r
25
S 20
E
ra 10
CO
ui00
30
25
1 2°
5 15
c
ro 10
CO
5
loo
S79 at Caloosahatchee River
20?Julian Day
Measured at near Surface
Simulated at near Surface
£±~
-*/v^ ■*■
w
200, ,. „
Julian Day
300
Measured at near Bottom
Simulated at near Bottom
n>
a -
200, ,.
Julian Day
w~
w
300
Figure 5.23 Comparison between simulated and measured salinity at S-79 in 2000.
138
S79 at Caloosahatchee River
a 20
c
■ 10
CO
^^\_
100
30
25
§■ 20
C
CO
15
10
100
20(3ulian Day
300
Measured at near Surface
Simulated at near Surface
ZL
/T*
200, „ _
Julian Day
300
Measured at near Bottom
Simulated at near Bottom
-&.
,- -^T^i ta
200, ,. „
Julian Day
4*
300
Figure 5.24 Comparison between simulated and measured salinity at BR31 in 2000.
139
S79 at Caloosahatchee River
"100
20l2)ulian Day
at Fort Myers
300
Measured at near Surface
Simulated at near Surface
200i .
Julian Day
300
Measured at near Bottom
Simulated at near Bottom
100
200
Julian Day
300
Figure 5.25 Comparison between simulated and measured salinity at Fort Myers in 2000.
140
S79 at Caloosahatchee River
Julian Day
at Shell Point
40 r
300
Measured at near Surface
Simulated at near Surface
TOO
20°, .. «
Julian Day
300
Measured at near Bottom
Simulated at near Bottom
Q.
a.
i
_£
100 200, ,. _ 300
Julian Day
Figure 5.26 Comparison between simulated and measured salinity at Shell Point in 2000.
J41
S79 at Caloosahatchee River
a.
a,
5
c
ro
20julian Day
near Sanibel Causeway
300
Measured at near Surface
Simulated at near Surface
Q.
Q.
5
15
(J)
200. ,
Julian Day
300
Measured at near Bottom
Simulated at near Bottom
TOO
Julian Day
300
Figure 5.27 Comparison between simulated and measured salinity near Sanibel
Causeway in 2000.
142
at Fort Myers
100
200
Julian Day
300
35
at Fort Myers
o
o
3 30
+-•
w
I—
Q)
Q.
£
0)
25 -
o
TO
?9
Simulated
Measured
40
145 150
Julian Day
155
160
Figure 5.28 Comparison between simulated and measured temperature at Fort Myers in
2000.
143
/
■ «c»
q V*
he I
" *&
&
7 o>St[
n 'A
1 \VW
"
i .^\
» • >
°!?> . ' -
": IIIIU1''
mitt
bo
fe7 ,>';V .*• £
I:7
n\,
ho,***
! > ' O^"
jus p ,.r . - './
«■■■ / .'■ >' / '' /y,^^p j)t)£ -
rf^f"
Ebb Tide
8/7/2000 0:0
Figure 5.29 Typical flow pattern of San Carlos Bay during ebb and flood tide for August,
7 on 2000.
144
/ '• K
w
• "it
I""'
s V * *
f.
us /
, ■ y-ugf .... \^'
"^s
1' y'y/W4mf§^ * r^
&
...
s
Residual surface flow
>
>^d
H
iw
J J)
Jliijf
A
->• •' ■■ -J3£u ^ >•<*/ -" •' -' /
SB^-fcN'
Residual bottom flow
— >
1 0 cm/s fen
n
■* <L:M>&
\v u
1
^:=s
^V-,7 W
Figure 5.30 One-year residual flow in San Carlos Bay in 2000.
145
Figure 5.31 One year residual salinity distribution in San Carlos Bay in 2000.
146
5.4.2 Applications of 2000 Hydrodynamic Simulations
Hydrologic Alterations
Hydrologic alterations in the San Carlos Bay and Caloosahatchee River have taken
many forms such as navigation channel and causeway. It has been suggested such large
transportation projects as the dredging of the IntraCoastal Waterway (ICW) and the
construction of Sanibel Causeway (SC) are linked to the decline of scallop populations in
Pine Island Sound (Estevez, 1998). A quantitative assessment of this suggestion is long
overdue. Moreover, to improve the natural environment of this area, it is necessary to
quantify the effects of these artificial hydrologic alterations. In this study, the calibrated
Charlotte Harbor model is used to evaluate the effects of these hydrologic alterations.
Figure 5.32 shows the location of eleven stations which are selected to quantify the
hydrologic alteration, and the locations of Sanibel Causeway (A) and IntraCoastal Waterway
(B). Three month simulations from April 9 to July 8, 2000 were conducted with two
hydrologic alterations in the Charlotte Harbor estuarine system. To test these effects, three
cases were considered: (1) present condition with Sanibel Causeway and existing bathymetry
(BASELINE), (2) hypothetical condition with Sanibel Causeway removed (NSC), (3)
hypothetical condition with ICW removed (NICW). Figure 5.33 shows the bathymetry and
shoreline for each case.
The simulated Charlotte Harbor circulation in the presence of and in the absence of
the Sanibel Causeway and IntraCoastal Waterway are compared in terms of the instantaneous
flow from July 3 to July 8, 2000. Figures 5.34 and 5.35 show the comparisons of water level
and salinity for each case at three selected stations: ST05 (Pine Island Sound), ST08 (San
Carlos Bay), and ST10 (Caloosahatchee River mouth).
147
The water level at all three stations show little difference among the three cases. The
salinity at Pine Island Sound (ST05) show little difference among the three cases, because
the impacts of these hydrologic alterations are very small and local. The results show that
water level and salinity transport are more affected by the absence of Intracoastal Waterway
(NICW) than by the absence of the Sanibel Causeway (NSC) except salinity at ST08 which
is located near the Sanibel Causeway. The freshwater from Caloosahatchee River and
saltwater from Gulf of Mexico are exchanged through the ICW. In the absence of the ICW,
salinity at San Carlos Bay (ST08) is increased, while salinity at Caloosahatchee River Mouth
(ST 10) is decreased at both surface and bottom layers because of the reduced flow and
salinity transport between the Caloosahatchee River and San Carlos Bay.
Calculated water level RMS differences between the baseline simulation and two
alteration cases from April to July 2000 are shown in Table 5.18 for 11 selected stations.
The RMS differences are less than 2 cm at all selected stations. The highest difference is
found at station 10 which is located at the Caloosahatchee River mouth. Calculated salinity
RMS differences between baseline simulation and the two cases during this period are shown
in Table 5.19. Salinity for the two alteration runs did not show much difference with the
baseline simulation except at station 10.
The 29-day residual flow and salinity patterns in San Carlos Bay are shown in Figure
5.36 (at a surface) and Figure 5.37 (at a bottom) during June 9 to July 8, 2000. Although the
results show some slight impact of the causeway on the residual flow as manifested by the
circulation gyres in the immediate vicinity of the causeway, there is no noticeable impact on
the salinity distribution in the San Carlos Bay area. The causeway islands did not block the
flow of saline ocean water from entering into the San Carlos Bay and reaching Pine Island
148
Sound because the causeway islands are already located in a very shallow region. While
there is strong residual flow along the IntraCoastal Waterway for the baseline simulation, this
strong residual flow vanished without this waterway. This can explain the effect of the
absence of IntraCoastal Waterway, which reduced the flow and salinity transport between
Caloosahatchee River and San Carlos Bay. Overall, the IntraCoastal Waterway and Sanibel
Causeway did not appear to show noticeable impact on the flow and salinity patterns in the
San Carlos Bay and Pine Island Sound.
o
o
o
un
CM
OJ
C\J
st-06
st-03
st-05 A
\
r~*
J L
J I L
±
J I I L
±
J I L
390000
395000
Easting (m)
400000
Figure 5.32 The locations of Sanibel Causeway and Intracoastal Waterway and stations
for comparing the effects of hydrologic alterations.
149
Depth
(NAVD88 :cm)
1500
450
400
350
300
250
200
150
100
Depth
(NAVD88 :cm)
■ 500
450
400
350
300
250
200
150
100
Figure 5.33 The comparison of bathymetry and shoreline for each hydrologic alteration case
scenarios which are Baseline, the absence of IntraCoastal Waterway, and the absence of
causeway.
150
ST05
E
o
CD
>
CD
S
CO
£
>
CD
CD
*-»
CO
2
CO
185
ST08
60
40 E-
20
o r
-20
-40
-60
185
186
186
187Julian Day388
Julian Day
189
190
190
BASELINE
No Causeway
NolCW
189
190
^JulianDay188
Figure 5.34 The comparisons of water level for three cases at three selected stations:
ST05 (Pine Island Sound), ST08 (San Carlos Bay), and ST10 (Caloosahatchee River
mouth).
151
ST05
35
Q.
330
c
I25
o
o
to
•C20
3
W
15
BASELINE
No Causeway
NolCW
186
Julian Day
ST08
35-
Q.
-S30
&
C
I25
a>
o
CO
"C20
3
3)
190
ST05
35
S30
c
I25
£
o
§20 1-
15
BASELINE
No Causeway
NolCW
186
. .- rO88
Julian Day
1 n rJ88
Julian Day
190
ST08
35
a.
-S30
c
s25
/ v V v
V '
E
0
?20
n
1
1
1
190
186
1 h J88
Julian Day
190
186
. ■• ,-J88
Julian Day
190
Figure 5.35 The comparisons of surface and bottom salinity for three cases at three
selected stations: ST05 (Pine Island Sound), ST08 (San Carlos Bay), and ST10
(Caloosahatchee River mouth).
152
Salinity
(PPt)
36
Figure 5.36 The comparisons of surface residual flow and salinity fields for three cases.
153
'p$r*r^&*. '-. ^'t^-—
Salinity
(PPt)
36
35
34
33
32
31
30
29
28
fe 27
i*ffim 26
Figure 5.37 The comparisons of bottom residual flow and salinity fields for three cases.
154
Table 5.18 The effects of hydrologic alteration on 2000 water levels. Values shown are
average RMS differences with baseline simulation for all selected stations.
Station Range (cm) RMS error RMS error RMS error for
Number for NSC forNICW NSCICW
1
126.58
0.41
0.16
0.38
2
115.75
0.76
0.31
0.93
3
115.01
1.11
0.40
1.31
4
115.19
0.55
0.44
0.92
5
116.69
0.36
0.19
0.49
6
114.70
0.43
0.23
0.60
7
114.53
0.55
0.30
0.80
8
112.49
0.63
0.33
0.90
9
110.60
0.62
0.68
1.15
10
84.55
0.40
1.36
1.22
11
111.49
0.61
0.39
0.94
Average
126.58
0.58
0.44
0.88
155
Table 5.19 The effects of hydrologic alteration on 2000 salinity. Values shown are average
RMS differences with baseline simulation for all selected stations.
Station
Layers
Range (ppt)
RMS error
RMS error
RMS error for
Name
for NSC
for NICW
NSCICW
1
surface
19.48
0.51
0.21
0.49
bottom
11.07
0.32
0.08
0.34
2
surface
18.16
0.78
0.20
0.80
bottom
11.80
0.24
0.06
0.24
3
surface
19.90
0.66
0.33
0.70
bottom
18.46
0.39
0.20
0.44
4
surface
28.45
0.40
0.41
0.50
bottom
26.58
0.32
0.21
0.31
5
surface
14.42
0.22
0.09
0.21
bottom
9.53
0.25
0.07
0.29
6
surface
16.46
0.27
0.14
0.28
bottom
12.07
0.23
0.08
0.24
7
surface
20.69
0.44
0.23
0.50
bottom
11.92
0.25
0.10
0.28
8
surface
20.95
0.44
0.43
0.61
bottom
20.26
0.42
0.40
0.57
9
surface
30.44
0.44
0.71
0.69
bottom
30.37
0.33
0.51
0.66
10
surface
30.01
0.32
1.62
1.79
bottom
30.45
0.38
1.33
1.24
11
surface
20.83
0.21
0.69
0.80
bottom
18.18
0.33
0.37
0.42
Average
surface
0.43
0.46
0.67
bottom
0.32
0.31
0.46
156
Freshwater Inflow and Salinity in the Caloosahatchee River
The Caloosahatchee River has been drastically altered for channelized flood-control
and navigational waterway. These changes have caused large fluctuations in freshwater
inflow volume, frequency of inflow events, timing of discharges, and water quality in the
downstream estuary (Chamberlin and Doering, 1998). Therefore, it is necessary to quantify
the impacts of freshwater inflow from S-79 on downstream estuarine system.
A minimum flow is defined by Ch.373.042(l) F.S (Florida State Law) as "the limit
at which further withdrawals would be significantly harmful to the water resources or
ecology of the area." Significant harm is defined in Chapter 40E-8 F.S. as "the degree of
impact requiring more than two years for the water (or biological) resource to recover".
Establishing quantifying relationship between freshwater inflow and the temporal and spatial
distribution of salinity in the river is the first step to determine MFL condition.
Salinity distribution in San Carlos Bay and adjoining water directly responds to the
fresh water inflow from Caloosahatchee River (Sheng and Park, 2001). In this study, the
calibrated Charlotte Harbor model is used to evaluate the effects of the fresh water inflow
from S79 to the salinity distribution in Caloosahatchee River. Using this calibrated model,
a series of vertical-longitudinal salinity profiles along the axis of the Caloosahatchee River
during one tidal cycle for wet and dry periods of river discharge at S-79 in 2000 are shown
in Figures 5.38 and 5.39, respectively. The fresh water (less than 1 ppt salinity) from S-79
reached the river mouth near Shell Point in wet season, while the fresh water stayed upstream
near BR3 1 during dry season. The salinity distribution at Caloosahatchee River shows much
difference corresponding to river discharge from S-79.
To quantify this relationship for Caloosahatchee River, the locations of specific
157
salinity values were calculated during the 2000 simulation periods. Figure 5.40 shows the
time histories of the river discharge rate and the locations of 1, 10, and 20 ppt surface salinity
along the Caloosahatchee River. As the river discharge varied, the 1 ppt location is varied
from S79 to near Shell Point. The tidal excursion of the salinity location is about 2 km for
lppt salinity and 5 km for 20 ppt salinity. The salinity locations result from a combination
of river discharge and tide and wind driven water circulation.
To test the relationship between river discharge and salinity distribution, the river
discharge at S79 was reduced 50% and increased 50% of current river discharge condition.
To remove tidal effect, 1-day averaged salinity location was compared. Figure 5.41 shows
the locations of 1-day averaged 1 ppt salinity location during the simulation period in
response to varying river discharge. The results show that 1 ppt salinity location move 3.4
km downstream and 8.5 km upstream, corresponding to a 50% reduction and a 50% increase
in river discharge, respectively.
The Caloosahatchee MFL rule (SFWMD) states that: "A MFL exceedance occurs
during a 365-day period, when (a) 30-day average salinity concentration exceeds 10 parts per
thousand at the Fort Myers salinity station or (b) a single daily averaged salinity exceeds a
concentration of 20 parts per thousand at the Fort Myers salinity station. Exceedance of
either subsection (a) or subsection (b), for two consecutive years are a violation." Figure
5.42 show current conditions of salinity distribution to compare MFL rule for
Caloosahatchee River during 2000 simulation periods. According to this result, the salinity
does not exceed the Caloosahatchee MFL rule during 2000 simulation periods except winter
dry season.
To quantify the Caloosahatchee MFL rule due to river discharge, the locations of 10
158
ppt surface salinity values were calculated during the 2000 simulation periods. Figure 5.43
shows the locations of 10 ppt surface salinity along the Caloosahatchee River due to river
discharge rate. The polynomial regression line was calculated from the relationship between
the locations of 10 ppt salinity and river discharge rate. According to this regression line, a
total river discharge of 15 m3/s at S79 produces 10 ppt salinity at the Fort Myers station.
An alternative and more acceptable approach to determine if the MFL has been
exceeded, is to use a numerical mass-balanced model in which flows from different sources
can be specified (Edwards et al., 2000). Therefore, the total river discharge at S79 required
to produce a given salinity at Caloosahatchee River can be estimated with a calibrated model.
Using the calibrated model, a diagram describing the relationship between river discharge
and the locations of two salinity values (1-day averaged 20 ppt and 30-day averaged 10 ppt)
in the river were generated in Figure 5.44. Twelve scenarios with constant river discharges
of 5, 10, 15, 20, 30, 50, 75, 100, 150, 200 m3/s at S79 were simulated for 60 days. The 60
day simulations allowed the model salinity to reach equilibrium condition for all specified
river discharges. To compare current Caloosahatchee MFL rule with simulated salinity
results, 30-day averaged 10 ppt salinity location and 1-day averaged 20 ppt salinity location
were calculated for each case. According to this diagram, a total river discharge of 18 m3/s
at S79 produces 30-day averaged 10 ppt salinity at the Fort Myers station.
To quantify the relationship between salinity at Fort Myers station and river
discharge, 1-day and 30-days averaged salinity at Fort Myers station were plotted vs. the
fresh water inflow at S79 (Figure 5.45). The result show the minimum flow to produce a
salinity of a 10 ppt at Fort Myers is about 18 m3/s, which is the same as that obtained form
Figure 5.44. The contribution of river discharge at S79 to spatial and temporal salinity
159
distribution was successfully quantified with the integrated model for Charlotte Harbor
estuarine system. This modeling approach would develop a management tool to establish
the MFL criteria, with long term salinity and river discharge data.
rShell Point
5/10/2000 01:00
-700
40000
-30000
5/10/2000 04:00
-700
-40000
-30000
100
0
-100
-200
j= -300
8" -400
Q
-500
-600
-700
5/10/2000 07:00
-30000
Fort Myers
-700
40000
-30000
-20000
Distance from S79
-20000
Distance from S79
-20000
Distance from S79
-20000
Distance from S79
BR31
-10000
-10000
-10000
-10000
S-79 Salinity
-* (PPt)
Figure 5.38 The vertical-longitudinal salinity profiles along the axis of the
Caloosahatchee River during wet season in 2000
Salinity
(PPt)
Salinity
(PPt)
28
24
20
16
12
8
4
0
Salinity
(PPt)
160
-700
40000
-30000
-20000
Distance from S79
-10000
-700
40000
-30000
-20000
Distance from S79
-10000
-40000
-30000
-20000
Distance from S79
-10000
-40000
-30000
-20000
Distance from S79
-10000
S-79 Salinity
> (PPt)
Figure 5.39 The vertical-longitudinal salinity profiles along the axis of the
Caloosahatchee River during dry season in 2000
Salinity
(PPt)
Salinity
(PPt)
Salinity
(PPt)
161
E1
ro
.n
o
to
I
150 r
100 -
150
200 250
Julian Day
300
20 ppt
Shell
"^oint
Fort
ppt Myers
BR31
S79
Figure 5.40 Time histories of river discharge at S79 and the locations of 1, 10, and 20 ppt
surface salinity along the Caloosahatchee River during 2000 simulation period.
162
0
i-
o 150
>
200
300
Location of 1 ppt surface salinity
45000 r
40000 -
£35000 r
K30000
E25000
o
^20000
0
C15000
*S
■^10000
Q
5000
0
current condition
50% increased river discharge
50% reduced river discharge
Julian Day
300
Figure 5.41 Time histories of river discharge at S79 and the 1 ppt salinity location along
Caloosahatchee River.
163
£
Q
45000
40000
35000
"'30000
25000
0)20000
o
c
*)j 15000
Q
10000
5000 -
1-day averaged surface salinity location of 20 ppt
rvn.
Shell
Point
Fort
Myers
30-day averaged surface salinity location of 1 0 ppt\
BR31
J L
X
J I I L
200
Julian Day
300
J L_ S79
Figure 5.42 The 1-day averaged 20 ppt surface salinity location and 30-day averaged 10 ppt
surface salinity location during the 2000 simulation period.
164
0 20000
O
C i /
ra §K«-\ • Minimum Flow = 15 m /s
■§15000
Q
10000H
5000
_L
Shell Point
Location of 10 ppt salinity
Polynomial regression line
-Fort Myers
•BR31
J I L
_L
J L
S79
0
50
100
150
200
River discharge (m /s)
Figure 5.43 The locations of 10 ppt surface salinity due to river discharge rate at S-79
during 2000 baseline simulation.
165
30 day averaged 1 0 ppt salinity
1 day averaged 20 ppt salinity
Shell Point
Fort Myers
■§15000
Q
10000
BR31
i i i i
I I I L
S79
50 100 150
River discharge (m /s)
200
Figure 5.44 The relationship between locations of specific salinity value vs. river
discharge at S-79.
166
♦ 1 -day averaged salinity
B 30-day averaged salinity
Minimum Flow = 18 m /s
- »---
■ | <■ ■ f ■ .|„ ■ | ,„ ■!■ - t - --t~
100
150
200
River discharge (m /s)
Figure 5.45 The relationship between salinity at Fort Myers station vs. river discharge at
S-79
CHAPTER 6
APPLICATION OF WATER QUALITY MODEL
The water quality model described in Chapter 4 was applied to the Charlotte Harbor
estuarine system. Model applications include the simulation of the three-dimensional
circulation, sediment transport and water quality processes during summer 1996, the entire
2000. These simulations provide calibration and validation of the Charlotte Harbor water
quality model using field data obtained by USGS, SWFWMD, and SFWMD. In addition,
model simulation provide assessment of (1) the effects of hydrologic alteration of causeway
and navigation channels, (2) the impact of river pollutant loading and its reduction on the
water quality in the Charlotte Harbor estuarine system, and (3) the causes for hypoxia of
bottom water in the upper Charlotte Harbor.
To reduce the computational time necessary to simulate annual response of the
Charlotte Harbor estuarine system, the parallel CH3D model developed by Sheng et al.
(2003) is used. The detail description of validation, CPU time, and speedup of the parallel
CH3D code are given in Appendix G.
6.1 Forcing Mechanism and Boundary Condition of Circulation
The tidal forcing along the Gulf of Mexico boundary is prescribed by the water level
data measured at Naples by NOS (http://www.co-ops.nos.noaa.gov/data_res.html). The
surface wind boundary condition is produced by using the hourly wind magnitude and
direction data collected at the National Data Buoy Center C-MAN stations at Venice and
167
168
USGS station at Naples and Fort Myers. The hourly wind magnitude and direction are
converted into EastAVest(-x) and North/South (-y) wind velocity components and then
interpolated onto the entire computational grid.
Daily river discharges for Peace River, Shell Creek, Myakka River, Caloosahatchee
River (S-79 spillway, Cape Coral and Whiskey Creek) and Estero Bay (Mullock & Hendry
Creek, Estero River, Spring Creek, and Imperial River) were measured by the USGS daily
(Table 5.1). Figures 6.1 to 6.3 show the water level at the tidal boundary, river flows at the
river boundary, wind speed and direction at the surface boundary, and air temperature for the
air-sea heat flux in 1996. The forcing mechanism and boundary condition for circulation
of 2000 simulation were explained at Chapter 5.
169
1 50 200
Julian Day
300 r
Caloosahatchee River
Peace River
Shell Creek
Myakka River
150
200
Julian Day
Figure 6.1 Tidal forcing and river discharges for 1996 simulations of Charlotte Harbor.
at Venice
170
_1_
-J I : i_
150 160 170 180 190 200
Julian Day
210
220
230
at Fort Myers
150 160 170 180 190 200 210 220 230
Julian Day
Figure 6.2 Wind velocity for 1996 simulations of Charlotte Harbor.
171
at Venice
150
Julian Day
200
at Fort Myers
150
200
Julian Day
Figure 6.3 Air temperature for 1996 simulations of Charlotte Harbor.
172
6.2 Initial and Boundary Condition for the Water Quality Model
For 1996 simulations, there were 21 stations sampled monthly for water quality data
in 1996 by SWFWMD and SFWMD (Table 6.1). The locations of the measurement sites are
shown in Figure 6.4. Most of the data were collected from May to July 1996. For Estero
Bay, there were 14 sites sampled for total nitrogen, total phosphorous, pH, dissolved oxygen,
temperature, turbidity and chlorophyll_a data in June 1996 by SFWMD. The initial water
column concentrations of several water quality parameters are determined from the EPA
data. The water quality data for May are used to produce the initial condition of 1996
simulations. The water quality data at each grid cell are calculated by interpolation of the
data at the three closest data stations, with a weighting function inversely proportional to the
distance from each of the three stations. Data collected at station CH001, CH029, and
CH004 are used to provide river loading data of Myakka River, Peace River, and Horse
Greek, respectively. There are three river boundary conditions at Caloosahatchee River,
which are S-79 spillway, Cape Coral, and Whiskey Creek. For these three river boundary
conditions, data at HB01, HB03, and HB04 were used. The water quality data at stations
EB002, EB004 EB01 3 and EBO 1 2 are used for four river boundary conditions of Estero Bay.
For 2000 simulation, the initial water column concentrations of several water quality
parameters were determined from SWFWMD and SFWMD data of January 2000. Figure
6.5 shows the locations of SWFWMD (CH-001 to CH-014) and SFWMD (CES01 to CES08)
water quality monitoring stations (Table 6.1). Data collected at station CH001, CH029,
CH004, CES01, CES05, and CES08 are used to provide river loading data of each river
boundary.
173
Table 6.1 Locations of water quality measured stations
station
Latitude Longitude
XUTM
YUTM
(l,J) in Grid Agency
CH-001
27 00 072
82 15 108
CH-002
26 57 216
82 12 300
CH-02B
26 58 048
82 1 1 048
CH-029
27 00 588
81 59 030
CH-004
26 56 396
82 03 324
CH-005
26 55 558
82 06 156
CH-05B
26 57 114
82 06 330
CH-006
26 54 006
82 07 090
CH-007
26 52 396
82 04 072
CH-009
26 49 132
82 05 294
CH-09B
26 53 132
82 09 288
CH-011
26 44 120
82 10 000
CH-013
26 41 300
82 18 010
CH-014
26 39 310
82 19 210
HB-001
26 41 480
81 49 277
HB-002
26 40 007
81 52 213
HB-003
26 38 090
81 54 294
HB-004
26 33 576
81 54 534
HB-005
26 30 570
81 59 024
HB-006
26 29 250
82 01 150
HB-007
26 32 264
82 07 324
CES01
26 43 199
81 41 233
CES02
26 43 354
81 42 284
CES03
26 43 001
81 45 382
CES04
26 40 541
81 50 017
CES05
26 38118
81 53 193
CES06
26 34 563
81 54 367
CES07
26 31 488
81 57 562
CES08
26 31 239
82 00 312
375674.130 2987267.910
380063.200 2982147.510
382419.400 2983448.440
402367.040 2988631 .380
394887.010 2980722.130
390352.680 2979406.960
389904.080 2981718.790
388859.690 2975881.310
393859.500 2973345.640
391542.120 2966995.710
384983.960 2974439.100
383975.740 2957800.680
370636.040 2952944.680
368386.980 2949305.790
417982.560 2953104.130
413179.170 2949843.840
409616.420 2946422.590
408897.610 2938704.790
401965.780 2933186.980
398261.910 2930385.140
387873.180 2936041.360
431400.582 2955824.308
429607.466 2956326.423
424351.809 2955279.834
417059.745 2951448.633
411552.356 2946470.516
409380.629 2940485.917
403804.420 2934742.217
399508.606 2934006.048
(38,123)
SWFWMD
(37,120)
SWFWMD
(42,119)
SWFWMD
(71,117)
SWFWMD
(60,116)
SWFWMD
(53,116)
SWFWMD
(54,118)
SWFWMD
(47,113)
SWFWMD
(57,111)
SWFWMD
(47,106)
SWFWMD
(38,113)
SWFWMD
(29,97)
SWFWMD
(9,98)
SWFWMD
(7,95)
SWFWMD
(79,58)
SFWMD
(76,58)
SFWMD
(73,58)
SFWMD
(70,55)
SFWMD
(63,50)
SFWMD
(45,43)
SFWMD
(24,70)
SFWMD
(91 ,56)
SFWMD
(89,56)
SFWMD
(84,56)
SFWMD
(78,57)
SFWMD
(74,56)
SFWMD
(71,55)
SFWMD
(65,54)
SFWMD
(58,51)
SFWMD
174
Light and temperature are the main limiting factors for the phytoplankton growth rate.
Temperature is calculated by solving the heat equation with the heat flux model as a surface
boundary condition, which was explained in chapter 3 and 4. The input data for the
temperature model include an air temperature and cloud cover rate. Continuous hourly air
temperature data collected at the NOS data collection platforms (DCP) and stored in the CO-
OPS databases are available for the Naples and Vince stations. Since there is no available
data for cloud cover, a constant cloud cover of 0.2 is used for all simulations.
To calculate PAR (Photosynthetically Active Radiance) in the light attenuation
model, the light intensity at the surface is used as the surface boundary condition. The light
intensity data (Langleys/day) are converted from the global and diffuse horizontal solar
irradiance data (W/m2) which were processed at the National Renewable Energy Laboratory
(NREL). Figure 6.6 shows the light intensity data used in the 1996 and 2000 water quality
model simulations.
A total of 215 sediment grab samples and 28 shallow cores were collected from the
Charlotte Harbor estuarine system during the period of December 27, 1964, to January 1,
1965 by Huang (1965). Based on the sediment size distribution, the entire study area is
characterized into five sediment types (Table 6.2).
Table 6.2 Sediment types for Charlotte Harbor water quality simulations
Type
D50 Range ( mm)
Category
Remark
very coarse
D50>0.50
5
coarse
0.25 > D50 > 0.50
4
medium
0.125 >D50>0.25
3
fine
0.0625 >D50> 0.125
2
silts or clay
0.0625 > D50
1
Sediments were predominantly sand, with low amounts of silt, clay, and organic
175
matter. Silt or clay deposits were only observed at the mouth of Peace River. This type of
sediment can be considered fine or cohesive sediment. Fine sediments exist in upper
Charlotte Harbor, the southern part of Pine Island Sound, and the northern part of San Carlos
Bay due to weak wave actions there. Sediment nutrient analyses were performed by the
FDEP Sediment Contaminant Survey with data from 33 sample stations from 1985 to 1989
(Schropp, 1998). These data include organic carbon, total nitrogen, and total phosphorous.
Based on water quality data, sediment data, bathymetry, and geometry, the Charlotte
Harbor study area is divided into 15 segments as shown in Figure 6.7 for the water quality
simulations.
176
o
o
o
O)
cm
C
IE
■c
o
Z o
o
o
CO
CM
OJ
CM
o
o
o
o
o
CD
CM
I
X
CH-lp'l
\-:. CH-02|
fCH-029
V4,. ^hdS iKU-.n
^ . CH-09B* •^r007
At'
%
ak •cfi-009
vj f
• ch-01 r
CH-013* (k«r€*^ J
CH-01 4 • W%f &*\ %i
HB-001 Jft^-yr-
v.fc*"
. •'HB-002
^^?or*
fl
<K:
v
1.
i i i i
_L
J I I L
J I I L
±
i i i
350000
375000 400000
Easting (m)
425000
Figure 6.4 Locations of 1996 water quality measurement stations operated by EPA
177
c
o
o
o
o
m
t-~
CM
o
o
o
m
cm
CD
CM
o
o
o
o
o
<J1
C\J
_L
X
CH
GH-02B,
fcH-029
X
„ •CH-Q06
CH-09B* ^Hrd07
WMEjjfaiJK •Cf-009
• CH-01,1
J_
CES003Ci?002
*&£ cisob-i
ES004
J_
J L
-L
J L
-L
_
350000
375000
400000
Easting (m)
425000
Figure 6.5 Locations of 2000 water quality measurement stations operated by SFWMD
and SWFWMD
178
2500 -
2000
£ 1500
c/)
C
S,
C 1000
+■>
JC
.2> 500
—
; 'Hi ,! [,
120 140 160 180 200
Julian Day
For 1996
220 240
For 2000
50 100 150 200 250 300
Julian Day
Figure 6.6 Light intensity at water surface for 1996 and 2000 simulations
350
179
\
&
\
u
"V
fe
i~\^--Y
WW
kV' 2 /-^
-*<w~"
4,.
#
K
#
45*sr
P
*ui/
,/t
X S*^*K*f k «fc
r
Figure 6.7 Segments for Charlotte Harbor estuarine system.
180
6.3 Simulations of Water Quality in 1996
CH3D-IMS are used to simulate the circulation, sediment transport, and water quality
dynamics of the Charlotte Harbor estuarine system during May 23 to August 21, 1996. To
create an appropriate initial condition throughout the computational domain, a spin-up
simulation was executed until the flow and salinity field reached a dynamic steady state. The
river discharge data for this area showed that the simulation period can be divided into a dry
season which lasts from January 1 until June 8 and a wet season which spans from June 9
until July 22 (Sheng and Park, 2002). A 30-day spin-up simulation was performed from
April 23 to May 23 (dry season) with all forcing mechanisms (tides, river discharges, wind)
to allow water level, velocity and salinity field to reach dynamic steady-state throughout the
computational domain. Using the surface elevation, velocity, and salinity at the end of the
spin-up simulation as the initial condition, the Charlotte Harbor circulation and water quality
models are then run from May 23 to August 21 with hydrodynamic, sediment and water
quality input data.
6.3.1 Calibrations
The calibration of water quality model was conducted using the monthly water
quality data obtained from the EPA STORET database. To compare the water quality results
with measured data, fourteen stations were selected: CH002 (near Myakka), CH005 (near
Peace), HB02, HB03 and HB06 (near Caloosahatchee), CH006, CH007, CH09B and CH009
in upper Charlotte Harbor, CHOI 1 in lower Charlotte Harbor, HB007 in Pine Island Sound,
HB006 in San Carlos Bay, and CHOI 3 and CH014 in the open sea.
Several initial water quality parameters for the Charlotte Harbor simulation are
obtained from "A Mechanistic Water Quality Model for the Tidal Peace and Myakka Rivers"
181
(Pribble et al, 1997) and Indian River Lagoon simulations (Sheng et al., 2001) shown in
Chapter 4.
In order to test the model response to variation of these specific parameters, the
sensitivity test was conducted for each water quality parameter by performing 90-day
simulations and comparison with the baseline simulation. These simulations used the same
initial and boundary conditions, and external forcing as those for the baseline simulation.
Therefore, variations in nutrient concentrations between the sensitivity test and the baseline
simulation can be directly related to variation of each parameter in a sensitivity test. Table
6.3 shows the parameters considered in the sensitivity analysis, baseline values in the
simulations and the parameter variations of each sensitivity test. The tests were performed
by varying each parameter within reasonable ranges according to literature survey. Table 6.4
shows the sensitivity analysis results which are presented in terms of percent RMS difference
in concentration, normalized by concentrations of each water quality species in the baseline
simulation. The SUM in Table 6.4 is linearly averaged value of RMS errors for each
sensitivity test.
The results of the tests showed the nitrogen half saturation constant (HALN) for
uptake as the most sensitive parameter in the water quality model, followed by the maximum
algae growth rate (AGRM). These two parameters are related and their major impact should
be detected in the phytoplankton biomass. However, the chlorophyll_a concentration is more
sensitive to ammonification than maximum algae growth rate or nitrogen half saturation
constant, which reveals the extension of nitrogen limitation to phytoplankton growth.
Morever, the effect of reducing the maximum growth rate (AGRM0.5) is more pronounced
in chlorophyll_a concentration than increasing this rate (AGRM2.0). Result of test
182
AGRM2.0 show that ammonia nitrogen (NH4) and nitrate+nitrite (N03) is rapidly uptaked
by phytoplankton and phytoplankton is increased due to increasing growth rate. If more
ammonia nitrogen and nitrate+nitrite are available in the system, phytoplankton will increase
more than that for this sensitivity test simulation. Related with algae growth rate and
nitrogen half saturation constant, algae respiration rate and algae mortality are also important
parameters. Therefore, the coefficients related with algae growth rate are most important
factors for overall water quality processes.
The third most important parameter revealed by the sensitivity test is an
ammonification rate. Soluble organic nitrogen (SON) is rapidly mineralized to ammonium
nitrogen. Nitrate (N03) level also increased due to nitrification. Due to high concentrations
of inorganic nitrogen, which is a food for phytoplankton growth, chlorophyll_a concentration
increased. On the opposite side, decreasing an ammonification rate promoted an increase in
SON, and decrease in NH4, N03, and chlorophyll_a.
The information obtained from the sensitivity tests enabled a more systematic and
efficient calibration of model coefficients described in Table 4.9 of Chapter 4. According
to the sensitivity test, the six most important parameters, which include maximum algae
growth rate, nitrogen half saturation constant, ammonification rate, algae mortality, algae
respiration rate, and sorption/desorption rate for SON and PON, are adjusted first for all
water quality species as part of the systematic calibration procedure. The other parameters
are adjusted as partially sensitive parameters for each specific water quality species. More
than 100 simulations were made during calibrations which include all the sensitivity
analyses. During these simulations, kinetic coefficients were adjusted within accepted
tolerances, estimated loads were reviewed and adjusted, and new processes were added or
183
modified in water quality model, if necessary. Listed in Table 6.5 are values for the
calibration parameters described in Chapter 4 and Table 4.3.
Table 6.3 Water quality parameters, baseline values, and range used in the sensitivity analysis
Test Run Parameter Literature Baseline Test run parameter
Range Parameter =Baseline parameter
*multiplier
AGRM2.0
AGRM0.5
Maximum algae
growth rate
0.2-8
2.0 - 2.2
2.0
0.5
HALN10.0
HALN0.5
Nitrogen half
saturation rate
1.5-400
25
10.0
0.5
HALP10.0
HALP0.5
Phosphorous half
saturation rate
1.0- 105
2
10.0
0.5
KAEX2.0
KAEX0.5
Algae respiration
rate
0.02 - 0.24
0.06
2.0
0.5
KAS2.0
KAS0.5
Algae mortality
0.01-0.22
0.07
2.0
0.5
WAS 10.0
WAS0.1
Algae settling
velocity
0.0 - 300
10
10.0
0.1
HALA2.0
HALA0.5
Algae half
saturation rate
200 - 2000
200
2.0
0.5
SONM2.0
SONM0.2
Ammonification
rate
0.001 - 1.0
0.015
2.0
0.2
NTTR2.0
NITR0.5
Nitrification Rate
0.004-0.11
0.08
2.0
0.5
DRON2.0
DRON0.5
Sorption/desorption
rate of SON/PON
0.02 - 0.08
0.03
2.0
0.5
DRAN2.0
DRAN0.5
Sorption/desorption
rate of PIN/NH4
0.02 - 0.08
0.03
2.0
0.5
SOPM2.0
SOPM0.2
Mineralization rate
0.001-0.6
0.02
2.0
0.2
DROP2.0
DROP0.5
Sorption/desorption
rate of SOP/POP
-
0.02
2.0
0.5
DRIP2.0
DRIP0.5
Sorption/desorption
rate of SRP/PIP
-
0.02
2.0
0.5
SODM2.0
Sediment oxygen
demand
0- 10.7
0.5 - 2.0
2.0
AKD2.0
AKD0.5
Oxidation rate
0.02 - 0.6
0.05
2.0
0.5
184
Table 6.4 Sensitivity analysis results in
quality calibration simulations
RMS d
ifference
w.r.t. b
aseline 1
or 199
6 water
ChlA
DO
NH4
NOX
TKN
P04
PHOS
TOC
SUM
BASE
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
AGRM0.5
17.63
4.29
55.20
171.6
3.31
25.53
8.47
9.51
36.9
AGRM2.0
5.68
2.54
12.68
12.72
0.97
5.00
1.91
1.81
5.42
HALN0.5
2.03
0.73
6.23
5.92
0.47
1.99
0.78
0.89
2.38
HALN10.0
18.56
4.26
65.73
204.0
3.71
37.54
10.61
10.99
44.4
HALPO.5
0.31
0.07
0.35
0.67
0.03
0.56
0.10
0.11
0.28
HALP10.0
4.02
0.77
6.07
13.27
0.45
12.03
1.89
1.76
5.03
KAEXO.5
2.13
0.92
7.17
5.24
0.54
2.58
0.87
0.92
2.55
KAEX2.0
5.43
2.08
18.59
17.27
1.35
6.87
2.33
2.55
7.06
KASSO.5
14.75
6.23
2.64
4.69
3.47
1.93
1.49
9.11
5.54
KASS2.0
19.58
5.59
7.56
13.49
3.99
2.47
2.20
9.29
8.02
WASSO.l
0.57
0.08
0.22
0.26
0.04
0.08
0.05
0.17
0.18
WASSOIO
5.51
0.81
2.43
2.94
0.47
0.84
0.50
1.67
1.90
HALAO.5
13.47
3.03
7.88
16.84
0.81
2.89
0.84
4.09
6.23
HALA2.0
13.13
3.11
2.76
4.54
1.00
1.20
0.28
3.62
3.70
SONM0.2
22.22
3.25
22.71
15.43
19.75
72.50
16.26
17.26
23.6
SONM2.0
12.47
3.66
32.01
42.19
13.49
19.84
7.96
8.96
17.5
NITRO.5
0.12
0.11
2.51
10.39
0.35
0.17
0.06
0.07
1.72
NTTR2.0
0.20
0.17
4.09
17.36
0.55
0.29
0.11
0.12
2.86
DRONO.5
3.19
0.90
6.03
6.27
10.87
6.46
2.44
3.13
4.91
DRON2.0
4.47
1.60
11.52
15.81
17.05
9.31
3.58
4.42
8.47
DRANO.5
1.46
0.40
1.74
1.65
0.71
2.94
1.12
1.31
1.42
DRAN2.0
2.32
0.68
2.96
3.01
1.12
4.58
1.74
1.99
2.30
SOPM0.2
2.24
0.29
3.42
7.15
0.29
10.14
2.96
1.07
3.44
SOPM2.0
1.68
0.21
2.68
4.21
0.23
9.15
2.68
0.91
2.72
DROPO.5
0.70
0.08
1.07
1.94
0.09
2.61
2.73
0.37
1.20
DROP2.0
0.55
0.07
0.83
1.61
0.07
2.13
2.21
0.29
0.97
DRIPO.5
1.99
0.27
3.16
6.42
0.27
8.43
3.32
0.94
3.10
DRIP2.0
2.34
0.31
3.94
5.74
0.34
14.75
5.73
1.28
4.30
SODM2.0
0.01
13.06
0.24
1.72
0.03
0.02
0.01
0.58
1.96
AKDDO.5
0.00
3.26
0.05
0.26
0.01
0.00
0.00
6.14
1.21
SUM
5.96
2.09
9.82
20.49
2.86
8.83
2.84
3.51
7.05
185
Table 6.5 The water quality model coefficients used for the Charlotte Harbor simulation
Coefficient
Description
Units
Literature
Range
Charlotte
Harbor
(6AD)T-20
temperature coefficient for
NH4 desorption
-
1.08
1.08
(6JT-20
temperature coefficient for
algae growth
-
1.01-1.2
1.08
(0A,)T-20
temperature coefficient for
ammonium instability
-
1.08
1.08
(eBOD)T-20
temperature coefficient for
CBOD oxidation
-
1.02-1.15
1.08
(6Dn)t-20
temperature coefficient for
denitrification
-
1.02-1.09
1.08
(6NN)T-20
temperature coefficient for
nitrification
-
1.02-1.08
1.04
Ood)7"20
temperature coefficient for
SON desorption
-
1.08
1.045
/o xT-20
VuONM/
temperature coefficient for
mineralization
-
1.02-1.09
1.08
re V-20
\VreSPS
temperature coefficient for
algae respiration
-
1.045
1.08
(0z)T-2O
temperature coefficient for
zooplankton growth
-
1.01-1.2
1.04
VH'a/max
algae maximum growth
rate
1/day
0.2-8.
2.0-2.1
VH'z/max
zooplankton maximum
growth rate
1/day
0.15-0.5
0.16
(NH3)air
ammonia concentration in
the air
Hg/L
0.1
0.1
achla
algal carbon-chlorophyll-a
ratio
mg C/mg
Chla
10- 112
100
anc
algal nitrogen-carbon ratio
mg N /mg C
0.05-0.43
0.16
^c
algal phosphorous-carbon
ratio
mg N /mg C
0.005-0.03
0.025
aoc
algal oxygen-carbon ratio
mg 02 /mg C
2.67
2.667
186
Coefficient
Description
Units
Literature
Range
Charlotte
Harbor
dan
desorption rate of adsorbed
ammonium nitrogen
1/day
-
0.03
don
desorption rate of adsorbed
organic nitrogen
1/day
-
0.03
dip
desorption rate of adsorbed
inorganic phosphorous
1/day
-
0.02
dop
desorption rate of adsorbed
organic phosphorous
1/day
-
0.02
d^l
molecular diffusion
coefficient for dissolved
species
cm2/s
4.E-6-1.E-5
l.E-5
Hbod
half-saturation constant for
CBOD oxidation
mgO-,
0.02-5.6
0.5
Hn
half-saturation constant for
algae uptake nitrogen
g/L
1.5-400
25
Hp
half-saturation constant for
algae uptake phosphorous
g/L
1.-105
2
Hnit
half-saturation constant for
nitrification
mg02
0.1-2.0
2.0
hv
Henry's constant
mg/L-atm
43.8
45
Is
optimum light intensity for
algal growth
|J.E /m2 /s
300-350
350
Kax
excretion rate by algae
1/day
0.02-0.24
0.06
Kas
mortality rate of algae
1/day
0.2-0.22
0.07
KAI
ammonia conversion rate
constant
1/day
0.01-0.1
0.01-0.02
KD
CBOD oxidation rate
1/day
0.02-0.6
0.05
^DN
denitrification rate constant
1/day
0.02-1.0
0.09
^NN
nitrification rate constant
1/day
0.004-0.11
0.08
Kon
rate of ammonification of
SON
1/day
0.001-1.0
0.015
187
Coefficient
Description
Units
Literature
Range
Charlotte
Harbor
Kop
rate of mineralization of
SOP
1/day
0.001-0.6
0.02
KVoi
rate constant for nitrogen
volatilization
1/day
3.5-9.0
7.
Kzs
mortality rate of
zooplankton
1/day
0.001-0.36
0.02
Kzx
excretion rate of
zooplankton
1/day
0.003-0.075
0.01
Pan
partition coefficient
between SAN and PEN
I/Jig
.5E-7-1.E-5
1E-4
Pon
partition coefficient
between SON and PON
1/M-S
1.0E-5
LE-5
Pip
partition coefficient
between SRP and PEP
i/ng
-
1E-5
Pop
partition coefficient of
SOP and POP
m
-
E-4
W^CBOD
CBOD settling velocity
cm/s
-
0.01
^*^algae
algae settling velocity
cm/s
0.0-300.
10.
Figure 6.8 contains the scattering plots for the calibration period. The location of
circles indicates the correlation between model predictions and observed data. A perfect
match between model and observed data is indicated by the diagonal line on each graph. The
circle above the line is over predicting the observation. Circles below the line indicate that
the model is under predicting the observation. Shown with each plot are the root mean
square (RMS) error and correlation coefficient (R2) mentioned before. Values shown in
parenthesis are RMS errors normalized with maximum measured data.
Overall, the model results show reasonable RMS errors for all constituents, although
the R2 values are not very high. The normalized RMS errors are less than 50%, except NH4,
P04, and TSS. Dissolved oxygen has the best agreement with measured data as shown by
188
the RMS error. Total suspended sediment concentration changed quickly with time on the
order of minutes and hours, hence comparison with monthly data did not show reasonable
RMS error. Dissolved ammonium nitrogen and soluble reactive phosphorous are strongly
affected by sorption and desorption processes, while particulate species are affected by
dynamic processes of settling and erosion of suspended sediment. The water quality model
calculates CBOD instead of total organic carbon. The simulated CBOD was compared with
measured total organic carbon because there is no measured CBOD data. Hence, the amount
of CBOD was under predicted as shown in the scatter plot, due to the difference between
CBOD and total organic carbon concentration in the system.
The correlation coefficient (R2) measures the strength of linear association between
simulated and measured data. Correlation coefficients for all species vary form 0.2 to 0.88.
189
A) Chrolophyll a
25
B) Dissolved Oxygen (DO)
RMS=7.51 (33.56%)
R2 =0.519
20
£ 15
w
1 10
C) Total Kjeldahl Nitrogen (TKN)
500 1000
measured
IU
RMS= 1.40 (17.63%) #X
R2 =0.308 ** ,X
* •<*<%/.
•* ,4r
• ui?v.
(1)
3
5
E
• -X' •
c/)
n
5 10 15 20 25
measured
IOUU
RMS=0.228(21.98
%) /
R2 =0.462
•^r
•
simulated
o o
0 o o
. . *x •
*3i» • • •
•
•
•
•
i , , , ,
measured
D) Dissolved Ammonium Nitrogen (NH ,)
250
10
RMS= 0.040 (44.48%)
R2 = 0.362
150(
100 200
measured
Figure 6.8 The scatter plots for water quality constituents during calibration period
190
E) Total Phosphorous
450
400
350
RMS= 0.0783 (31.26%)
7 R2 =0.242
F) Soluble Reactive Phosphorous (P04)
250 1
100 200 300 400
measured
G) Total Organic Carbon (CBOD)
1 ■•* » i i ■ I ■ . i i 1 i i i ■ I i ■
£U
RMS= 3.320 (32.13%) /
R2 = 0.806 »A*
15
/>•'•
■a
f •
0)
/ * •
*•>
X •
m
/ • . •
=j
10
s •
£
/ •*:.#
w
A* • \ .
5
n
/ J* *'» *
AW.
0 50 100 150 200 250
measured
H) Total Suspended Sediment (TSS)
150
RMS= 25.67 (36.45%)
R2 =0.124
"0 5 10 15
measured
Figure 6.8 Continued
20
««h<
50 100
measured
150
191
6.3.2 Results of 1996 Water Quality Simulation
The temporal variation of water quality species at each measured station is compared
with measured data in Figures 6.9 to 6.20. The thicker solid lines represent the simulated
water quality parameters near surface and dashed lines are those for bottom layer. The
measured water quality parameters are represented with rectangular symbols which represent
several vertical layers. The numbers of vertical layer of measured data are all different for
each species and each station. The measured data were plotted at all available vertical
measured location. The results show simulated and measured Chlorophyll_a, dissolved
oxygen, TKN, dissolved ammonium nitrogen, total phosphorous, and soluble reactive
phosphorous. During the simulation period, there was relatively little temporal variation in
the water quality parameters. One major exception is at the CH-005 and CH-006 station
which showed a significant vertical stratification of dissolved oxygen concentration.
The dissolved oxygen processes are important in any aquatic environment, because
living organisms depend on oxygen in one form or another to maintain their metabolic
processes. In Charlotte Harbor estuarine system, bottom water hypoxia has been reported
periodically by Environmental Quality Laboratory (EQL) since the mid-1970 (Heyl, 1996).
In the water quality model, dissolved oxygen is a function of photosynthesis and respiration
by phytoplankton organisms, sediment oxygen demand, reaeration, nitrification and
denitrification, decomposition of organic matter, tide and wind mixing, and river loading. As
shown in Figures 6.9 to 6.20, concentrations of dissolved oxygen exhibit a temporal and
vertical variation in response to variations in phytoplankton biomass and nitrogen species in
the northern part of Charlotte Harbor, while relatively little variation in the southern part of
Charlotte Harbor during the simulation period. Figure 6-21 shows the snapshots of the near
192
simulated bottom dissolved oxygen distribution in the study area on August 22, 1996 (at the
end of 90-day simulation). The result corresponds well with the measured low (< 2mg/L)
dissolved oxygen in the bottom water at the Peace River mouth. The rest of estuary does not
show low dissolved oxygen less than 2 mg/1.
Due to wind and tidal mixing and the shallow water depth, the Charlotte Harbor
estuarine system generally exhibits a vertically well mixed distribution of DO. In the upper
Charlotte Harbor, where the lowest DO in the system is usually found, stratification may
occur and a pycnocline may form if the condition is right - high river inflow from the Peace
river and low wind mixing (Sheng and Park, 2002). With this strong stratification, surface
water DO from reaeration would be blocked and high sediment oxygen demand in summer
season could create the hypoxia observed in this area. Model results showed that strong
stratification and low DO developed during the simulation period.
Phytoplankton dynamics are very important processes simulated in this study. It is
closely related to the nutrient recycling through uptake during growth, and excretion/decay
during respiration. The measured phytoplankton as algal mass per volume, was converted
to phytoplankton carbon with an algae to carbon ratio of 100. The chlorophyll_a
concentrations of Myakka and Peace River have maximum value of on July 23 (Julian Day
204) and those for the other stations have no trends during the simulation periods. The
highest daily fluctuation of chlorophyll_a is occurred in HB006 due to strong tidal
fluctuation and high chlorophyll_a concentration from the Caloosahatchee River.
In order to compare the nitrogen cycle simulations with the nitrogen species data
provided by EPA, simulated soluble organic nitrogen, dissolved ammonium and ammonia
nitrogen concentrations are combined and compared with the EPA total Kjeld nitrogen data.
193
The dissolved ammonium nitrogen data were also compared with measured data. For
phosphorous species, total phosphorous concentration and soluble reactive phosphorous are
used to compare with the EPA data. According to time series plot of each species in Figures
6.9 to 6.20, model results appear to capture the overall trend of the EPA data.
Figure 6-22 shows the snapshots of the near surface chlorophyll_a distribution in
study area for August 22 (the end of 90-day simulation). Only the Peace River mouth area
exhibits very high chlorophyll_a concentration of 3000p-g (phytoplankton carbon). Higher
phytoplankton concentrations are generally found near the river mouths and low
concentrations are found in the Gulf of Mexico.
Figures 6.23 to 6.26 show the snapshots of the near surface dissolved ammonium
nitrogen, soluble organic nitrogen, soluble reactive phosphorous, and soluble organic
phosphorous distributions in the study area at 2 pm on August 21, 1996 (at the end of 90-day
simulation). It is interesting to note that high dissolved ammonia concentrations but low
dissolved organic nitrogen concentrations are found in the area between upper Charlotte
Harbor and the Boca Grande Pass. High concentrations of particulate organic nitrogen and
adsorbed ammonium are first resuspended by the strong currents in the Boca Grande Pass
area, causing high dissolved ammonium concentration in the water column which are then
transported towards the upper Charlotte Harbor area. High concentrations of phosphorous
species are found in the upper Charlotte Harbor because Peace River drains the Hawthorn
phosphatic formations.
194
160
180
200
220
| 30
<o 20
Q.
O
o
10
160
180
200
220
2000
§ 1500
— ' 1000 -
H 500 •
160
i^w^wv»/^rf' ^^m^^^vvv-J~^~-"''^v^.'Vvwm^
180
200
220
500
g. 400
J 300
x- 200
Z 100
0
160 180 200 " 220 "~
1000
800
■a
ROO
d
Q.
400
t-
200
0
500
^^
400
"5>
-i
300
**
o
200
□.
100
160
180
200
220
160 180 200 220
Figure 6.9 Temporal water quality variations at CH002 station in 1996.
195
10 r
O
Q 2
0
*rwww\^I^
»wv \\jf 'vwq
160
180
200
220
160
180
200
220
2000
f* 1500
500
0
1000 *^^-^'
/WV\*|^
500 r
.-. 400
J 300
£• 200
z 100
0
160
180
200
220
I
uMMjU
■
a(L «
mMmaw
iMMMy^M^yvvWito
160
180
200
■ 220 "
1000
800
"9> 600
i 400
200
0
500
160
180
200
220
160 180 200 220
Figure 6.10 Temporal water quality variations at CH004 station in 1996.
196
\m
■•
180 200
y^mfki'V)0
220
160
"&>
au
-i
"■ —
■
oj
20
>.
.c
CL
o
c
o
10
JP«MA/
tfjtW/Wwwi
■
™rv«*\\A/v'w^
WJMK
mMJ}M^
flL .iA-'
.c
0
u
160
180
200
220
2000
f* 150°
-' 1000
1^^-^.^WaV
P 500
0
500
^ 400
J 300 f-
3-' 200
Z 100
160
180
200
"220"
0 m»AA*AAWjmmy*^^
200
220
160
180
200
220
500
160 180 200 220
Figure 6.1 1 Temporal water quality variations at CH005 station in 1996.
197
160
180
200
220
re
>>
.c
D.
2
o
.c
o
JU
20
■
10
0
^!«VAAflMW!©5M^
1 "
,vJkNW3v«JlwY
vaII M/ - -^^ ^ArvvJ
T\-V? **\-
^-H"— vvK\j
L^-''r
160
180
200
220
2000
1* 1500
* 1000
£ 500^
ltfiMwJW^rv^'tJ*JWM^^
500
160
160
180
200
220
1UUU
800
3
a
i-
600
400
200
0
.aft, . miw/
ICuhJU ■ ftt
w^^^j^^^A^
160
180
200
220
500 r
160 180 200 220
Figure 6.12 Temporal water quality variations at CH006 station in 1996.
198
E
O
Q
10
8
6
4
2
0
500
g. 400
1 300
X
200
z
100
0
1000
f
2
a.
800
600
400
200
0
500
1
400
300
O
a
200
100
0
T"'*
160
180
200
220
160
160 180 200 220
160
180
200
220
160 180 200 220
Figure 6.13 Temporal water quality variations at CH007 station in 1996.
199
O
a
| 30
m 20 ■
>
| 10
o
5 o
160
180
200
220
■g*^^^^^
160
180
200
220
2000
U 1500
■3" 1000
z
H 500
500
^ 400
t 300
x"» 200
Z 100
KJLS**-***
^^/V^^'-J
160
180
200
220
laaaaaftttMawwvflt^^
160
180
200
220
1000
800
■&
600
3
Q-
400
1-
200
0
500
/— s
400
-1
300
o
200
a
100
0
160
180
200
220
160 180 200 220
Figure 6.14 Temporal water quality variations at CH09B station in 1996.
200
10
«-. 8
g 6 fc^v^^^5^^^
O
Q
5
>.
a.
o
4 t
2
160
180
200
220
160
180
200
220
2000 r
ffc 1500
w 1000
500
BjVA^V,— j~*^V*WW*.
NVM^WW^f
r^JvJWV-v^v/'/*/v^W'
,^.^^Vv*WVW«^'ww<M^wn^/^'^^*^^^
160
180"
200
220
500 F
j-. 400
J 300 f
_.* 200
°"^ ~T60 180 ' 200 220
1000
800
■ft
600
Ti
a.
400
i-
200
0
500
^^
400
a
300
O
200
Q.
100
0
160
180
200
220
160 180 200 220
Figure 6.15 Temporal water quality variations at CH009 station in 1996.
201
■a
E
o
D
10
8
6 |
4
2
0
160
180
200
220
5 30
TO
~>.
SI
a.
2
o
.c
O
20
10
■a.
5
I
z
0
2000
1500
1000
500
0
500
400
300
200
100
T60- 180 200 220
***~~*<w*^*i»*>iW<t1fiM^^
160
180
200
220
™^ 1 «n ' Tori ' oAa ™ ' oon "
160
180
200
220
=5.
1000
800
600
o
a.
400
200
0
500
400
300
200
| l^-^WW^'^W
160
180
200
220
l|~~""^^^
Figure
160 180 200 220
6.15 Temporal water quality variations at CH010 station in 1996.
202
10
8
E
6
O
4
Q
2
0
I
30
s
"— ^
ro
20
>.
.n
Q.
O
10
O
Z
o
0
2000
0
1500
«. a hi
At iLllf
mi 1 1 AA a
^VvvH^
*%P
^jW^
^#%
m!0&i^d>MmN
■
160
180
200
220
160
180
200
220
i
— 1000
z
H 500
'VVVwA^^^
160
180
200
220
500 r
1
3
400
300
z
200
100
0
1000
f
Q.
H
800
600
400
200
0
500
1
2
400
300
if
O
Q.
200
100
0
/^M/I/VI/ma/mm/m^^
160
180
200
220
160
180
200
220
160 180 200
Figure 6.17 Temporal water quality variations at HB002 station in 1996
220
203
10
Q
o 4
160
| 30
n 20
>.
f 10
5 o
2000
fi 1500
& 1000
z
K 500 [■
160
kW.'.^*™^
0
500
-. 400
1 300
j? 200 ■
Z 100 •
0
160
160
1000
^ 800
"^ 600
a 400
H 200 t
500
^ 400
§ 300
Q' 200
a 100
0
160
180
200
220
180
200
220
T80~ 200
WW
220
180
200
220
180
200
220
'^W-Vw'Bs,~*^^
160 180 200 220
Figure 6.18 Temporal water quality variations at HB006 station in 1996.
204
E
O
Q
10
8
6
4
2
0
f**ZXf^
^$^*^
K/\M
ns/vj
160
180
"200"
220
0 160 ~~^80~~ ~200 220"
Figure 6.19 Temporal water quality variations at HB007 station in 1996.
205
"a
J,
o
160
180
200
220
■a
Q.
O
I—
O
r
o
30
20
10
0
2000
=• 1500
-" 1000
500
0
500
400
300
200
100
0
1000
800
600
400
200 ■
0
500
400
300
200
100
0
160
180
200
220
■a.
a.
160
■a
3
t
o
a.
160 180 200 ~ 220
Figure 6.20 Temporal water quality variations at CH013 station in 1996.
206
Near-Surface Dissolved Oxygen (mg/L)
Near-Bottom Dissolved Oxygen (mg/L)
August 21, 1996
5. \
Figure 6.21 Simulated dissolved oxygen concentration in Charlotte Harbor estuarine
system on August 21, 1996.
207
Near-Surface Chlorophyll a (ug/L)
Near-Bottom Chlorophyll a (ug/L)
Figure 6.22 Simulated Chlorophyll_a concentration in Charlotte Harbor estuarine system
on August 21, 1996.
208
Near-Surface NH4 (ug/L)
Near-Bottom NH4 (ug/L)
Figure 6.23 Simulated dissolved ammonium nitrogen concentration in Charlotte Harbor
estuarine system on August 21, 1996.
209
Near-Surface SON (ng/L)
Near-Bottom SON (ug/L)
Figure 6.24 Simulated soluble organic nitrogen concentration in Charlotte Harbor
estuarine system on August 21, 1996.
210
Near-Surface SRP (ng/L)
Near-Bottom SRP (|ig/L)
500 /
400 \
\
X \
J
500
400
300
200
100
V
August 21, 1996
Figure 6.25 Simulated soluble reactive phosphorous concentration in Charlotte
Harbor estuarine system on August 21, 1996.
211
Near-Surface SOP (ng/L)
Near-Bottom SOP (ug/L)
60
50
40
30
20
10
August 21, 1996
v *^
Figure 6.26 Simulated soluble organic phosphorous concentration in Charlotte
Harbor estuarine system on August 21, 1996.
212
6.4 Simulations of Water Quality in 2000
Model simulation of circulation, sediment transport and water quality dynamics in
Charlotte Harbor estuarine system during the summer 1996 was conducted as calibration
process tunning of model parameters and inputs followed systematic calibration procedure.
Based on the results presented, the water quality model was considered calibrated and
validated, and additional one year simulation was performed to validate the model. This
validated model was used to assess the effects of Sanibel Causeway and the navigation
channel in San Carlos Bay as well as river load reductions.
6.4.1 Validation
For validation runs, hydrodynamic and water quality model coefficients were held
fixed at the calibration values, and results of a one-year 2000 simulation were compared with
field data collected in January to December 2000. The data include phytoplankton, nitrogen,
phosphorous, dissolved oxygen at several locations in the Charlotte Harbor estuarine system.
Figure 6.27 contains calibration period scattering plots which include the root mean
square (RMS) error and correlation coefficient (R2). Values shown in parenthesis are
normalized RMS errors with maximum measured data. Without adjusting any water quality
model coefficients, the normalized RMS errors are less than 45%. Dissolved oxygen has the
best agreement with measured data as shown by the RMS error. Correlation coefficients for
all species vary form 0. 124 to 0.80. These plots allow us to assess weaknesses of the model,
and to suggest areas needing further improvement.
213
A) Chrolophyll a
25
to
RMS= 7.51 (33.56 %)
R2 =0.519
20h *
■a
$> 15-
co
E-' I- /
- iot. l-jyzs** • • •
• . vvn>s • • •
"0 5 10 15 20
measured
C) Total Kjeldahl Nitrogen (TKN)
1500
■D1000
3
E
« 500
B) Dissolved Oxygen (DO)
10
CD
£
CO
RMS= 1.40 (17.63%)
R2 = 0.308
25
0,
RMS=0.228(21.98
%) /
R2
= 0.462
mf
•
• •
•
•
• I • • 4 Sm •
• *y *
Vm
9
•;
•
•
•
"
pA-
r *
/
,,!,,,,
, , , ,
measured
D) Dissolved Ammonium Nitrogen (NH4)
250
RMS= 0.040 (44.48 %)
R2 = 0.362
500 1000 150( 0 100 200
measured measured
Figure 6.27 The scatter plots for water quality constituents in 2000.
214
E) Total Phosphorous
450
400
350
RMS= 0.0783 (31.26%)
R2 = 0.242
■o
E
RMS= 3.320 (32.13%)
R2 = 0.806
100 200 300 400
measured
G) Total Organic Carbon (CBOD)
20
5 10 15
measured
Figure 6.27 continued
20
F) Soluble Reactive Phosphorous (P04)
250
RMS= 0.0294 (42.87%)
R2 = 0.280
200-
50 100 150 200 250
measured
H) Total Suspended Sediment (TSS)
150
RMS= 25.67 (36.45 %)
R2 =0.124
50 100
measured
150
215
6.4.2 Results of 2000 Water Quality Simulation
The simulated water quality species at each measured station in 2000 as shown with
measured data in Figure 6.28 to Figure 6.39. During the simulation period, the water quality
data show the seasonal variation, and these seasonal variations are produced quite well by
the water quality model. In the water quality model, dissolved oxygen is a function of
photosynthesis and respiration by phytoplankton organisms, sediment oxygen demand,
reaeration, nitrification and denitrification, decomposition of organic matter, tide and wind
mixing, and river loading. As shown in Figures 6.28 to 6.39, simulated dissolved oxygen
concentrations show spatial and temporal variations in reasonable agreement with measured
data. The average monthly near-surface concentrations declined from 8.5 to 6.7 mg/L from
January to July and then began to rise in upper Charlotte Harbor. Near-bottom average
monthly concentrations in this area were highest in February, declined slowly through May,
and then declined more rapidly until September. The hypoxia conditions during summer are
attributed to strong stratification, which cause restricted re-aeration, and SOD. After the
breakup of the stratification, the DO concentration increased from October to December.
Algae photosynthesis and re-aeration maintain surface dissolved oxygen level, while
vertical mixing controls the transfer of dissolved oxygen to bottom water. Sediment oxygen
demand is a function of temperature which has strong seasonal variation. The high water
temperature in summer season will increase SOD in bottom sediment. If there is no
stratification, surface water dissolved oxygen will become mixed with bottom water
dissolved oxygen quickly. The Charlotte Harbor estuarine system generally exhibits a
vertically well mixed distribution of DO due to wind and tidal mixing and the shallow water
depth. The dissolved oxygen in most part of the estuary does not show any strong
216
stratification except in Peace and Caloosahatchee Rivers. In the upper Charlotte Harbor,
which usually has the lowest level of DO, some stratification may occur due to high
consumption by SOD near the bottom and super-saturation near the surface with strong river
discharge from Peace River. To quantify the causes of the bottom water hypoxia in this area,
dissolved oxygen concentration at CH006 was compared with river discharge at Peace River,
salinity, temperature and, reaeration and SOD fluxes in Figure 6.40. Although, the SOD flux
is higher than the reaeration flux in summer (Julian Day 120 to 280) due to high
temperature, the DO stratification did not occur during this period. The DO stratification
period (Julian Day 220-280) matches with the salinity stratification caused by strong river
discharge from Peace River.
To compare salinity stratification and vertical DO distribution, Simulated
longitudinal-vertical salinity and dissolved oxygen concentration along the Peace River at
1 pm on June 18 (Julian Day 170) and October 6 (Julian Day 280), 2000 were plotted in
Figures 6.41 and 6.42, respectively. The salinity at these two time period represent the effect
of salinity stratification on vertical DO distribution since the SOD and re-aeration fluxes are
similar at these periods. The results show the strong relationship between salinity and DO
stratifications. Therefore, the hypoxia in the upper Charlotte Harbor is primarily caused by
the combination effects of SOD and stratification with strong river discharge and high
temperature.
In the Caloosahatchee River upstream near the S79 (CES02), there is very strong
daily fluctuation in the surface dissolved oxygen concentration due to tidal fluctuation and
high dissolved oxygen concentration from the Caloosahatchee River. Chlorophyll_a
concentrations ranged from 1 to 100 mg/L and averaged 8.5 mg/L. Both productivity and
217
biomass were greater during summer near the mouth of tidal rivers which has middle range
salinity of 6 to 12 ppt. The chlorophyll_a concentration of Caloosahatchee River has a
maximum value of 98 mg/L on July 12 (Julian Day 193), 2000 and then quickly drop to 5
mg/L because of low chlorophyll_a concentration of river loading and fast flushing due to
strong river discharge from Caloosahatchee River. Simulated nitrogen and phosphorous
concentrations appear to capture the overall trend of the measured data collected by
SFWMD and SWFWMD.
Figures 6.43 and 6.44 show the snapshots of the near surface chlorophyll_a and the
near bottom dissolved oxygen concentration distribution in the study area on February 9,
May 9, August 7, and November 5, to represent seasonal characteristics. In February,
phytoplankton was low in the entire estuary and DO was a generally high. In May,
phytoplankton significantly increased in the Caloosahatchee River due to strong river
discharge with high nutrient concentrations. Hence, DO was reduced substantially in all
portions but was higher in the north than in the south, except in Pine Island Sound. In
August, there was significant increase in phytoplankton and decrease in dissolved oxygen
concentration in the bottom water. This low dissolved oxygen in bottom water reached the
north portion of upper Charlotte Harbor. The new water quality model, which contains
improved dissolved oxygen processes at the air/sea interface and water-sediment interface,
successfully reproduced the bottom water hypoxia in both temporal and spatial plots.
218
1
E
o
Q
I 30
3.
>»
a
o
S
o
20
10
0
1500
1000
500
^— i -*-«-^^^
^»^
•6)
200
150
100 -
50 ■
400
300
200
100
0
200
=; 150
■3 100
o
0. 50
a.
****
100
"500"
300
-^^^>-^^_
100
^00
300
[-
100
200
300
■
1 — .1^
"*"**«»
""■■ DlJlljlh
1 -y.
in iiiiiiiiiiiiiJiiiJ^rfP^
1 OKU
«*JLgI
"^ H^^»» . a. .^._ ^B
100
200
300
^^^Jj>,y*L-,
100 200 300
Figure 6.28 Temporal water quality variations at CH002 station in 2000.
219
§
o
Q
| 3°
3.
TO
>.
.C
Q.
2
o
O
100
200
20
10
0
1500
1000
ksr^*^^^^^^
* 500
0
200
i
z
150
100
50
oPVwJl
5
■B)
if
o
100
200
100
200
300
300
300
WHM'Jftl
100
200
iiiiwmwimwi
A
300
Figure
100 200 300
6.29 Temporal water quality variations at CH004 station in 2000.
220
O
Q
■a
a,
Q.
O
i-
o
i
o
1
3
30
20
10
0
1500
1000
500
0
200
150
100
I*
Z 50
0
400
O 300
3 200
Q.
*- 100
i*-ff^l — nr^J^^^*^^
100
200
300
/-T
»**"*"
100
200
"300"
"5>
2
wJLfciriffMjLt^r riv
100
200
300
•-if"**^^
100
200
300
5
o
a
Figure
100 200 300
6.30 Temporal water quality variations at CH005 station in 2000.
221
■5)
E
O
a
% 30
100
200
300
Q.
2
o
o
20
10
Piftw** »,
100
200
300
1500
"S> 1000
1 500
I-
200
yw'
,_^^|^
«U#W/»fc
m i «njl» ,w ^ i^^'^i # ■*![ in i
100
200
300
100
200
300
400 r
~ 300 [
%
3 200
Q.
*- 100
^t-^-»*-^^jl^a*^^ — .
100
200
300
200
=; 150
o)
« 100
o
Q. 50
100
200
300
Figure 6.31 Temporal water quality variations at CH006 station in 2000.
222
10
£
8
6
-^
o
4
Q
2
0
1
30
3
-w'
TO
20
><
J=
Q.
O
10
i-
O
£
o
0
1500
"S> 1000
* 500
I-
■5)
■a.
"ft
o
Q.
150
100
50
0
400
300
200
100
0
200 r
150
100
50
0
100
200
300
100
200
300
~_*f-'-"vJ
100
200
300
k^Nrt*
100
200
300
JUttAm " i -
ll^Mt^^k
100 200 30(
Figure 6.32 Temporal water quality variations at CH007 station in 2000.
223
100
200
300
%
30
3
w
TO
20
><
.C
Q.
O
10
O
.C
o
0
1500
^
"&
1000
d
,— •*
7
*
600
h-
'*>il*i«mmm*M^m>i'm^f0>^
«^*^^^
100
200
300
>„^~~»v-'"
100
200
300
200
f
2
I
2
150
100
50
0
400
1
zt
300
200
i-
100
)J^^k^mm»^^
100
200
300
100
200
300
200
^ 150
3
O
a
100
50
0
fr ^""""If**^*"^
Figure
100 200 300
6.33 Temporal water quality variations at CH008 station in 2000.
224
E
O
Q
10
8
6
4
2
0
100
200
300
i 30
• 20
§" 10
1500
"3) 1000
* 500
^jj**<lim^m»<,'i-»iii«*^
riMHP
H^.^^**--^^^
100
200
300
0
200
=; 150
* 100
_■ ■ "***J^'II>*J"> I ' |L M iii Irt.nf-J-l
100
200
"300"
I
z
50
0
400
TOO ' ' 200 ' ' ' ' 500 ~
ss- 300
3 200
Q.
*- 100 ■
0
200
g 150
3 100
o
Q. 50
100
200
300
>^JL*^ I
100 200 300
Figure 6.34 Temporal water quality variations at CH009 station in 2000.
225
f
o
Q
"ft
(0
>.
Q.
2
10
8
6
4
2
0
100
200
300
30
20
0 W ' 200 500 ^~
1500
1000
500
0
200
=• 150
•S 100
X
Z 50
0
400
300
200
100
■^J^^^^^
100
200
^00"
1 00 200 300
vtrfiyWMllW,^^^,
100
200
300
1
o
200
150
100 [
50
^y*%%»^^ ■
Figure
100 200 300
6.35 Temporal water quality variations at CH010 station in 2000.
226
o
Q
3
(0
><
Q.
2
o
.c
o
100
200
3.
2000
1500
1000
500
0
400
300
200
100
0
600
100
200
;to;*S;*s*^^
100
200
=; 150
~ 100
o
a
50 ■ _
u 100
300
300
300
200 300
Figure 6.36 Temporal water quality variations at CES02 station in 2000.
227
400 r
~. 300
■5.
3 200
Q.
*- 100
0
200
^ 150
« 100
o
Q. 50
100
200
300
4UU
a
i |
^
300
-
/P l/V ,,|
? *V
eft
5
200
A
^ l^n/1
■ "
I
*v^ ■
A IM
Jj jtitMiiiM
liiiiAlii ujAl
z
100
i'-vHi1 'MiWmma
mnm 11
iiliiiJlWW *v^_
^^f^y-
0
pff^fmPHH
mf if
iilllMi"
■ V^*
100
200
300
■
■
■
""^^^ J^^^Jj^^i''' |
■
\~s
100
200
300
jmBS^jmL
A
/^ W-v I
100 "" 200 300
Figure 6.37 Temporal water quality variations at CES03 station in 2000.
228
|
E
O
Q
§ 10°
3 80 F
a
= 60
100
200
300
•q. 40
o
o 20
5 o
" 100 200 %6o
"ft
z
2000
1500
1000
500
0
400
=; 300
^*«<€Mft
MPp
0H
l^P*»*^»w>B^^
100
200
300"
200
■a.
2
0
400
300
200
100
100 -^^^^^
0 TOO ' 2$0 300
"-*****,
WWjW i f&HRKlmmim
100
illliliWilliiMHjIWIlWM
200
gggg^ggggggg i ^£»f*>m
300
i
3
o
Q.
Figure
200
150
100 i
50
0
J-« "^
100 200 300
6.38 Temporal water quality variations at CES08 station in 2000.
229
10
s=> 8
%
E
O
Q
E 6 " **to,*J">>**^
100
200
300
% 30
3
>.
sz
a.
o
20
10
0
.^""^^frlt)^
1500
^> 1000
200
250
300
lI v^^^"^'^^^ifc»lyl^V^***w*^^^l J .nni'Mi i**mj
100
200
300
200 r
150
100
50
hy^r^V-^^mt^K^^s^^ *V*^
fti**mH**\nrt>,"«'*r'iH/(f
100
200
300
3.
200 r
150
100
50
i
o
a
0
200
150
100
H^4w»"*^^
100
200
300
50
0 ' rr
Figure
100 200 300
6.39 Temporal water quality variations at CHOI 3 station in 2000.
230
"e
(D
O
.2
Q
>
35
g 30
Q.
£ 25
W 20
15
Surface
Bottom
*¥*V
50
100
150
200
250
300
O
E
re
k_
0)
a
E
0)
•S 15
re
E
S
c
o
E
a>
re
v
O.
1>
s
c
T3
1
o
in
re
"e
i
Q
O
C/3
1 50 200
Julian Day
300
Figure 6
possible
fluxes at
.40 The comparison between simulated dissolved oxygen concentration and the
causes of hypoxia: river discharge, salinity, temperature, and re-aeration and SOD
CH006 water quality measurement station.
231
6/18/2000 13:00 (Julian Day 170)
CH006 CH005 CH004
-22500 -20000 -17500 -15000 -12500
Distance from Peace River upstream (m)
10000
>5000
CH006
CH005
CH004
-22500 -20000 -17500 -15000 -12500
Distance from Peace River upstream (m)
-10000
Figure 6.41 Simulated longitudinal-vertical salinity and dissolved oxygen concentration
along the Peace River at 1 pm on June 18 (Julian Day 170), 2000.
232
10/6/2000 13:00 (Julian Day 280)
CH005 CH004
-22500 -20000 -17500 -15000 -12500
Distance from Peace River upstream (m)
10000
25000
-22500 -20000 -17500 -15000 -12500
Distance from Peace River upstream (m)
10000
Figure 6.42 Simulated longitudinal-vertical salinity and dissolved oxygen concentration
along the Peace River at 1 pm on October 6 (Julian Day 280), 2000.
233
Chlorophyll a (p.g/L)
Chlorophyll a (ug/L)
ft • ^
Chlorophyll a (ug/L)
Chlorophyll a (ug/L)
Figure 6.43 Simulated near-surface chlorophyll_a concentration in Charlotte Harbor
estuarine system on February 9, May 9, August 7, and November 5, 2000.
234
Near-Bottom Dissolved Oxygen (mg/L)
Near-Bottom Dissolved Oxygen (mg/L)
V \ . >
$e
Near-Bottom Dissolved Oxygen (mg/L)
Near-Bottom Dissolved Oxygen (mg/L)
August 7, 2000
I
November 5, 2000
&*$■%!
J. tAa?
Bvl
J
^
Figure 6.44 Simulated near4Dottom dissolved oxygen concentration in Charlotte Harbor
estuarine system on February 9, May 9, August 7, and November 5, 2000.
v
235
6.4.3 Application of 2000 Water Quality Simulations
Hydrologic alteration
The hydrodynamic and water quality models of the Charlotte Harbor estuarine system
have been successfully developed and validated. The models can be used to address the
effect of hydrologic alteration on the water quality as shown in Chapter 5. Using the
validated water quality model, we performed several model simulations with the causeway
islands removed and with the IntraCoastal Waterway removed during April 9 to June 10,
2000. The results are compared with those under existing conditions with both of them in
place. The scenarios and stations for comparison are the same as those for comparison of
flow and salinity in Chapter 5.
The chlorophyll_a concentrations for both cases were compared with those for the
baseline simulation in Figure 6.45. The results at all three stations show that chlorophyll_a
concentration is not noticeably affected by the absence of Intracoastal Waterway (NICW
case) or the causeway (NSC case). Just like the chlorophyll_a concentration, the other water
quality species do not show much effect by these hydrologic alterations.
To quantify the effect of these hydrologic alterations on the spatial distribution of
water quality species, the snapshots of chlorophyll_a and dissolved ammonium nitrogen
concentrations for these two cases were compared with those for the baseline simulation in
Figures 6.46 and 6.47. There is not noticeable impact on the water quality species
distribution in the San Carlos Bay area, consistent with the negligible effect of hydrologic
alteration on flow and salinity in Chapter 5.
Table 6.6 shows the temporal average water quality species concentrations at the
eleven comparison stations for the baseline run. Results of the hydrologic alteration cases
236
were compared to the baseline results and RMS differences calculated, then normalized with
these averaged water quality species concentration in Table 6.6. Normalized RMS
differences of water quality species concentrations for the no causeway case from April to
June, 2000 are shown in Table 6.7. Table 6.8 shows the normalized RMS differences for the
no ICW case, while Table 6.9 shows the normalized RMS differences when both the
causeway islands and the ICW are removed. The RMS differences are less than 2% at all
selected stations for hydrologic alteration. Therefore, it can be concluded that neither the
causeway islands nor the Intracoastal Waterway had noticeable effect on the water quality
in the San Carlos Bay and Pine Island Sound area.
237
ST05
BASELINE
30
Causeway removed (NSC)
i25
Intra Coastal Waterway removed (NICW)
a.
"-" 20
CO
^ 15
£
a
,aAWW/Vu,u
g 10
^ ^^w
o
,. AAAA/W
1 ^^,,jy/W\MM/\IVVw^
■= 5
-,^a^A*W^
v V\jW™ 'V'* "
O
1
II 1
0
12C
) 140. _ 160 180
"9>
CO
2
o
30
ST08
25 -
"a.
a.
•— ' 20
CO
% 15
Q.
2 10
o
■= 5
O
0
ST10
120
120
140. _ 160
Julian Day
180
160
180
14Q. _
Julian Day
Figure 6.45 Comparisons of simulated surface chlorophyll_a concentration for three cases
at three selected stations: ST05 (Pine Island Sound), ST08 (San Carlos Bay), and ST 10
(Caloosahatchee River mouth)
238
Chlorophyll a
20
15
10
5
• S xk
Figure 6.46 Comparisons of simulated surface chlorophyll_a concentration fields in San
Carlos Bay after 90 days of simulation for three cases.
239
' m
Figure 6.47 Comparisons of simulated surface dissolved ammonium nitrogen (NH4)
concentration fields in San Carlos Bay after 90 days simulation for three cases.
240
Table 6.6 The temporally averaged water quality species concentrations for the baseline
2000 simulation.
Station number
ChlA
DO
NH4
NOX
TKN
P04
PHOST
TOC
(mg/L)
(mg/L)
(u-g/L)
(Hg/L)
(Hg/L)
(Hg/L)
(lig/L) (
mg/L)
1
6.04
6.39
99.53
31.69
728.98
9.11
103.50
1.61
2
6.51
6.39
103.06
29.60
765.36
9.62
105.38
1.68
3
7.28
6.46
101.99
27.46
763.13
9.63
110.97
1.69
4
9.14
6.50
103.79
24.03
784.51
9.69
141.30
2.07
5
7.37
6.18
115.69
26.32
864.87
10.60
90.78
2.35
6
7.59
6.33
110.88
25.46
839.52
10.39
109.21
1.97
7
7.24
6.40
107.63
26.88
809.50
10.10
108.62
1.79
8
8.57
6.48
103.77
22.69
785.02
10.35
116.44
1.92
9
11.10
6.50
103.05
20.03
787.49
9.61
145.89
2.72
10
13.71
6.44
101.44
15.66
784.27
9.35
148.24
3.85
11
10.14
6.38
100.42
16.16
771.49
11.44
117.82
2.38
AVG
8.61
6.40
104.66
24.18
789.47
9.99
118.01
2.18
Table 6.7 Normalized RMS differences of water quality species
concentrations at 1 1
stations during April to June 2000, i
showing the effect of no causeway islands.
Station number
ChlA
DO
NH4
NOX
TKN
P04
PHOST
TOC
(%)
(%)
(%)
(%)
(%)
(%)
(%)
(%)
1
0.27
0.25
0.83
1.33
1.36
0.26
1.58
0.24
2
0.37
0.28
1.17
1.16
1.91
0.29
1.29
0.27
3
0.65
0.24
0.83
1.56
1.36
0.39
1.9
0.34
4
0.85
0.24
1.04
0.58
1.77
0.21
2.25
0.31
5
0.58
0.68
1.35
1.14
2.55
0.45
2.2
0.55
6
0.47
0.43
1.15
1.19
2.02
0.33
1.95
0.3
7
0.44
0.32
1.15
1.17
1.95
0.25
1.62
0.23
8
0.5
0.27
0.94
1.59
1.46
0.4
2.22
0.41
9
0.93
0.2
0.9
0.41
1.54
0.2
1.93
0.39
10
0.76
0.18
0.74
0.29
1.23
0.15
1.44
0.36
11
0.33
0.24
1.09
1.22
1.44
0.28
1.39
0.25
AVG
0.56
0.3
1.02
1.06
1.69
0.29
1.8
0.33
241
Table 6.8 Normalized RMS differences of water quality species concentrations at 1 1
stations during April to June 2000, showing the effect of no ICW.
Station number ChlA DO NH4 NOX TKN P04 PHOST TOC
(%) (%) (%) (%) (%) (%) (%) (%)
1
0.28
0.05
0.12
0.42
0.24
0.11
0.46
0.13
2
0.33
0.08
0.23
0.65
0.38
0.11
0.33
0.19
3
0.34
0.08
0.18
0.41
0.33
0.22
0.63
0.19
4
0.97
0.12
0.37
0.75
0.41
0.24
1.89
0.37
5
0.37
0.30
0.44
1.13
0.75
0.16
0.53
0.49
6
0.52
0.19
0.39
1.16
0.59
0.16
0.53
0.35
7
0.46
0.12
0.34
0.95
0.51
0.14
0.56
0.26
S
0.50
0.17
0.24
0.38
0.47
0.37
0.84
0.30
9
1.39
0.20
0.62
0.86
0.51
0.29
2.67
0.68
10
1.51
0.41
0.74
0.82
0.68
0.51
5.31
2.11
11
0.48
0.13
0.32
0.41
0.56
0.26
1.03
0.41
AVG 0.65 0.17 0.36 0.72 0.49 0.23 1.34 0.50
Table 6.9 Normalized RMS differences of water quality species concentrations at 1 1
stations during April to June 2000, showing the effect of no causeway island and ICW
Station number ChlA DO NH4 NOX TKN P04 PHOST TOC
(%) (%) (%) (%) (%) (%) (%) (%)
1
0.40
0.27
0.95
1.69
1.36
0.26
1.30
0.30
2
0.39
0.34
1.39
1.77
2.11
0.30
1.33
0.39
3
0.72
0.28
0.96
1.89
1.25
0.43
1.46
0.40
4
0.57
0.24
1.34
1.24
1.47
0.30
1.69
0.39
5
0.40
0.93
1.77
2.18
3.23
0.54
2.58
0.97
6
0.42
0.59
1.51
2.22
2.43
0.39
2.19
0.59
7
0.40
0.42
1.44
2.01
2.23
0.26
1.75
0.42
8
0.69
0.38
1.03
1.96
1.12
0.48
2.06
0.52
9
1.07
0.23
1.42
1.15
1.17
0.35
1.80
0.74
10
1.40
0.35
1.36
0.89
0.75
0.50
4.28
2.19
11
0.47
0.30
1.26
1.58
0.86
0.37
0.75
0.32
AVG 0.63 0.39 1.31 1.69 1.63 0.38 1.93 0.66
242
River load reduction
To provide a tool for studying management options and corresponding responses of
the Charlotte Harbor estuarine system, which is one of the primary objectives of this study,
the validated integrated model was used to evaluate the effectiveness of load reductions for
improving estuarine water quality. To achieve this goal, it is necessary to use a common set
of initial conditions for the water column and the sediments so that any differences observed
between the results of different load reduction simulations would be attributable to the
differences in load reduction. To analyze the potential impact of reduced nutrient loadings
to the system, model simulations were carried out using 100% load reductions of nitrogen
species concentrations, 100% load reductions of phosphorous species concentrations, and
100% load reduction of nitrogen and phosphorous species concentrations at Peace River and
Caloosahatchee River.
One major assumption in the load reduction runs is that the SOD values are assumed
to be the same as the baseline run. This is because the fact that SOD values for the baseline
run were provided from field experiments and the model's inability to directly related the
SOD values inside estuary to nutrient loading from the rivers. Hence the results presented
in the following should be interpreted with caution, since large errors may be associated with
the above assumption. In particular, since the SOD values are unchanged, load reduction is
not expected to produce any improvement in hypoxia, hence the DO results are practically
unchanged in the model results. In reality, load reduction is expected to lead to increased
DO.
For the Peace River load reduction simulations, CH004 and CH006 stations are
selected to compare the water quality species before and after load reduction. Figures 6.48
243
and 6.49 show the water quality species at CH004 and CH006 before and after 100%
nitrogen load reduction. Due to reduction of nitrogen loading, the chlorophyll_a
concentration decreased at both stations because the reduction of dissolved ammonium
nitrogen as the limiting factor for phytoplankton. The amount of the decrease was 12 |ig/L
at the CH004 station and 4 [igfL at the CH006 station. The dissolved oxygen concentration
is little decreased at the CH004 station because the amount of dissolved oxygen from
photosynthesis is reduced due to decreasing of phytoplankton biomass, while those from re-
aeration and SOD fluxes remained the same as the baseline simulation.
Figures 6.50 and 6.51 show the water quality species at CH004 and CH006 before
and after 100% phosphorous load reduction from the Peace River. The chlorophyll_a
concentrations decreased at both stations. The phytoplankton food limiting factor is changed
to soluble reactive phosphorous after SRP concentration reached to less than phosphorous
half saturation rate. The amounts of decrease 8 [ig/L at CH004 station and 4 |ig/L at CH006
station were smaller than those for 100% nitrogen load reductions. Dissolved ammonium
nitrogen concentration was increased because phytoplankton consumed a smaller amount of
NH4 concentration, as phytoplankton food limit was decreased by SRP concentration.
For the Caloosahatchee River load reduction, CES02 and CES08 stations are selected
to compare the water quality species before and after load reduction. Figures 6.52 to 6.53
show the water quality species at CES02 and CES08 before and after 100% nitrogen load
reduction, while Figures 6.54 to 6.55 show the corresponding results for 100% phosphorous
load reduction. With reduction of nitrogen or phosphorous loading, the chlorophyll_a
concentration decreased at CES02 while there is no significant difference at CES08. The
Chlorophyll_a concentration at CES02 reached 94 |ig/L due to large amount of chlorophyll_a
244
from the river loading. Most of the water quality species are strongly affected by the loading
from Caloosahatchee River, as shown in the results at these two stations. The impacts
resulting from load reductions were confined in Caloosahatchee River and became
insignificant once outside the Caloosahatchee River.
As mentioned earlier, dissolved oxygen concentration was relatively unaffected by
the loading reduction because the SOD kinetic process used in water quality model is an
empirical formula which is a function of temperature, dissolved oxygen, and sediment type.
Sediment oxygen demand (SOD) depends on the deposition and decomposition of organic
matter on the seabed, and the exchange of nutrients and oxygen across the sediment-water
interface. To provide detailed trends in hypoxia in response to organic loads, it is necessary
to apply sediment flux model (DiToro and Fitzpatrick, 1993) with observed data for CH4,
H2S, and organic matter in sediment column and river boundary as described in Figure 4.4.
Although applying of sediment flux model provides rational predictions of sediment response
to environmental alterations, it requires additional information as compared to the use of
empirical SOD model.
Without any available data, the sediment flux model was tested for organic matter
river load reduction in Peace River using DiToro's method described in Appendix F. Figure
6.56 shows the comparison of chlorophyll_a and dissolved oxygen concentrations at CH006
station between baseline and 100% organic matter loading reduction simulations. The result
does not show much difference in dissolved oxygen concentrations between the baseline and
100% organic matter load reduction simulations, because SOD flux from methane and
nitrogen calculated by the sediment flux model is not enough to create hypoxia in CH006.
The measured total organic carbon concentration at CH006 is much lower than that at the
245
Peace River boundary (CH029). Therefore, the organic carbon concentration from Peace
River does not reach to the hypoxia area (CH006). In the summer season, the dissolved
oxygen concentration of the river load reduction scenario is even lower than that for the
baseline simulation due to reduced chlorophyll_a concentration. Chlorophyll_a
concentration from river loading in summer season is quickly reduced because of the nutrient
load reduction from the river. The photosynthesis is decreased due to chlorophyll_a
decrease, but CBOD is increased in the water column due to mortality of phytoplankton and
settling to the sediment layer. This increase in CBOD will increase the SOD flux at the
sediment-water interface.
The measured data show that total organic carbon concentration at CH006 was very
low during the simulation period. Hence, the carbonaceous oxygen demand may not be a
major factor of SOD flux in upper Charlotte Harbor. Sulfide flux could be a very important
component of SOD in anoxic estuarine water (Chapra, 1997) such as the upper Charlotte
Harbor. Hence, sulfide and iron fluxes should be included in the sediment flux model to
better represent sediment oxygen demand for river load reduction simulations. In addition,
more field data and modeling effort to determine sediment fluxes should be focused on the
specific conditions of upper Charlotte Harbor, especially specifying the parameters for the
sediment layer and the sediment- water interface.
To test the effect of reducing organic matter in river loading with current water
quality model, the analysis of the potential impact of reduced organic matter loadings to the
system was carried out by model simulations using 50% and 75% SOD load reductions at
Peace River. Figure 6.57 shows the comparison of dissolved oxygen concentrations at
CH004 and CH006 before and after 50% SOD reduction and 75% SOD reduction. The
246
dissolved oxygen concentration in the upper Charlotte Harbor is very sensitive with SOD
reduction. With 50% SOD load reduction, the surface and bottom dissolved oxygen
concentration are increased from maximum 2 mg/L to 0.2 mg/L. With 75% load reduction,
no hypoxic event was found during this simulation and the dissolved oxygen distribution
exhibit minimal vertical stratification. Although there is no hypoxia event, some localized
low-levels of near-bottom dissolved oxygen and vertical stratification were maintained even
during the 75% SOD load reduction scenario.
To quantify the relationship between SOD coefficient and hypoxia, the area which
exposed in bottom water hypoxia condition (DO is less than 2 mg/L) was calculated with 0.5
(75% reduction), 1.0 (50% reduction), and 2.0 (baseline) rate constant of SOD at 20 °C
during the simulation period (Figure 6.56). With 2.0 as the rate constant of SOD, over 60
km2 of the upper Charlotte Harbor was hypoxic during the summer season, while hypoxia
condition was not observed with 0.5 as the SOD rate constant. The maximum areal extent
of hypoxia 67 km2 was on August 21, 2000. This was also accompanied by a high degree
of stratification as indicated in Figure 6.40.
With further scientific understanding, the integrated modeling approach would enable
the development of a science-based management tool which is built on process-based
understanding rather than simple regression. Subsequent refinement of this integrated model
can be used to address ecosystem management issues such as controlling estuarine
eutrophication and determining allowable external nutrient loading levels to restore water
quality in estuarine system.
247
10
.— -
8
E
6
•^^
O
4
Q
2
°,
1
50
3
40
(D
^T
30
>.
SZ
Q.
20
O
O
10
SZ
O
U,
260
280
baseline near surface
baseline near bottom
-100 % N load reduction near surface
' 1 00 % N load redcction near bottom
VA*vvw*w*s#//w\tf*MMwWA
60
2000
200
^ 150
* 100
50
i60
180
200
220
240
260
280
260
280
180
200
ijiiiiliiijiiiljjiiiiiiiiliiiiiiiiiii
220
i
dlWilillLi '■'■
liiuilVai'i'iili'uiiiiii'ii'iVuimiii
240
260
280
300
320
300
320
300
320
300
320
ui60 180 200 220 240 260 280 300 ~ 320
Figure 6.48 The water quality species concentrations at CH004 water quality measured
station before and after 100 % nitrogen load reduction from the Peace River.
248
10
^^
8
E
6
O
4
D
2
°,
1
50
3
40
CO
—
30
>>
a.
20
o
- ***$* ' SJrS
60
180
200
'^%M^r%
220
240
'fi^
260
280
10
o
baseline near surface
baseline near bottom
100 % N load reduction near surface
1 00 % N load redcction near bottom
^VWtilii^^HWWi
1 60
180
200
220
240
260
280
200
220
240
260
280
400
i
s
300
200
Q.
100
Q
300
320
300
320
320
300
320
160
180
Tor
220
240
260
280
300
320
1
O
Q.
200
150
100
50
Q
V^.
Figure
station
160 180 200 220 240 260 280 300 320
6.49 The water quality species concentrations at CH006 water quality measured
before and after 100 % nitrogen load reduction from the Peace River.
249
5
o
Q
■"TW ^
ui60
ML . \^x^ %^jM'ku^..jf II
200 220 240 260 280
1 '•»..,
*%
50
40
m
= 30
q. 20
2
o
10
o
180
baseline near surface
baseline near bottom
100 % N load reduction near surface
1 00 % N load redcction near bottom
300
320
MwtomHttmi&NNtMNHHNlkNItlllt
160
180
200
"221T
240
260
280
300
320
320
300
320
1000
800
"j* 600
a 400 I
200
ui60
500
e. 400 r
2 300
mftmt^mt^^
180
200
220
240
260
280
300
320
O
a
200
100
q
^*^m*kmiiiML
Hm-
m^^^Mj:^lm^
iMMftto
1 60 180 200 220 240 260 280 300 320
Figure 6.50 The water quality species concentrations at CH004 water quality measured
station before and after 100 % phosphorous load reduction from the Peace River.
250
o
10
8
6
4
2
Q
'^^^^m^mL^'
ML3
W^^
60
180
200
220
240
260
280
300
320
*> 50
3 40
n
= 30
q. 20
o
S 10
5 °.6ir
2000
fi 1500
w 1000 l
z
* 500
baseline near surface
baseline near bottom
- 1 00 % N load reduction near surface
' 1 00 % N load redcction near bottom
^*Vftnt»»»»»
160
m mmmm^0^Smmmmm0S»j
ttUUtU MttHUUk,,, ,/(|i
180 200
220
240 260
280
300
320
•""'••vmutHMinmii
4 -»»»»
1000
•
800
.
1> 600
.
i
q. 400
■
H
200
180
200
<G&m0Baaai£m^kmu*
220
240
260
320
o
Q.
SUU
400
300
200
100
■
..^..^Wf^few.-.
4*
>0
180
200
220 240
260" 280 360 32
Figure 6.51 The water quality species concentrations at CH006 water quality measured
station before and after 100 % phosphorous load reduction from the Peace River.
251
100
baseline near surface
baseline near bottom
1 00 % N load reduction near surface
1 00 % N load redcction near bottom
400
^ 300
* 200
X
Z 100
• J
i
tyi|i||jjijj|tj|j '
ij|f|fUj|L, |
*'i
60
180
200
220 240
260
280
300
320
ui60 180 200 220 240 260 280 300 320
Figure 6.52 The water quality species concentrations at CES02 water quality measured
station before and after 100 % nitrogen load reduction from Caloosahatchee River.
252
■a
E
O
D
>>
SZ
D.
O
10
8
6
4
2
MWWWIflMWM
w^w *^«HWtlHw»w^^
Hmm+tw+^H
'i60
180
200
220
240
260
280
300
320
50
40
30
20
10 f-
baseline near surface
baseline near bottom
- 1 00 % N load reduction near surface
1 00 % N load redcction near bottom
1**M*w<*nv™,w/A^/llfl
WWWMlwilWHHi^
60
180
200
220
240
260
280
300
320
3
z
i-
2000
1500
1000
500
^m^mm#mtumm\w*
i60
180
200
220
240
260
280
300
320
5
a.
400
300
200
100
^^m^^^^^S^^^SSZ
w*"M**Hvw'wu
ui60
400 r
300
200
100
180
200
220
240
260
280
300
320
ikMHUuHHnttnmttm" >tf* m '* "■«*" ***mfiWtm*nM '"*%l»M»%^|/r^^
1
O
Q.
ui60
200 r
150
100
50
180
200
220
240
260
"280*
300
*320
0
MHHttttt0lto*i * i WMUMHH w
.ji^fcittii
i 60 180 200 220 240 260 280 300 320
Figure 6.53 The water quality species concentrations at CES08 water quality measured
station before and after 100 % nitrogen load reduction from Caloosahatchee River.
253
10
0 8
E
O
Q
_:
.ll/V I
llnJn '
60
180
200
220 240 260 280 300 32
baseline near surface
baseline near bottom
1 00 % N load reduction near surface
1 00 % N load redcction near bottom
1000
800 r
"^ 600
a 400
200 r
0
,«"•-'"
"* ^ **W.
1 60
180
200
220
240
260
280
300
320
500
£. 400
J 300
J 200
°- 100
A,
, jfcyTf , , fa ,
.v'.':'
sj^%i
#T'W|
W W ""%**
*~J .»— «<ft*»W»»H»Vm~-
ui60 180 200 220 240 260 280 300 320
Figure 6.54 The water quality species concentrations at CES02 water quality measured
station before and after 100 % phosphorous load reduction from Caloosahatchee River.
254
E
o
Q
=1
a
2
o
O
10
8
4
2 f
immiiimwmw ww
240
260
280
300
320
50
40 r
30
20
10
60
•5.
a.
2000
1500
1000
500
%*. —
1 60
"a.
3
400
300
200
100
Cj
400
300
200
100
60
1 60
o
Ql
200
150
100
50
baseline near surface
baseline near bottom
100 % N load reduction near surface
1 00 % N load redcction near bottom
*— iww»^».M<>(#ffWW^NW^MMiw*<»*w«tMwwMMM1 — riff**^ jt^'-^^^gjij'^i^ViIfy^***' Will I'f^lWf'^TyVl'iW'VttUi.u
180
200
220
240
260
280
300
320
^^iimmmm^''fmtmm^
1 ■ 'numwmwm
180
200
"22T
240
260
280
300
320
180
200
"22T
240
260
~2&0
300
320
HH^Min^llll^^
180
200
220
240
260
280
300
320
IWMXIMilitJl fcWWlWWMMMM
,mi, ).-v,'i,"i-,-,--^--l'-;ri,,',;"' ■"■ ■■"■->--
ui60 180 200 220 240 260 280 300 320
Figure 6.55 The water quality species concentrations at CES08 water quality measured
station before and after 100 % phosphorous load reduction from Caloosahatchee River.
255
50
100
150 200 250
Julian Day
300
10r
c
D)
I
O
8
0
50
at surface for Baseline simulation
at bottom for Baseline simulation
at surface for 1 00 % organic matter reduction
at bottom for 1 00 % organic matter reduction
100
250
-I I I !_
300
i i i i
150 200
Julian Day
Figure 6.56 Dissolved oxygen and chlorophyll_a concentrations at CH006 water quality
measured station before and after 100 % organic matter load reduction from Peace River
using DiToro's sediment flux model.
256
50 % SOD reduction at CH004 station
1
E
c
<D
U>
>•
X
O
■a
I
o
en
w
6
c
><
X
O
J
O
(/>
W
§
c
a)
5.
x
O
-o
I
o
</>
ian Day
50 % SOD reduction at CH006 station
10r
8
6
4
2
0
luflan Day 200 250 300
75 % SOD reduction at CH004 station
Surface, for Baseline
Bottom, for Baseline
Surface, for Reduction
Bottom, for Reduction
m
ian Day
75 % SOD reduction at CH006 station
irian Day
Figure 6.57 Dissolved oxygen concentrations at CH004 and CH006 water quality
measured stations before and after 50 % SOD reduction and 75% SOD reduction.
257
80 r
70
eg
E
60
-*
(0
50
<D
l_
<
40
A3
X
o
30
Q.
>
X
20
10
0
SOD = 2.0 g/m3/day
SOD = 1.0g/m3/day
- - SOD = 0.5 g/m3/day
150 200
Julian Day
Figure 6.58 The comparison of hypoxia area at Upper Charlotte Harbor according to
varying SOD constant rate at 20°C.
CHAPTER 7
CONCLUSION AND DISCUSSION
An integrated modeling system, CH3D-HMS (Sheng et al.,2002), which includes a 3-
D hydrodynamics model, a 3-D sediment transport model, and a 3-D water quality model,
has been enhanced and applied to the Charlotte Harbor estuarine system. Circulation and
water quality in the entire estuarine system have been simulated and validated with field data
from numerous sources in 1986, 1996, and 2000. To reproduce bottom water hypoxia in
upper Charlotte Harbor, models of oxygen balance and oxygen fluxes at the air-sea and
sediment-water interfaces are enhanced. To achieve a better understanding of the temporal
and spatial variation of temperature and light, and their effect on water quality processes,
the 3-D temperature model, with a heat flux model at the air-sea interface, and the physics-
based light attenuation model (Christian and Sheng, 2003) were applied. Major conclusions
of this study are summarized in the following:
1) A fine-resolution numerical grid which accurately represents the complex
geometrical and bathymetric features in Charlotte Harbor estuarine system was generated and
used to simulate the hydrodynamic, salinity and temperature characteristics in 1986 and
2000. A sensitivity study was conducted to examine the sensitivity of model results to such
factors as boundary conditions, model coefficients, bathymetry, advection scheme, and grid
resolution. Flow, salinity and temperature patterns produced by hydrodynamic model agree
well with existing data. The normalized RMS error analysis demonstrated model's ability
258
259
to simulate water level, currents, salinity and temperature within 7.5 %, 20%, 6.5 % and 7%
accuracy, respectively.
2) The calibrated circulation model was applied to assess the impact of the removal
of the Sanibel Causeway and IntraCoastal Waterway on the flow and salinity pattern in the
San Carlos Bay and Pine Island Sound and to develop minimum flow criteria for the
Caloosahatchee River. The results show that these hydrologic alterations do not appear to
show noticeable impact on the flow and salinity patterns in the San Carlos Bay and Pine
Island Sound. The salinity at the Caloosahatchee River Mouth is reduced by about 1.36 ppt
in the absence of the IntraCoastal Water way.
3) The relationship between salinity at Fort Myers and river discharge was established
by comparing 1 -day and 30-day averaged salinity values at Fort Myers and fresh water inflow
at S79 (Figure 5.35). According to this relationship, a total river discharge of 18 m3/s at S79
produces a 1-day averaged salinity of 20 ppt at Fort Myers.
4) The newly enhanced integrated model, CH3D-IMS, was used to perform 1996 and
2000 hydrodynamic, sediment transport, and water quality simulations of the Charlotte
Harbor estuarine system. The normalized RMS errors of model simulation vary generally
within 10 - 45% for all stations and all species. The agreement between simulated and
measured DO is particularly good with less than 20% error everywhere. The daily and
seasonal vertical distribution and fluctuation of DO are successfully simulated by applying
measured SOD at the upper Charlotte Harbor.
5) The upper Charlotte Harbor system has been suffering summer hypoxia in bottom
water in the past decade. In this study, CH3D-IMS was enhanced and calibrated to analyze
this phenomenon. The calibrated model was used to examine the dynamics of various factors
260
that can affect hypoxia, including freshwater inflow, tides, SOD, and water column oxygen
consumption. The model results suggest that hypoxia in the upper Charlotte Harbor is
primarily caused by the combined effects of SOD and vertical salinity stratification, and also
enhanced by water column oxygen consumption.
6) The integrated model was applied to assess the effects of hydrologic alterations
and to provide a preliminary evaluation of pollutant load reduction goal (PLRG). The results
show that these hydrologic alterations do not appear to show noticeable impact on the water
quality patterns in the San Carlos Bay and Pine Island Sound. The dissolved oxygen
concentration of the estuarine system does not response with different load reduction because
of the use of the same SOD values for all load reduction simulation. In reality, SOD is
expected to be reduced when nutrient loading is reduced. However, due to lack of detailed
data of the sediment diagenesis and lack of understanding of the relationship between SOD
and nutrient loading, the model cannot simulate the response of SOD to load reduction.
Further research is needed to develop the ability to simulate the effect of nutrient load
reduction.
7) A systematic calibration procedure has been developed for a more efficient and
more objective calibration of the water quality model. A consistent framework for
systematic calibration is formulated which include following step: model initialization,
sensitivity analysis, parameterization, and formulation of calibration criteria.
8) Although the integrated model developed and applied in this study have performed
well simulating hydrodynamic and water quality components in the Charlotte Harbor
estuarine system, the model contains various uncertainties, assumptions and simplifications
which need further investigation and improvement. Several possible improvements in the
261
developed integrated model and year-long simulation of hydrodynamic and water quality
dynamics in the system are:
a) A finer grid system is needed to represent the navigation channel, causeway, small
islands, and complex shoreline. With grid resolution used in this study, the model cannot
solve the dynamics of the shallow regions less than 1.5 m in depth of the system, where wave
interactions become important.
b) Sediment oxygen demand (SOD) depends on the deposition and decomposition
of organic matter on the seabed, and the exchange of nutrients and oxygen across the
sediment-water interface. In order to predict trends in hypoxia in response to organic loads,
it is necessary to include oxidation-reduction processes with CH4, H2S, and organic matter
(DiToroetal. 1990).
c) More accurate river discharge data are needed. The river discharge data at Peace
and Myakka rivers are underestimated since not all freshwater inflows are accounted for by
the data. To solve this problem, the integrated model need to be coupled with watershed
model to supply missing freshwater discharge
d) Phytoplankton dynamics are different with different phytoplankton species. To
better understand phytoplankton process and eutrophication characteristics, gathering more
phytoplankton species data and applying models of multiple phytoplankton species are
needed
e) Submarine groundwater is an important mechanism in delivering chemical
elements to surface waters, and can be of the same order of magnitude as surface water
sources (Oliveria et al, 2003). Rutkowski et al. (1999) suggest that the nutrient flux by
submarine groundwater is on the same order of magnitude as a small river. To address this
262
issue, a nutrient advection by coupling with groundwater model with the integrated model.
f) Spatially more detailed and temporally more frequent hydrodynamic and water
quality data inside the Charlotte Harbor and at the river mouths are needed. Data of nutrients
and organic material in the sediment column should be gathered to enable simulation of
hypoxia in the upper Charlotte Harbor.
APPENDIX A
FLOWCHARTS FOR CH3D-IMS
The following flow charts illustrate the order of operations performed in the
integrated modeling for Charlotte Harbor estuarine system. Figures A. 1 illustrate the flow
chart of the main CH3D program (ch3d.f). Main program, ch3d.f, call the subroutine
ch3dm2.f which is the driving subroutine for time stepping of the solution as well as defining
the time varying forcing functions and generating output shown in Figure A.2. Water quality
model and sediment transport model are coupled in this subroutine. Figure A.3 and A.4
show the initializing sediment parameters and main routine of sediment transport model,
respectively. Water quality model was illustrated in Figure A.5 through A.7.
263
Chart 1: main program (ch3d.f)
264
Set screen output destination
(ch3dfileopen.f)
Output the version number of the fortran/script files
(ch3d_versionsf)
Read optional command line arguments
(ch3dclo.f)
Read grid & bathymetry
(ch3dgdep f)
Read hydrodynamic input file
(ch3dir_hyd.f)
Read salinity input file
(ch3dir_salf)
Curvilinear coordinate transformation constants
(ch3dtr.f)
Modify bathymetry (hmin and hadd)
(ch3dih.f)
Nondimensionalize and set constants
(ch3dnd.f)
Set cell boundary direction flag and sweeping indices
(ch3dii.f)
Read temperature input file
(ch3dir_tem.f)
Figure A.l Flow chart for the main CH3D program.
265
Initialize arrays
(ch3dif.f)
Compute constants
(ch3div.f)
Set wind stress
(ch3dws.f)
Is flushing model included?
(_FLUSHING_MODEL_)
F *
^ Is tidal forcing ~~^_
<C_ read from file?
-^(ITIDE < 0) ^ "
--^ T
Read tidal forcing from file
(ch3dtd.f)
F
1
r
Details in Chart 2
Main time iteration loop
(ch3dm2.f)
f END J
Figure A.l continue.
Read flushing input file
(ch3dirjlu.f)
Set initial concentration
field (ch3dics.f)
266
Chart 2: main time-iteration loop (ch3dm2.f)
Called from ch3d.f : Chart 1
-lag for initiaT
surface elevation
XISURF = 3L
Interpolate the initiatial
surface elevation
(ch3dinitsurf.f)
Set depth array
(ch3ddp.f)
Initializing sediment model
Chart 3-1
Initialize sediment
model
(ch3dsedi.f)
Initializing water quality model:
Chart 4-1
initialize
water quality model
(ch3dwaini.f)
Set temperature
at river boundary
(ch3dtk.f)
Initialize river
arrays(ch3dri.f)
Calculate 3-D velocity
at river boundaries
(ch3dri_3d.f)
Is precipiation
^/evaporation included?^
(IRAIN * 0)
Initialize rainfall
/evaporation arrays
(ch3drain.f)
Figure A.2 Flow chart for the driving subroutine for the time stepping of the solution.
267
Initialize heavily used
transport coefficients
(ch3dcoef.f)
Start of time loop
Time varying inflow
river discharge routine
(ch3drv.f)
Calculate water
density&arroclinic
pressure in
sigma plane
(ch3dde_i.f)
(ch3ddej.f)
ISIGZ=1
CSD
Calculate water
density&arroclinic
pressure in
z plane
(ch3dde_z_i.f)
(ch3dde_z_j.f)
1s vertical eddy-
coefficient varying?
(IEXP*01
Computer variable
turbulent eddy coefficients
(ch3ded.f)
Figure A. 2 continue.
268
Update internal and external variables
(ch3dreset.f)
Set wind field
(ch3ddwt.f)
T
oes rainfall vary witff-^
Update the time
varying rain (ch3drt.f)
time?(IRAIN > 0 \^
^F
Is real tide being
used?(ITIDE<0)
Read tide data and
interpolate (ch3dti.f)
Ts harmonic tide beinc]
used? (ITIDE>0L
Update old depth
Tidal forcing calculated
from harmonic input file
(ch3dtd.f)
Compute Ul"*1 and S* for2-D model
(ch2dxy_i.f)
Compute Vl~' and S"-1 for 2-D model
(ch2dxy_j.f)
Update new depth arrays
(ch3ddp.f)
Figure A. 2 continue.
269
Update velocities
at river boundaries
(Ch3dri_3d.f)
Compute velocities in x/y/z directions
(ch3dxyz_i.f)
(ch3dxyzj.f)
(ch3dxyz_k.f)
Compute inertia&diffusion
for 3-D model in x/y directions
(ch3ddi_i.f)
(ch3ddij.f)
Compute inertia&diffusion
for 2-D model in x/y directions
(ch2ddi_i.f)
(ch2ddi_j.f)
Is salinity
being simulated?
(IT>ISALT &
ISALT#0)
Salinity transport
(ch3dsa.f)
-Is temperature^
being simulated?
(IT>ITEMP &
ITEMP*0)
Temperature
transport (ch3dte.f)
Suspended sediment
transport (ch3dsed.f)
Sediment model:
Chart 3-2
Figure A. 2 continue.
(2D_2)
270
Water quality model
Chart 4-2
Water quality transport
and chemical reactions
(ch3dwq.f)
Conservative
species transport
(ch3dcs.f)
Conservative
species transport
for 2-D model
(ch3dcs.f)
Check conservation
(ch3dcheckcons.f)
Output Hydrodynamics&salinity
model results(ch3dot.f)
Output sediment&wave
model results
(ch3dot_sed.f)
Run "poor mans RPC
via tecplot (ch3dpmrpc.O
End of time loop
Return to chart
Figure A. 2 continue.
271
Chart 3-1: Sediment initialization and fetch (ch3dsedi.f)
Called from ch3dm2.f: Chart 2
Read sediment input file and wave
induced bottom shear stress
(ch3dirs.f)
Initialization
(ch3disd.f)
Compute fetch
(ch3dft.f)
.Return to ch3dm2.f: Chart 2)
Compute wave induced bottom
shear stress
(ch3dbs.f)
Figure A.3 Flow chart for the initializing sediment transport model
272
Chart 3-2: Sediment model (ch3dsed.f)
Called from ch3dm2.f: Chart 2
T
Compute wave induced bottom
shear stress
(ch3dbs.f)
Calculate settling, deposition velocities and erosion rate for
fine and coarse sediment
Fine group advection, diffusion
(ch3dsd1 .f)
Coarse group advection, Diffusion
(ch3dsd2.f)
Calculate sediment bottom change
Return to ch3dm2.f: Chart 2)
Figure A.4 Flow chart for main sediment transport model
273
Chart 4-1 : Initializing Water Quality Model (ch3dwqini.f)
Called from ch3dm2.f :chart 2
Initialize WQ variables
Read water quality input file
(ch3dir_nut.f)
Read dissolved oxygen data for several
stations (fort. 310) and time and space
interpolate (ch3drdoxy.f)
F •*
Read incident light data (fort.307)
and time interpolate (ch3drlight.f)
Read temperature data
for each segment(fort.306)
and time interpolate
(ch3drtemp.f)
Read algae growth rate for each segment
(fort.309) and time interpolate (ch3dragrm.f)
Read color data for each segment
(fort.308) and time interpolate (ch3drcolor.f)
Figure A.5 Flow chart for initializing water quality model
274
Sgad initial WQ at aerobic sediment layer.
(INIT_WATER t 0)
Read light information:Table for absorption rate
according to wave length from fort.321
Read all initial WQ parameters
at computational grid points (fort.99)
Read initial WQ for water column(fort.303)
and spatial interpolate (ch3dinterp.f)
Read initial WQ for aerobic sediment column
(fort.301) and spatial interpolate (ch3dinterp.f)
Read initial WQ for anaerobic sediment layer
(fort.302) and spatial interpolate (ch3dinterp.f)
Set initial WQ area for conservation check
and initialize assimilation terms
Read river boundary data from fort.304 or fort.305
time interpolate (ch3drrivwq.f)
F K
Return to ch3dm2.f : chart 2
Figure A.5 Continue.
275
Chart 4-2 : Water Quality Model (ch3dwq.f)
Called from ch3dm2.f :chart 2
i
Set up assimilation terms
and Reset WQ variables
Read river boundary WQ data (fort.305)
and time interpolate (ch3drrivwq.f)
Calculate horizontal & vertical advection, horizontal diffusion
in water column and specify tidal and river boundary (ch3dtrp.f)
Calculate solar radiation angle
(solar_angle.f)
Compute incident light
(ch3dlig.f)
Read incident light data (fort.307)
and time interpolate (ch3drlight.f)
Read color data for each segment
(fort.308) and time interpolate (ch3drcolor.f)
Read dissolved oxygen data for several
stations (fort.310) and time and space
interpolate (ch3drdoxy.f)
Read temperature datafor each segment
(fort.306)and time interpolate(ch3drtemp.f)
Read algae growth rate for each segment
(fort.309) and time interpolate (ch3dragrm.f)
Figure A.6 Flow chart for main water quality model.
276
Setup Z- vertical grid
include water and sediment column
(gridld.f)
Compute temperature
and light functions
(temlit.f)
Temperature and light coefficient
: Chart 5
Compute vertical diffusion and chemical reaction
for plankton species in water column
(ch3dalg.f)
Prepare variables and convert to z-direction
1 dimensional variables in water and sediment column
for WQ simulations (preset.f)
Compute vertical diffusion and chemical reaction
for nitrogen species in water and sediment column
(ch3dnitro.f)
Compute vertical diffusion and chemical reaction
for phosphorous species in water and sediment column
(ch3dphosp.f)
Compute DO and CBOD
(ch3doxy.f)
Post processing of WQ variables
(postset.f)
Check conservation fro WQ species
(ch3cons_wq.f)
Output water quality model results
(ch3ot_nut.f)
Return to ch3dm2.f : chart 2
Figure A. 6 continue.
Chart 5: Sets up temperature and light functions for water quality model
and finds light intensity in a particular water column (temlit.f)
277
Set up temperature functions
For water quality
Set coefficient for algal growth
based on light intensity
Return to ch3dwq.f: Chart 4-1
Figure A.7 Flow chart for computing the temperature and light attenuation functions for
water quality model.
APPENDIX B
DIMENSIONLESS EQUATIONS IN CURVILINEAR BOUNDARY-FITTED AND
SIGMA GRID
Non-dimensionalization of governing equations make it easy to compare the relative
importance of various terms in the equations. The governing equations are non-
dimensionalized using the following reference scales: Xr and Zr are the reference lengths
in the vertical and horizontal directions; [/, is the reference velocity; p , pr and
Ap = p- p0 are the reference density, mean density and density gradient in a stratified
flow; AHr and AVr are the reference eddy viscosities in the horizontal and vertical
directions; Dw and DVr are the reference eddy diffusivities in the horizontal and vertical
directions. The dimensionless variables can be written as (Sheng, 1983)
(x\y\z*) = (x,y,zXr/Zr)/X,
(u\v\w*) = (u,v,wXr/Zr)/Ur
m =coXrIUr
t*=tf (B.l)
(f>;) = (r;,<)/(Poizr[/r) = (r;>;j/rr
C = gCi(furxr)=cisr
A\r = A/ ' AVr
D'H=DH/ DHr
D*v=Dv/DVr
278
279
These dimensionless variables can be combined to yield the following dimensionless
parameters
Rossby number : R =
Froude Number : F =
° 4^
u.
r M
F
Densimetric Froude Number : J7 = !_ /g ?^
Vertical and Horizontal Ekman Number : Ev = —^- , EH = —&-
ft ft
A A
Vertical and Horizontal Schmidt Number : Srv =—^-,Sru = -Jt-
Dvr DHr
(3 = ^
A,
2 v2~ —
KFrJ
In three-dimensional modeling, complex bottom topographies can be better
represented with the application of o-stretching (Sheng, 1983). This transformation allows
the same vertical resolution in the shallow coastal areas and the deeper navigation channels.
The vertical coordinate, z, is transformed into a new coordinate, o, by (Phillips, 1957).
^_ z-£(x,y,t)
h(x,y) + C(x,y,t) (B3)
where h is the water depth and Q is the surface elevation.
In this new vertical coordinate, the vertical velocity is calculated by the following
equation.
280
DC
w = H co + (\ + a)—?- + a
Dt
dh dh
u — + v —
dx dy
)
(B.4)
u dz da . ,
where w = — in the z-plane, CO = in the o-plane
dt dt
Using non-orthogonal boundary-fitted horizontal grid, it is possible to better represent
the circulation and transport processes in estuarine systems with complex shoreline
geometries. Using the elliptic grid generation technique developed by Thompson (1982) and
Thompson et al. (1985), a non-orthogonal boundary-fitted grid can be generated in the
horizontal dimension. To solve for flow in a boundary-fitted grid, it is necessary to
transform the governing equations from original coordinates (x, y) to the transformed
coordinates ($, r\). The spatial coordinate system in the computational plane (£, r|) is
dimensionless while the coordinates in the physical plane (x, y) have dimensions of length.
During the transformations, the velocities are transformed into contra-variant velocities.
In the boundary-fitted, curvilinear, o-stretched, non-dimensional coordinate system,
the continuity and momentum equations are
dt
^(^»+4(^»
do
(B.5)
281
1 dHu
H~dT
(
s d£ * drj
S~ , 8
rll + , V
^V 60
^
■Vn
_a_
a
( y{ 4^HUU + y„ \[8~0HUV) + T- ( Vf V^# "V + ^ yfg^Hw}
Ro
-So"
«£(
a//i
(B.6)
wv
ao-
ap + 12a^
a<f a;/.
Jcr +
^fr^flfr^H
+
— 7- — A^ — +EHAH (Horizontal Diffusion of u)
H da\ da j
i a//v
ff ar
d# * a^
+
.21
\
U + , V
"n/So
+AL
So#
a_
a_
dHvw
So
f:
«t
da +
+
eu a
H2 do-
da j
21 ^P 22 ^/°
dv\
\- — +EHAH (Horizontal Diffusion of v)
da )
2ldH ,2dH
[l#°
+ crp
(B.7)
is the determinant of the matric tensor, gu , which is defined as
(B.8)
§"-
x: +
i + ye
xsx, + y*yn
_xnx{ + y,yz xv + y,
Sn 812
621 <?22
(B.9)
whose inverse is
282
g'J =
4 + yi -(x^n + y{yn)
-<XnXe+yvye) x]+y]
11 12
8 8
8 8
(B.10)
As shown in Sheng (1986), the contravarient components («') and physical
components (u(I)) of the velocity vector in the non-Cartesian system are locally parallel or
orthogonal to the grid lines, while the covarient components («,) are generally not parallel
or orthogonal to the local grid lines. The relationship between the physical velocity and
contravarient velocity is given by (Sheng, 1986)
u(i)
with summation on i.
The salinity and temperature transport equations can be written as
dHS
dt
HSCV da
3 fDv^
vda
K
dHwS
da
&
VioW
±(4J-0HuS) + ±(^0HvS)
+
+
Sch\I8o
d ( dS
2 3S"
)
d ( l—rr 2. dS f— „ 22 dS ^
(B.ll)
(B.12)
and
283
BHT
dt
HSCV da
3 (dX
v da
K
H
+
go
dHwT
da
ToHuT) + — (JT0HvT)
*Z
Scny[g~oldZ
drj
8oHgn—+Jg0Hg12 —
d£ drj
scH^ldrl
i — _. 21 dT / — „dT^
d£ drj
(B.13)
The sediment transport equation can be written as
dHc:
dt H ■ Scv da
H
I (D V
V
-/?,
y
d(Hco-wJci
da
+
Ev
ScHyJgo
ScH\jgo
f
K
dc
goHg"-j: + JgoHgi--±
V
d f
^
, 3c,
3?7
drj
„dc"
\
g0Hg2^ + Jg~0Hg
(B.14)
where c, represents cohesive (i=l) and non-cohesive (i=2) sediment concentrations and wVi
is settling velocity for sediment group i.
APPENDIX C
COMPARISON OF WATER QUALITY MODELS-WASP, CE-QUAL-ICM, and
CH3D-WQ3D
The CH3D water quality model (CH3D-WQ3D) was compared with existing water
quality models, specially Water Quality Analysis ans Simulation Program (WASP), the
integrated Compartment water quality model developed by the US Army Corps (CE-QUAL-
ICM). The methods of coupling with hydrodynamic and sediment transport models, the
simulated parameters, the assumptions, the chemical/biological processes, and the limitations
of each model are discussed and compared in Table C.l. The first criterion in model
comparison is its ability to simulated hydrodynamic, sediment transport, and water quality
with the efficient coupling.
The WASP box modeling framework has proven to be an excellent water quality
model for riverine systems, where the steady state assumption is applicable. In addition, this
simple box model can be successfully used to perform numerical experiments like sensitivity
test. However, in marine environments, it should not be used without proper linkage with
3-D hydrodynamics and sediment models, because tide wind and baroclinic forcing underact
in an unsteady balance.
As pointed out by Chen and Sheng (1994), the loosely coupled models such as CE-
QUAL-ICM, cannot account for nutrient release by sediments in episodic events. Because
these models were not coupled with a dynamic model for sediment transport, they could not
284
285
accurately consider sediment-process effects such as resuspension, deposition, flocculation,
and settling on nutrient dynamics in estuaries. Furthermore, this model use equilibrium
partition with function of biomass for hydrolysis process. This may be a reasonable
assumption when the time step of the water quality simulation is large compared to the time
it takes to reach equilibrium. However, sometimes a water quality model may use a small
time step during which the absorbed and the dissolved nutrient may not achieve equilibrium.
In this case, a kinetics model, such as sorption/desorption kinetic process in CH3D-WQ3D,
is needed. Analytical chemists repeatedly found that complete recovery of contaminants
from soil/sediments frequently requires lengthy extraction periods, abrasive mixing, and
strong solvents (Witkowski and Jaffe, 1978). These observations are in contradiction with
the equilibrium models which assume that sorption-desorption reactions are accomplished
instantaneously.
The light attenuation model need chlorophyll a and suspended sediment
concentrations to calculate light attenuation in water column. The water quality model use
light intensity for limitation of algae growth rate and photosynthesis processes. Therefore,
without fully coupling these models, it could not be accomplished to communicate each
others with both direction.
The chemical/biological processes in all three models are basically similar and
compatible with each others. Most of them are the adjustable coefficients. Thus, the ability
to simulate water quality depends on modeler experience, and necessity of model complexity.
Also the data availability is the key to ensure reliability and accuracy of the results. The CE-
QUAL-ICM include much more simulating species than those for the other model. Without
measured data for those species, these complex nutrient cycles could create more uncertainty
286
than simple nutrient species model. CE-QUAL-ICM use sediment flux model developed by
Ditoro and Fitzpatric (1990) for sediment oxygen demand. Although, applying of sediment
flux model provides rational predictions of sediment response to environmental alterations,
it increases information requirements and computation time compare to employment of user-
specified fluxes (Cerco and Cole, 1995). They point out that employment of user-specified
fluxes show the better calibration of the water quality model than employment of the
sediment flux model.
Overall, with the similarity and compatibility of chemical/biological processes which
is depend on data availability, coupling with hydrodynamic model, sediment transport model,
temperature model, and light attenuation model is the key to ensure reliability and accuracy
of model. Specially, the Charlotte Harbor estuarine system which has a strong linkage
between hydrodynamics and water quality dynamics suggests the importance of coupling
with hydrodynamics and sediment transport. Therefore, CH3D-WQ3D is better to simulate
water quality in Charlotte Harbor estuarine system than the other water quality models.
287
Table C.l Comparison of water quality models
model
WASP5.X
CE_QUAL_ICMvl.O
CH3D-WQ3D
Hydrodynamic
model
Read into the
model as input
parameters
read into the model
as input parameters
fully coupled with
circulation, salinity
and temperature
transport model
Sediment
transport model
Read into the
model as input
parameters
loosely coupled with
simple model
fully coupled with
sediment dynamic
model
Light attenuation
model
loosely coupled
with empirical
formula
loosely coupled with
empirical formula
fully coupled with
physics-based light
attenuation model
Nitrogen cycle
NH3, N03, ON
NH4, N03, SON,
LPON, RPON
NH4, N03, SON,
PON, PIN, NH3
phosphorous
cycle
SRP, OP
SRP, SOP, LPOP,
RPOP
SRP, SOP, POP,
PIP
carbon cycle
Using CBOD
SOC, LPOC, RPOC
Using CBOD
silica cycle
No
Available Silica,
Particulate Biogenic
silica
No
metal cycle
No
Iron & Manganese
No
phytoplankton
One species
Three species
One species
zooplankton
Using zooplankton
grazing rate
Using phytoplankton
predation rate
Yes
Metabolism
Respiration and
non-predator
mortality
One parameter
Respiration and
non-predator
mortality
re-aeration
Function of current
velocity and depth
Constant
Function of current
velocity, depth, and
wind speed
sediment oxygen
demand
Empirical formula
Sediment flux model
Empirical formula
Variation of
Ammonification
Function of
temperature
Function of algae
biomass
Function of
temperature
Variation of
mineralization
Function of
temperature
Function of algae
biomass
Function of
temperature
288
Hydrolysis
No particulate
Equlibrium partition
Sorption/desorption
species
with function of
algae biomass
kinetic
phosphorous
Constant
Variable ratio
Constant
/carbon ratio
described by
empirical
approximation
APPENDIX D
NUMERICAL SOLUTION TECHNIQUE FOR WATER QUALITY PROCESSES
In the finite difference solution of the water quality model, the advection and
horizontal diffusion terms are treated explicitly, whereas the vertical diffusion and
biogeochemical transformations are treated implicitly. Fractional step methods, which
guarantee numerical stability and prevent negative concentrations, are applied in the
numerical solution (Chen and Sheng, 1994). The horizontal diffusion and horizontal and
vertical advection terms are solved first. The numerical solutions proceeds with the
calculation of vertical diffusion, and then biogeochemical transformation reactions. Finally,
the sorption/desorption reaction terms are solved. Equation (D.l) shows a schematic of the
numerical solution algorithm method used in this study.
At
Nn2-Nn{
- [Horiz. Advection + Vertical Advection + Horiz. Diffusion]"
■ = [Vertical Diffusion]"2 + [Q]"2 (D.l)
At
= [Sorption]" + [desorption]
Nn+l-N"2 r„ . in+1 r, . in+1
At
By solving the sorption/desorption terms separately from other terms, it is possible
to treat these terms implicitly. To illustrate this, the difference equations for the
sorption/desorption reactions are examined:
289
290
d ^ d =HdxN?-dx-Px.c-N?1)
N>Hl _ Nn2 (D.2)
' ^ ' =<dxN?-dx.px-c-N?x)
where Nd and Np are concentrations of dissolved and particulate nutrients such
as,NH4 and PIN, SON and PON, SRP and PIP, and SOP and POP; dx is desorption rate for
a specific nutrient species; andp^ is partition coefficient for a specific nutrient species.
In solving the second step of the fractional step method for organic/inorganic
phosphorous and nitrogen species, the difference equations for the entire water column and
sediment column are solved simultaneously. The kinetic processes for particulate nutrient
and plankton species need to include a settling process, which accounts for the limited
vertical motion of these species.
For dissolved species
For particulate species
dN
d _
dt dz
D.
dN
\
<i
dz
+ Q (D.3)
dND a
f
dz J
+ Q (DA)
dt dz
The exchanges of particulate nutrients at the water-sediment interface are determined
by the sediment resuspension and deposition fluxes while the exchange of dissolved nutrient
species between the water and sediment column are automatically included in the numerical
solutions, with diffusion term. In order to ensure that the water quality model is consistent
with the sediment model, in terms of bottom exchanges, the erosion and deposition rates
calculated in the sediment model can be imported into the governing equations for
particulate nutrients. Therefore, the particulate nutrient concentrations are solved with the
unit of percentage (jig/jig). Let p be any particulate species (Np) mass per unit mass of
sediments, then,
N=p-c
291
(D.5)
where c is the suspended sediment concentration. Substitute Np with pc into second step of
fractional step method:
dpc _ d
dt dz
wspc + Dv
dpc
~dz~
+ Q
dc dp d
p — + c— = —
dt dt dz
wc + DV —
vdz
dz
v dz
+ Q
(D.6)
(D.7)
Since the vertical one-dimensional sediment equations are:
dc d ( dc^
— = — wtc + Dv —
dt dzy ' dz
(D.8)
Equation (D.7) becomes:
dt
wc + D —
'dz
dz dz
dA
v dz
+ Q
(D.9)
This equation allows us to import the erosion and deposition rates calculated in the
sediment model into the particulate nutrient species. Since the boundary condition for the
sediment model at the water-sediment interface is:
wsc + Dv-^- = D-E
dz
Equation (D.9) at the water-sediment interface becomes:
(D.10)
dt dz dz
dz
+ Q
(DM)
where D and E are deposition and erosion rate, which are calculated at sediment model.
Due to the difference in the partial pressures of oxygen and carbon and because of
other physical and biochemical factors, the transformation processes in the water column and
in the sediment column are different. There are no phytoplankton and zooplankton species
292
in the sediment column. When these plankton species die, they are treated as particulate
organic species. In the sediment column, there exist two distinctive layers: an aerobic layer
and an anaerobic layer, within which the transformation processes are not the same. For
example, in the anaerobic layer, there is no nitrification process due to a lack of oxygen,
while in the aerobic layer, there is no denitrification process due to the availability of oxygen.
The vertically, one dimensional z-grid is shown in Figure D.l.
a>
o
Q.
-o
KMG
KMG-1
CO
o
3
o
o
3
KMS+2
KMS+1
KMS
KMR+1
KMR
V
KMG=KM+KMS+KMR
Water Column
KMS=KMR+KMO
7^
o
-V
Aerobic Sediment Column
KMR
J3
Anaerobic Sediment Column
Figure D.l The vertical one-dimensional z-grid
Since the horizontal transport is generally very weak in the sediment column, the
mass flows of nutrient species are mainly vertical, and the governing equations are:
For dissolved species in a sediment layer
dt dz
0M
dz >
+ Q (D.12)
293
For particulate species in a sediment layer
dNn
wrN+M p-
c P dz
+ Q (D.13)
dt dz
where Nd is dissolved nutrient species such as NH4, SON, SRP, SOP concentrations (per unit
volume of porewater), wc is the consolidation velocity of sediments, M is the molecular
diffusivity, and 0is the porosity.
APPENDIX E
NUTRIENT DYNAMICS
Nutrients are essential elements for life processes of aquatic organisms. Nutrients
of concern include carbon, nitrogen, phosphorous, silica and sulfur. Among these nutrients,
the first three elements are utilized most heavily by zooplankton and phytoplankton. Since
carbon is usually available in excess, nitrogen and phosphorous are the major nutrients
regulating the ecological balance in an estuarine system. Nutrients are important in water
quality modeling for several reasons. For example, nutrient dynamics are critical
components of eutrophication models since nutrient availability is usually the main factor
controlling algae bloom. Algae growth is typically limited by either phosphorous or
nitrogen. (Bowie et al., 1980)
Nutrient inputs to estuarine systems are related to point and non-point sources from
land, atmospheric deposition, and fixation. Additionally, internal loadings such as from
resuspended sediments containing inorganic and organic forms are also important. The
specification and quantification of each of these contributions are the first steps towards the
determination of nitrogen and phosphorous budgets in an estuarine system.
Nutrient cycles are highly dependent on the hydrodynamics and sediment dynamics
of the estuarine system. Resuspension events, combined with desorption processes, can
significantly change the input and budget of nitrogen and phosphorous in the system. On the
other hand, deposition and sorption may contribute to major losses of nitrogen and
294
295
phosphorous from the water column. The hydrodynamics not only derive the sediment
processes, but also affect the sorption/desorption reactions, through turbulent mixing.
E.l Nitrogen Cycle
Nitrogen can be classified into two groups: dissolved nitrogen and particular nitrogen.
The criterion of this division is established in the laboratory using filtering technique. The
dissolved nitrogen include ammonia nitrogen (NH3), dissolved ammonia nitrogen (NH4),
nitrite and nitrate nitrogen (N03), and dissolved organic nitrogen (SON). Particulate
nitrogen includes particulate inorganic nitrogen (PIN), and particulate organic nitrogen
(PON). Phytoplankton nitrogen (PhyN) and zooplankton nitrogen (ZooN) related biomass
to nitrogen concentration through a fixed stoichiometric ratio: nitrogen-to-carbon ratio (ANC).
The model nitrogen cycle (Figure E.l) includes the following processes.
1) Ammonification of organic nitrogen
2) Nitrification of ammonium
3) Volatilization of ammonia
4) Denitrification of nitrate
5) Uptake of ammonia and nitrate by phytoplankton
6) Conversion of phytoplankton nitrogen to zooplankton nitrogen by grazing
7) Excretion and mortality by phytoplankton and zooplankton
8) Settling for particulate nitrogen
9) Sorption/desorption reactions
296
tVolatili
NH3
Volatilization
NH3-(NH3),J
Instalbility
Excretion (Kzx'ZOON)
PIN
Kal — SS. — NH4
Hal + pH
Sorption/Desorption
d„(PIN-p„*C-NH4
" "
Zooplankton
Mortality
(1-KPDN)*K -ZOON
NH4
Nitrogen
Diffusion* Erosion/
Amonification
Mortality
KPDhTK 'ZOON
SON
Mortality
(1-KPDN)*Kas'PHYN
Uptake Pm"ua*PHYN
Excretion K./PHYN
Nitrification
Sorption/Desorptio
,(PON-p 'C'SO
on
PON
Mortality
KPDN'K -PHYN
Phytoplankton
Kim ^?— AW4
Hnit+DO
♦ Diffusion
N03
Uptake
(1-PJ'u-PHYN
f.|iIhsS:-;:< 4
•■ . .<■ "' .:,
PIN
Sorption/Desorption
°,„(PIN-P„-C-NH *
Instalbility
NH4
Amonification
SON
1
Water Column
fcrosioii/ ^Diffusion
iorption/Desorptio
„(PON-p "C
>rption
•SON!
NH3
Kal ^ AW 4r
Hal + pH
PON
Nitrification
Knn - AW 4
< < Hnil + DO
N03
♦ Diffusa
"t"""*""""""
N03
Denitrification
Kiln "n°3 A-03
Hno3+ DO
N2
Aerobic Layer
Anaerobic Layer
Figure E.l Nitrogen Cycle
Ammonification is the biological process of formation of ammonium from soluble
organic nitrogen. It is the first step of nitrogen mineralization, in which organic nitrogen is
converted to the more mobile, inorganic state. The rate of ammonification is expressed as
a first order reaction (Rao et al, 1984) :
0,
ammonification ** ONM ' " " ^
(E.l)
where Kom is the rate constant of ammonification which is a function of water temperature,
pH, and the C/N ratio of the residue (Reddy and Patrick, 1984).
The second step of mineralization of organic nitrogen is nitrification, which is an
oxidation of ammonia to nitrate( NO^ ) directly or to nitrite ( NO: ) and then to nitrate:
297
nh: + 1 .5a -> no; +ih++ h.o
(E.2)
Nitrification requires oxygen as the electron acceptor. Therefore, nitrification is a
strictly aerobic process, occurring only in the water column and in the aerobic layer of the
sediment column. This process is related to a sink of dissolved oxygen in the system as
shown in eq 4.11. The kinetics of nitrification are modeled as function of available
ammonia, dissolved oxygen, and temperature.
^nitrification ~ ~ ^ NN ~7l nTTi" 4 (E-3)
Mnit+DO
where KNN is nitrification rate which is a function of temperature; and Hni, is the half-
saturation constant for the bacteria growth.
The dissolved form of ammonium in water is generally not stable and can exist in its
gaseous form, or ammonia nitrogen. Because the ammonia concentration in the atmosphere
is very low, ammonia in the water column can escape to the air. This is a volatilization
process of ammonia. The volatilization of ammonia is a sink for nitrogen in an aquatic
system. A first order rate equation can be used to describe the kinetics of the ammonia
volatilization process (Chen and Sheng, 1994):
&***. = Kvol [KNH3 - NH°°* ] (E.4)
where KV0L is rate constant of volatilization which is a function of temperature. It can
be derived from the so-called two-film model (Jorgensen, 1983); hv is henry's constant; and NH"'m
is the ammonia concentration in the air.
Denitrification is defined as the biogeochemical transformation of nitrate nitrogen to
gaseous end products such as molecular nitrogen or nitrous oxide (Reddy and Patrick, 1 984).
298
While nitrification occurs in the water column and aerobic layer of the sediment column,
denitrification occurs only in the anaerobic layer of the sediment column. The denitrification
process can be described by the standard Michaelis-Menten equation (Bowman and Focht,
1974).
jj
U denitrification =~^dn~ " ~ ~ "03 (E.5)
Hno3 + DO
where Kdn is the denitrification rate, which is a temperature function; and Hn^ is the half-
saturation constant for denitrification.
In the nitrogen species, sorption processes refer to conversion from a soluble to a
solid phase of inorganic (NH4 to PIP) and organic (SON to PON) species, while desorption
reactions describe the inverse process. Sorption/desorption processes, combined with
resuspension events can significantly alter the nitrogen cycle in the system. The kinetics of
sorption/desorption reactions are dependent on nitrogen species characteristics, sediment
properties, pH, temperature, and dissolved oxygen concentration (Simon, 1989) The most
commonly used mathematical representation of sorption/desorption processes is the linear,
reversible, isotherm (Berkheiser et al., 1980; Reddy et al., 1988):
JtNad--Dr-Nud+Sr-Ns (E.6)
where Dr is the desorption rate constant which is a temperature function; Sr is the sorption
rate constant; Nad is the adsorbed nitrogen concentration such as particulate inorganic and
organic nitrogen; N, is the dissolved nitrogen concentration such as dissolved ammonium
nitrogen and dissolved organic nitrogen.
The ratio between the desorption and sorption rates gives the partition coefficient
dissolved and particulate forms, because dNad/dt = 0 at equilibrium.
299
Sr _ <d _
o = Pc (E.7)
Dr N
where A/^and AT" are the adsorbed and dissolved nitrogen, respectively, at the equilibrium
condition; and /?,. is the partition coefficient. Therefore, the kinetic equation for
sorption/desorption reaction is:
JtNad=-Dr-(Nad-pc-Ns) (E.8)
Inorganic nitrogen is incorporated by phytoplankton during growth and release as
ammonium and organic nitrogen through respiration and non-predatory mortality. The
phytoplankton nitrogen can be converted to zooplankton nitrogen by grazing process. The
kinetic processes for particulate nitrogen species need to include settling process, which
accounts for the limited vertical motion of particulate nitrogen. For this species, it is
reasonable to assume the same settling velocity of the suspended sediment particles.
The mass balance equations for nitrogen state variables are written by combining
nitrogen transformation processes.
Ammonia nitrogen (NH3)
include ammonia conversion and volatilization processes
For water column
d ._„ v pH
dt Hal + pH
hv-NH3-NHu3""] (E.9)
For sediment column:
dt Hal + pH
where Kal is the ammonia conversion rate constant which is a temperature function; Hal is
half-saturation constant for ammonia conversion.
Dissolved ammonium nitrogen (NH4)
300
include phytoplankton uptake and respiration, zooplankton respiration,
ammonification, nitrification, ammonia conversion, and sorption/desorption reaction.
For water column:
-NH4=-[(Pn.Ma- Kax ) • PhyC - Ku ■ ZooC] ■ ANC + K0NM ■ SON
-Km — NH4-Kal P— NH4 (E-11)
Hnil+DO 4 - Hal+PH 4
+dan(PIN-pan-c-NH4)
For sediment column:
-NH4=+KONM.SON-KNN.-^.NH4-Kar--^--.NH4
m Hni,+D0 Hal + pH (E.12)
+dan(PIN-Pan-c-NH4)
where dan is sorption/desorption rate of NH4 from sediment particles; pan is the partition
coefficient between NH4 and PIN; and c is the suspended sediment concentration.
Nitrate and nitrite nitrogen (NQ3)
include nitrification, denitrification and phytoplankton uptake processes.
For water column:
|-/V03 = +KNN ■ D° ■ NH4 - KDN ^ NO,
dt Hnil+DO Hlw3+DO 3 (e.13)
-ANCia-Pn>Ma'PhyC]
For sediment column:
Soluble organic nitrogen (SON)
include ammonification and sorption/desorption reaction.
For water column:
-SON = -KONM -SON + don(PON-pm-C'SON) (E.15)
For sediment column:
301
^SON = -K0NM ■ SON + d(m -(PON- Pon -c-SON) (E.16)
where d(m is sorption/desorption rate of SON from sediment particles; andpon is the partition
coefficient between SON and PON
Particulate organic nitrogen (PON)
include mortality of phytoplankton and zooplankton, settling, and a sorption-
desorption reaction.
For water column:
-PON = ANC\Kas-PhyC + Kzs.ZooC]--wsp-PON
-don(PON-Pon-c-SON)
For sediment column:
— PON = ws • PON - don ( PON - pon ■ c ■ SON ) (E. 18)
dt dz
where wsp is settling velocity for particulate species, which is same with that of suspended
sediment particles.
Particulate inorganic nitrogen (PIN)
include settling and a sorption/desorption reaction.
For both water and sediment columns:
^PIN = -j-wsp-PIN-dan(PIN-pan-c-NH4) (E.19)
E.2 Phosphorous Cycle
Phosphorous can be classified into two groups: dissolved phosphorous and particulate
phosphorous. The criterion for this division is established in the laboratory, using a filtering
technique. The dissolved phosphorous include soluble reactive phosphorous (SRP) and
dissolved organic phosphorous (SOP). Particulate phosphorous includes particulate
inorganic phosphorous (PIP), and particulate organic phosphorous (POP). Phytoplankton
302
phosphorous (PhyP) and zooplankton phosphorous (ZooP) related biomass to phosphorous
concentration, through a fixed stoichiometric ratio: phosphorous-to-carbon ration (APC).
The model phosphorous cycle (Figure E.2) includes the following processes:
1) Mineralization of organic phosphorous
2) Uptake of soluble reactive phosphorous by phytoplankton
3) Conversion of phytoplankton phosphorous to zooplankton phosphorous by grazing
4) Excretion and mortality by phytoplankton and zooplankton
5) Settling for particulate phosphorous
6) Sorption/desorption reactions
Excretion (Kzx'ZOOP)
PIP
Sorption/Desorption „_„
<upip-p -c-srpT "HK
Zooplankton
Mortality
(1-KPDP)*K 'ZOOP
Mineralization
Mortality
KPDP'K -ZOOP
SOP
Mortality
(1-KPDPCK 'PHYP
Uptake P -u "PHYP
Excretion K. /PHYP
Sorption/Desorption
V(pop-p -c-sopT
POP
Mortality
KPDP'K "PHYP
Phytoplankton
Phosphorous
Diffusion ♦ Erosion/
* Diffusion
Diffusion *
Water Column
Erosion/ 4 Diffusion
~
t deposition
PIP
Sorption/Desorption
"dJPIP-p -C'SRPy
_
deposition
SRP
Mineralization
SOP
Sorption/Desorption „_„
VJPOP-p-C'SOrf KUK
♦ Diffusion
T" *"
p.p Sorption/Desorption
♦ Diffusion Diffusion^
Mineralization
SRP
Aerobic Layer
Diffusion^
"T '
SOP
Sorption/Desorption
a„(POP-P -c-sopT
POP
opv ^op
Anaerobic Layer
Figure E.2 Phosphorous Cycle
The mineralization process is mediated by bacteria, which transfers dissolved organic
phosphorous to soluble reactive phosphorous, through the uptake of SOP and excretion of
303
SRP. Since bacteria abundance is related to algae biomass, the rate of organic phosphorous
mineralization is related to algae biomass. The mineralization of SOP is a relatively fast
process, it can take a few hours, compared to the mineralization of carbon and nitrogen,
which takes place in a few days (Golterman, 1973). Mineralization is highest when algae are
strongly phosphorous limited and is lowest when no limitation occurs. Thompson et al.
(1954) found that the mineralization of dissolved organic phosphorous is influenced by pH
value. An increase in pH causes a temporary increase in the rate of mineralization of
dissolved organic phosphorous. Temperature can also affect the speed of the mineralization
by stimulating the mineralization process with high temperatures.
The mineralization rate of SOP is usually modeled by a first order equation, as
follows (Jorgensen, 1983)
^mineralization ~ "■ opm ' ^U* (E.20)
where Koptn is a rate constant for mineralization of SOP, which is a function of pH and
temperature.
Soluble reactive phosphorous is incorporated by phytoplankton, during growth and
release, as soluble reactive phosphorous; and organic phosphorous through respiration and
mortality. The phytoplankton phosphorous is converted to zooplankton phosphorous by
grazing processes. The settling and sorption/desorption processes are similar to the nitrogen
species.
The mass balance equations for phosphorous state variables are written by combining
these phosphorous transformation processes.
Soluble reactive phosphorous (SRP)
include mineralization, uptake by phytoplankton, mortality of zoo and phytoplankton,
304
and sorption/desorption reaction.
For water column:
T- SRP = KoP>„ ■ SOP + Apc [-jua ■ PhyC + K^ ■ PhyC + K.x ■ ZooC]
m (E.21)
+dip(PIP-pip-c-SRP)
For sediment column:
JtSRP = Kopm ■ SOP + dip (PIP -Pip-c- SRP) (E.22)
where dip is sorption/desorption rate of SRP from sediment particles; and/?,p is the partition
coefficient between SRP and PIP.
Soluble organic phosphorous (SOP)
include mineralization and sorption/desorption reaction.
For both water and sediment columns:
-SOP = -Kopm -SOP-d^POP-p^c-SOP) (E.23)
where dl)p is sorption/desorption rate of SOP from sediment particles; andpop is the partition
coefficient between SOP and POP.
Particulate organic phosphorous (POP)
include respiration of zooplankton and phytoplankton, settling, and a
sorption/desorption reaction.
For water column:
d d
— POP = APC ( Kux ■ PhyC + K^ -ZooC)- — wsp- POP
-dop-(POP-Pop-c-SOP)
For sediment column:
(E.24)
jtPOP = - — wspPOP-dop\POP-pop-c-SOP) (E.25)
305
Particulate inorganic phosphorous (PIP)
include settling and a sorption/desorption reaction
For both water and sediment columns:
-PIP = -—wSp-PIP-dip-(PIP-Pip.cSRP) (E.26)
APPENDIX F
SEDIMENT FLUX MODEL
DiToro et al. (1990) developed a model of the SOD process in a mechanistic fashion
using the square-root relationship of SOD to sediment oxygen carbon content. Using similar
analysis as applied to carbon, they also evaluate the effect of nitrification on SOD. In this
model, carbon and nitrogen diagenesis are assumed to occur at uniform rates in a
homogeneous layer of the sediment of constant depth (active layer). The sediment oxygen
demand and sediment fluxes are calculated by simple first-order decay processes of the
concentrations of particulate organic carbonaceous material and of particulate organic
nitrogenous material in this active layer as follow:
dC
poc
= -kpocCpoc+Mc (F.l)
at
dC
, K pon^ pon ~ ln N \L --J
where CP0C is a concentration of POC in sediment (g 02/m3)
CP0N is a concentration of PON in sediment (g 02/m3)
KP0C is a decay rate of POC in sediment (g 02/m3)
KPON is a decay rate of POC in sediment (g 02/m3)
MN is a source term for CP0C (g/m3-day)
Mc is a source term for CPON (g/m3-day)
Sediment carbon and nitrogen diagenesis fluxes, Jc and JN, are the most important
306
307
parameters in the equations for sediment oxygen demand and sediment fluxes as defined by
h = ~kpocCpocH
i -b r h F3)
JN ICpon(^pon"
where H is the depth of the active layer (m)
Using these diagensis fluxes, DiToro and Pitzpatrick developed sediment flux model
which include NSOD, CSOD, ammonia and nitrate flux, methane flux, and sulfide flux.
F.l Sediment Flux Equations inCH3D-WQ3D
The reactive portion of particulate organic carbon and particulate organic nitrogen
in the sediment are presented by CBOD and NH4 in CH3D-WQ3D model system. The
equation (F.3) can be rewritten as
Jc=-kgxy-Q™CBODstd.H
(R4)
JN=-KM<i,-°-NHAsed-H
where koxy is a oxidation rate in sediment (g 02/m3)
kni[ is a nitrification rate in sediment (g 02/m3)
®l~y° is a temperature effect of oxidation
@Jt' is a temperature effect of nitrification
In this study, sediment fluxes include sediment oxygen demand, benthic dissolved
methane flux, and benthic methane gas flux. Sediment oxygen demand (SOD) can be
determined for the set of equations developed by DiToro and Fitzpatrick (1993).
Carbonaceous sediment oxygen demand (g/m2/d):
CSOD = J2KDCSJC
f
1-sec/z
KrO,
SOD
Note: The square root term is replaced by Jc ifJc <2KDC,
Nitrogenous sediment oxygen demand (g/m2/d):
(F.5)
308
/
NSOD = 1.714 J>
Total sediment oxygen demand (g/m2/d):
1 - sec h
V
KNQ2
SOD
(F.6)
SOD = CSOD + NSOD (F.7)
where 02 is a concentration of dissolved oxygen in overlying water column (g 02/m3).
kD is a dissolved methane diffusion mass transfer coefficient (g 02/m3)
kc is a reaction velocity for methane oxidation (g 02/m3)
kN is a reaction velocity for ammonia oxidation (g 02/m3)
C, is a methane solubility (g 02/m3)
The magnitude of the fluxes at the sediment water interface of aqueous methane and
gaseous methane are predicted to be
•* CHA(aq) V *> S C
sec/z
Kc02
SOD
"'CH4(g)~''C V O 5 C
(F.8)
(F.9)
The original DO kinetic equation in CH3D-WQ3D is modified to incorporate the
sediment exchange predictions of the DiToro model, given by equation (F.10).
dDO
dt
-Oxydation - Nitrification + reaeration + photosynthesis - respiration
SOD JcH4(aq) ^ "* CH4(g)
(F.10)
H H J,u' H
where Gfrac is a fraction of gaseous methane produced in the sediment.
His a water depth.
The first five source and sink terms in the equation (F.10) are the same with the
original DO kinetic equation, while the last three terms represent oxygen demand due to
sediment exchange processes, where SOD and aqueous and gaseous methane flux rates.
APPENDIX G
MODEL PERFORMANCE TEST WITH PARALLEL CH3D-IMS
To reduce the computational time necessary to simulate multiple-year seasonal
response of the Charlotte Harbor estuarine system with a serial CH3D model, a parallel
CH3D model was developed and validated. The parallel approach is applied to the parallel
CH3D model via parallel constructs added to the serial CH3D model (Davis and Sheng,
2000; Sheng et al., 2003). These parallel constructs are implemented by adding additional
macros to the original serial source code. By defaults, the parallel CH3D model uses
OpenMP constructs although due to the flexible nature of the implementation process, either
Sun Microsystems-style or Clay-style constructs can be used. The parallel source code
closely resembles the serial source code and maintains perfect compatibility with the serial
code. In other words, the parallel CH3D model can be executed using the same input files
as the serial CH3D model and parallel CH3D output format is identical to the output format
of the serial CH3D model. To determine how well the parallel CH3D model performs in
Charlotte Harbor estuarine system, the serial and parallel CH3D simulations performed for
the parallel model validation.
G.l Validation of Parallel Model
To validate the parallel CH3D model, simulations are performed using both the serial
and parallel CH3D model. CH3D model output from all of the parallel simulations are then
compared to their respective serial simulations to fully validate the parallel CH3D model.
309
310
The computational grid for Charlotte Harbor estuarine system contains 92 x 129 horizontal
cells (Figure 5 . 1 ) and 8 vertical layers, with a total 1 1 648 grid cells which include 5367 water
cells and 6281 land cells. Grid spacing varies from 40 to 2876 meters (average 598 meters).
Using this grid, the 3-D hydrodynamic and water quality simulation simulate the circulation,
sediment transport, and water quality dynamics of the Charlotte Harbor estuarine system
from May 23 to June 22, 1996. The initial condition for simulation is provided by a 30-day
spin up simulation (April 23 to May 23) previously performed during the dry season with all
forcing mechanisms (tides, river discharges, wind) to allow water level, velocity and salinity
field to reach dynamic steady-state throughout the computational domain. A more detailed
description of the simulation can be found in Chapter 6.
Simulated time series and field parameters at the end of 30-day simulation of each
parallel simulation are compared to the all corresponding simulated parameters from the
serial CH3D simulation which include water level, current velocity, temperature, salinity,
sediment concentration, and all water quality species. For all comparisons, the simulated
time series and field parameters obtained from the serial CH3d model are identical to the
simulated parameters obtained from the parallel CH3D model. Thus the parallel CH3D
model is completely validated.
G.2 CPU Times of Parallel Routines
The 1 -month simulation is simulated using both serial and parallel CH3D models on
the SGI orgin system witch is Silicon Graphyics Origin 3400, 400 MHz MIPS R 12000 (IP
35). This platform has 16 processors and 8 GB main memory size. The parallel model is
executed using from 1 to 4 processors on this system. Table F.l report the CPU times per
iteration for the parallel CH3D model executed on this platform. This timing results show
311
that the time necessary for CH3D simulation can be significantly reduced using multi-
processor computers coupled with parallel techniques.
Table G.l CPU time for the parallel, shared memory, CH3D procedures on SGI origin
platform. Time shown are per time step iteration of the model using computational grid
(92x 1 29) and are given in seconds, n is the number of processors used and speed up is shown
in parenthesis.
Serial
n=1
n=2
n=3
n=4
Main WQ
1.221(1.00)
1.287(0.95)
0.680(1.80)
0.469(2.60)
0.370(3.30)
Main Sediment
0.126(1.00)
0.125(1.01)
0.078(1.62)
0.060(2.10)
0.052(2.41)
Turbulence
0.100(1.00)
0.107(0.93)
0.059(1.70)
0.045(2.21)
0.038(2.61)
Sediment transport
0.090(1.00)
0.089(1.01)
0.047(1 .94)
0.032(2.85)
0.025(3.62)
Dimensionalize
0.075(1 .00)
0.083(0.90)
0.047(1.62)
0.033(2.28)
0.026(2.89)
Baroclinic (J)
0.071(1.00)
0.075(0.95)
0.042(1 .70)
0.038(1.87)
0.036(1.97)
Baroclinic (I)
0.070(1.00)
0.067(1.05)
0.038(1.86)
0.025(2.75)
0.020(3.49)
Salinity
0.055(1.00)
0.059(0.95)
0.030(1.85)
0.020(2.51)
0.016(3.43)
N.L/Diffusion (J)
0.055(1.00)
0.055(1.00)
0.033(1.65)
0.030(1 .84)
0.029(1.87)
N.L/Diffusion (I)
0.042(1 .00)
0.043(0.98)
0.025(1.65)
0.018(2.34)
0.015(2.82)
Layer Vel. (v)
0.040(1.00)
0.041(0.97)
0.027(1.48)
0.021(1.86)
0.019(2.12)
Layer Vel. (u)
0.028(1 .00)
0.029(0.97)
0.016(1.73)
0.011(2.60)
0.009(3.28)
Layer Vel. (w)
0.011(1.00)
0.011(0.99)
0.006(1.70)
0.004(2.37)
0.003(2.85)
Interpolation
0.011(1.00)
0.011(1.00)
0.006(1.79)
0.005(2.62)
0.004(3.52)
integrate Vel. (U)
0.008(1 .00)
0.008(1.00)
0.006(1.34)
0.004(1.69)
0.004(1.87)
integrate Vel. (V)
0.006(1.00)
0.006(0.97)
0.004(1.35)
0.005(1 .32)
0.005(1 .26)
Bottom sheer stress
0.001(1.00)
0.001(1.01)
0.001(1.89)
0.000(2.76)
0.000(3.60)
Wave H/T
0.001(1.00)
0.001(1.03)
0.000(1.95)
0.000(2.93)
0.000(3.87)
All Parallel Routine
1.973(1.00)
2.060(1.00)
1.152(1.71)
0.843(2.34)
0.701(2.82)
Total Routine
2.022(1 .00)
2.107(1.00)
1.204(1.68)
0.895(2.26)
0.753(2.69)
G.3 Parallel Speedup
The parallel speed up for the Charlotte Harbor 1 -month simulation described earlier
on the SGI Origin platform is shown in Figure G.l. As the lines get close to the theoretical
maximum, the parallel model is performing better. The speedups of the individual parallel
procedures are shown as values inside parenthesis in the previous table G. 1 and also illustrate
312
how computationally intense routines have higher speedups. Applying parallel method for
CH3D model achieved a 1.71x and 2.82x speedups using 2 processors and 4 processors,
respectively.
2 3
Numbers of Processors
Figure G.l Parallel speedup gained in performing the simulation on the SGI Origin
platform
APPENDIX H
TIDAL BENCH MARKS FOR CHARLOTTE HARBOR
Table H.l Tidal datum referred to Mean Low Low Water (MLLW), in meter.
Station Station Latitude Longitude MHHW MHW MLW NAVD
Number Name
8725853
Venice
27 04.3
82 27.3
0.671
0.589
0.113
0.496
8725791
Peace
River
26 59.3
81 59.6
0.615
0.537
0.119
0.501
8725781
Shell
Creek
26 58.8
81 57.6
0.668
0.570
0.135
0.472
8725541
Bokeelia
26 42.4
82 09.8
0.526
0.481
0.072
0.495
8725520
Fort Myers
26 38.8
81 52.3
0.401
0.335
0.191
0.318
8725391
Sanibel
26 29.3
82 00.8
0.689
0.614
0.145
0.570
8725110
Naples
26 07.8
81 48.7
0.874
0.797
0.184
0.696
313
REFERENCES
Ambrose, R. B. Jr., Wool, T.A., Martin, J.L., Connolly, J.P., Schanz, R.W. (1994) WASP5,
A hydrodynamic and water quality model-model theory, user's manual, and programmer's
guide. U.S. EPA Environmental Research Laboratory. Athens, Georgia.
American Public Health Association(APHA) (1985). Standard methods for the examination
of water and wastewater. APHA. Washington, D.C.
Banks, R.B. (1975). Some features of wind action on shallow lakes, ASCE, Journal of
Environmental Engineering Division, Vol. 101, No. EE5, p8 13-827
Banks, R.B. and Herrera, F.F. (1977). Effect of wind and rain on surface reaeration, ASCE,
Journal of Environmental Engineering Division, Vol. 103, No. EE3, p489-504
Baker, B. (1990). Draft Caloosahatchee water quality based effluent limitations
documentation (Lee County). Florida Department of Environment Regulation. Water Quality
Tech. Ser. 2. No. 121: 18.
Berkheiser, V.E., Street, J.J., Rao, P.S.C., and Yuan, T.L. (1980) Partitioning of inorganic
orthophosphate in soil-water systems. CRC Critical Review in Environmental Control.
Berliand, M.E., and Berliand, T.G. (1952). Determining the net longwave radiation of the
Earth with the consideration of the effect of cloudiness, Isv. Acad. Nauk., SSSR Ser. Geofis,
No. 1.
Boler, R.N., Malloy, R.C., and Lesnett, E.M. (1991) Surface water quality monitoring by the
Environmental Protection Commission of Hillsborough County. In (S.F. Treat and Clark
P.A., eds.) Proceedings of Tampa Bay Area Scientific Information Symposium 2, 1991.
Tampa, FL.
Bowie, G.L., Mills, W.B., Porcella, D.B., Campbell, C.L., Pagenkopf, J.R., Rupp, G.L.,
Johnson, K.M., Chan, P.W.H., Gherini, S.A. and Chamberlin, C.E. (1985). Rates, constants,
and kinetics formulations in surface water quality modeling. In: (Editors), U.S.
Environmental Protection Agency, Office of Research and Development, Athens, GA
Bowie, G.L., Chen, C.W. and Dykstra, D.H. (1980) Lake Ontario ecological modeling, phase
HI. Tectra-Tech, Inc., Lafayete, CA.
314
315
Bowman, R.A. and Focht, D.D. (1974). The influence of glucose and nitrate concentrations
upon denitrification rates in sandy soils. Soil Biol. Biochem. 6: 297-301.
Bricaud, A., Morel, A., and Prieur, L. (1981). Absorption by dissolved organic matter of the
sea (yellow substance) in the uv and visible domains. Limnology and Oceanography,
28(5):43-53.
Budyko, M. I. (1974). Climate and Life. Academic Press Inc., New York, 508pp.
Camp, Dresser, & Mckee (1998). The Study of Seasonal and Spatial Patterns of Hypoxia in
the Upper Charlotte Harbor. Prepared for S WFWMD, Prepared by Camp, Dresser, & Mckee
Canale, R.P., DePalma, L.M., and Vogal, A.H. (1976). A plankton-based food web model
for Lake Michigan, in (R.P. Canale, ed.) Modeling Biochemical Processes in Aquatic
Ecosystem, Ann Arbor Science Publishers Inc., P.O. Box 1452, Ann Arbor, Michigan 48 160.
Cerco, D.F., and Cole, T.M. (1994). Three-dimensional eutrophication model of Chesapeake
Bay. Technical Report El-94-4, U.S. Army Engineer Waterways Experiment Station,
Vicksburg, MS.
Cerco, D.F., and Cole, T.M. (1995). User Guide to the CE-QUAL-ICM three-dimensional
eutrophication model, Technical Report EL-95-15, U.S. Army Corps of Engineers,
Washington, DC.
Chamberlain, R.H. and Doering, P.H. (1998). Freshwater inflow to the Caloosahatchee
Estuary and the resource-based method for evaluation. Proceedings of the Charlotte Harbor
public conference and technical symposium, Technical Report No. 98-02: 81-90.
Chapra, S.C. (1997). Surface water-quality modeling. The McGRAW-HILL companies,
INC., New York.
Chen, C.W., andOrlob, G.T. (1975). Ecologic simulation of aquatic environments. Systems
Analysis and Simulation in Ecology, Vol. 3, B.C. Pattern (ed.). Academic Press, New York:
475-588.
Chen, X. and Sheng, Y.P.(1994). Effects of hydrodynamics and sediment transport processes
on nutrient dynamics in shallow lakes and estuaries. Ph.D. dissertation. University of Florida,
Gainesville, Florida
Christian, D.J. and Sheng, Y.P.(2003). Modeling the Effects of Hydrodynamics, Suspended
Sediments, and Water Quality on Light Attenuation in Indian River Lagoon, Florida. PhD
dissertation, University of Florida., Gainesville, Florida.
Clark, N. E., Eber R. M. Renner J. A., and Saur J. F. T. (1974) Heat exchange between ocean
and atmosphere in the eastern North Pacific for 196-71. NOAA Tech. Rep. NMFS SSRF-
682, U.S.Dept. Commerce, Washington. ,D.C.
316
Davis, J.R. and Sheng, Y.P. (2000). High performance estuarine and coastal environmental
modeling: The CH3D example. In Estuarine and Coastal Modeling, Vol. 6, p470-486.
American Society of Civil Engineers.
Davis, J.R. and Sheng, Y.P. (2001). High Performance Modeling of Circulation and
Transport in the Indian River Lagoon, Florida. PhD dissertation, University of Florida.,
Gainesville, Florida.
Day, J.W. Jr., Hall, C.A.S, Kemp, K.M., and Arancibia, A.J. (1989). Estuarine Ecology.
John Wiley & Sons. New York.
DeCosmo, J., Katsaros, K.B., Smith, S.D., Anderson, R.J., Oost, W.A., Bumke, K., and
Chadwick, H.M. (1996). Air-sea exchange of water vapor and sensible heat: The Humidity
Exchange Over the Sea (HEXOS) results. J. Geophys. Res.-Oceans, 101: 12,001-12,016.
Degrove, B. (1981). Caloosahatchee River wasterload allocation documentation. Florida
Dept. Of Env. Reg., Water Quality Tech ser. 2. No. 52: 17.
Degrove, B. and Nearhoof, F. (1987). Water quality assessment for Caloosahatchee River.
Florida Dept. Of Env. Reg. Water Quality Tech. Ser. 3. No. 19.
DiToro, D.M., Paquin, P.R., Subburamu,K., and Gruber, D.A. (1990). Sediment oxygen
demand modehMethan and ammonia oxidation. Journal of Evironmental Engineering,
116(5), p945-986.
DiToro, D.M. and Fitzpatrick, J.J. (1993). Chesapeaka bay sediment flux model. U.S. Army
Corps of Engineers, Waterways Experiment Station, Tech. Report EL-93-2.
Dixon, L.K., and Gary, J.K. (1999). Cause of light attenuation with respect to seagrasses in
upper and Lower Charlotte Harbor. Marine Laboratory Technical Report No. 650. Prepared
for Southwest Florida Water Management District.
Doering, P.H. (1996). Temporal variability of water quality in the St. Lucie Estuary, South
Florida. Water Resource Bulletin 32:1293-1306.
Echternacht, K.L. (1975). A study of the precipitation regimes of the Kissimmee River-Lake
Okeechobee watershed. Florida Department of Environmental Regulation Technical Service
(3)1. Tallahassee, Florida.
Eckart, C. (1958). Properties of water. Part H The equation of state of water and sea water
at low temperatures and pressures. American Journal of Science, 256(4): 224-240.
Edwards, R.E., Lung, W., Montagna, P. A., and Windom, H.L. (2000). Final review report.
Caloosahatchee Minimim Flow Peer Review Panel, September 27-29, 2000. Report to the
south Florida Water Management District, West Palm Beach, Florida.
317
Environmental Quality Laboratory, Inc. (1979). Hydrobiological monitoring report to the
Southwest Florida Water Management District, Port Charlotte, Florida.
Environmental Quality Laboratory, Inc. (1987). Hydrobiological monitoring program, data
report for the period from March 1986 through Feburaury 1987 covering the lower Peace
River abd Charlotte Harbor. Port Charlotte, Florida, Environmental Quality Laboratory, Inc.:
92.
Eppley, R.W. (1972). Temperature and phytoplankton growth in the sea. Fish. Bull.,
70:1063-1085.
Estevez, E. D. (1986). Infaunal macro invertebrates of the Charlotte Harbor estuarine system
and surrounding inshore waters, Florida.. U.S. Geological Survey Water-Resources
Investigations Report 85-4260: 116.
Estevez, E.D. (1998). The story of the greater Charlotte Harbor Watershed, Charlotte Harbor
Estuarine Program, North Fort Myers, Florida.
Fraser, T.H. (1986). Long-term water-quality characteristics of Charlotte Harbor, Florida.
U.S. Geological Survey Water-Resources Investigations Report 86-4180: 43.
Fraser, T.H. and Wilcox, W.H. (1981). Enrichment of a subtropical estuary with nitrogen,
phosphorous, and silica, in Neilson, B.J., and Cronin, L.E., eds., Estuarine and Nutrients,
Humana Press, New Jersey: 481-498.
Friehe, C. A. and Schmitt, K. B. (1976). Parameterization of air-sea interface fluxes of
sensible heat and moisture by the bulk aerodynamic formulas. Journal of Physical
Oceanography, 6, 801-809.
Gallegos, C.L. (1993). Determination of optical water quality requirements in the Indian
River near Ft. Pierce, FL, with emphasis on the impact of colored water discharges.
Technical report, South Florida Water Management District.
Gallegos, C.L. and Correl, D.L. (1990). Modeling spectral diffuse attenuation, absorption,
and scattering coefficients in a turbid estuary. Limnology and Oceanography, 35(7): 1484-
1502.
Gallegos, C.L. and Kenworthy, W.J. (1996). Seagrass depth limits in the Indian River
Lagoon (Florida, USA). Estuarine, Coastal, and Shelf Science, 42:267-288.
Garret, J.R. (1977). Review of drag coefficients over oceans and continents. Monthly
Weather Review 105:915-929.
Golterman, H.L. (1973). Vertical Movement of Phosphorous in Freshwater, in (E.J. Griffith,
A. Beeton, J.M Spencer, and D.T. Mitchell, eds.) Environmental Phosphorous Handbook,
John Wiley & Sons., New York.
318
Goodwin, C.R. (1996). Simulation of tidal-flow, circulation, and flushing of the Charlotte
Harbor estuarine system, Florida. U.S. Geological Survey Water-Resources Investigations
Report 93-4153.
Goodwin, C. R. and Michaelis, D.M. (1976). Tides in Tampa Bay, Florida: June 1971 to
December 1973. U.S. Geological Survey Open-File Report FL-75004: 338.
Hammet, K. M. (1990). Land use, water use, streamflow characteristics, and water-quality
characteristics of the Charlotte Harbor inflow area, Florida. U.S. Geological Survey Water-
Supply Paper 2350-A: 64.
Hammet, K. M. (1992). Physical processes, salinity characteristics, and potential salinity
changes due to freshwater withdrawals in the tidal Myakka River, Florida. U.S. Geological
Survey Water-Resources Investigations Report 90-4054:20
Heyl, M.G. (1997). Hypoxia in upper Charlotte Harbor. Proceedings of the Charlotte Harbor
public conference and technical symposium, Technical Report No. 98-02: 219-228.
Hwang, T.C. (1966). A Sedimentologic Study of Charlotte Harbor, Southwestern Florida.
Master's thesis, Florida State University, Tallahassee, Florida.
Jaworski, N.A. (1981). Source of Nutrients and the scale of eutrophication problems in
estuaries, in Neilson, N.J., Jersey, Humana Press: p83-110.
Jerlov, N.G. (1976). Marine Optics. Elsevier Scientific Publishing Company, New York.
Jorgensen, S.E. (1976). A eutrophication model for a lake. Ecological Modeling 2: 147-165.
Jorgensen, S.E. (1983). Modeling the ecological processes. In: (G.T. Orlob, ed.)
Mathematical Modeling of Water Quality: Streams, Lakes, and Reservoirs. John Wiley and
Sons. New York. Pp 116-149.
Jorgensen, S.E., Sorensen, B.H., and Nielsen, S.N. (editors) (1996). Handbook of
Environmental and Ecological Modeling. CRC Press, Inc.
Jorgensen, S.E., and Gromiec, M.E. (editors) (1989). Mathematical sub-models in water
quality systems. Development in Environmental Modeling, 14. Elsevier Science Publishers.
Ketchem, B.H. (1967). Phytoplankton nutrients in estuaries, in Lauff, G. H., ed., Estuaries.
National Oceanic and Atmospheric Administration
Kirk, J.T.O. (1984). Dependence of relationship between inherent and apparent optical
properties of water on solar altitude. Limnology and Oceanography, 29(2):350-356.
319
Lam, D.D.L., Schertzer, W.M., and Fraser, A.S. (1983). Simulation of Lake Erie water
quality responses to loading and weather variations, I.W.D. Scientific Series, 134,
Environment Canada, Ottawa:299.
McEwan, J., Gabric, A.J. & Bell, P.R.F. (1998). Water quality and phytoplankton dynamics
in Moreton Bay, Southeastern Queensland, n. Mathematical modeling. Marine and
Freshwater Research, Vol. 40, No. 3, p215-225.
McPherson, B. F. and Miller, R.L. (1990). Nutrient distribution and variability in the
Charlotte Harbor estuarine system, Florida. Water Resource Bulletin 26: 67-80.
McPherson, B. F., Miller, R. L. and Stoker, Y. E. (1996). Physical, Chemical, and Biological
Characteristics of the Charlotte Harbor Basin and Estuarine System in Southwestern Florida-
A Summary of the 1982-89 U.S. Geological Survey Charlotte Harbor Assessment and Other
Studies. U.S. Geological Survey Water-Supply Paper 2486.
Metha, A.J. (1986). Characterization of cohesive sediment properties and transport processes
in estuaries. In:A.J. Metha (ed.), Estuarine Cohesive Sediment Dynamics. Springer, New
York:290-325.
Mehta, A.J. (1989). On estuarine cohesive sediment suspension behavior. Journal of
Geophysical Research 94(C 10): 14303-14314.
Mellor, G. L., (1996), Introduction to Physical Oceanography, Springer, Verlag New York.
Miller R. L. and McPherson B. J., (1995). Modeling Photosynthetically Active Radiation in
Water of Tampa Bay, Florida, with Emphasis on the Geometry of Incident Irradiance.
Estuarine, Coastal and Shelf Science, 40, 359-377 .
Morel, F. (1983). Principles of aquatic chemistry. John Wiley and Sons, New York, NY: 150.
Morel, A., and Gentili, B. (1991). Diffuse reflectance of oceanic waters: Its dependence on
sun angle as influenced by the molecular scattering contribution. Applied Optics 30:4427-
2238.
Morrison, G. (1997). Proposed trophic state goals and nitrogen management objectives for
the tidal reaches of the Peace and Myakka rivers. Draft. Prepared by Southwest Florida Water
Management District, Surface Water Improvement and Management (SWIM) Section.
O'Connor, D.J. and Thomann, R.V. (1972). Water quality models: chemical, physical and
biological constituents. In: Estuarine Modeling: an Assessment. EPA Water Pollution
Control Research Series 16070 DZV, Section 702/71.
Oliveria, J., Burnett, W.C., Mazzilli, B.P., Brags, E.S., Farias, L.A., Christoff, J., and
Furtado, V. V. (2003). Reconnaissance of submarine groundwater discharge at ubatuba coast,
Brazil, using rn-222 as a natural tracer, Journal of Radioact., 69:37-52.
320
Orth, R.J. and Moore, K.A. (1983). Chesapeake Bay: An unprecedented decline in
submerged aquatic vegetation. Science, v. 222, No. 4619: 51-53.
Phillips, N.A. (1957). A coordinate system having some special advantage for numerical
forecasting. Journal of Meteor. 14:184-185.
Pickard , G. L . and Emery, W. J. (1990). Descriptive physical Oceanography. Pergamon
Press, New York. Fifth (SI) Enlarged Edition: 320.
Pribble, J.R., Wade, D.L., Squires, A.D., and Janicki, A.J. (1998). A mechanic water quality
model for the tidal Peace and Myakka rivers. Proceedings of the Charlotte Harbor public
conference and technical symposium, Technical Report No. 98-02: 242-258.
Prieur, L. and Sathyendranath, S. (1981). An optical classification of coastal and
oceanographic waters based on the specific spectral absorption curves of phytoplankton
pigments, dissolved organic matter, and other particulate materials. Limnology and
Oceanography, 26(4):67 1-689.
Rao P.S.C., Jessup, R.E., Reddy, K.R. (1984). Simulation of nitrogen dynamics in flooded
soils. Soil Science 138:54-62.
Reddy, K.R., Jessup, R.E., Rao, P.S.C. (1988). Nitrogen dynamics in a eutrophic lake
sediment. Hydrobiologia 159:177-188.
Reddy, K.R., and Patrick, W.H. (1984). Nitrogen transformations and loss in flooded soil and
sediments. CRC Critical Reviews in Environmental Control 13(4):273-309.
Redfield, A., Ketchman, B., and Richar, F. (1966). The influence of organisms on the
composition of sea-water. The Sea VolII. Interscience Publishers, New York, 26-48.
Reed, R. K. (1977). On estimating insolation over the ocean. Journal of Physical
Oceanography, 7, 482-485. Research 49(3): 227-239.
Rutkowski, C.W., Burnett, W.C., Iverson, R.L., and Chanton, J.P. (1999). The effect of
groundwater seepage on nutrient delivery and seagrasss distribution in the northern Gulf of
Mexico. Estuaries, 22:2033-1040.
Schropp, S.J. (1998). Charlotte Harbor Sediment Quality Data Review and Evaluation.
Proceedings of the Charlotte Harbor public conference and technical symposium, Technical
Report No. 98-02: 35-47.
Sheng, Y.P. (1983). Mathematical modeling of three-dimensional coastal currents and
sediment dispersion: model development and application. Coastal Engineering Research
Center Tech. Rep. CERC-TR-83-2. U.S. Army Corps of Engineers.
321
Sheng, Y.P. (1986). Modeling bottom boundary layers and cohesive sediment dynamics.
In:(A.J. Mehta, ed.) Estuarine Cohesive Sediment Dynamics. Pp.360-400. Springer- Verlag.
Berlin.
Sheng, Y.P. (1987). On modeling three-dimensional estuarine and marine hydrodynamics.
In Nihoul and Jamart, editors, Three-dimensional models of marine and estuarine dynamics,
pages 35-54. Elswvier, Amsterdam.
Sheng, Y.P. (1989). Evolution of a 3-D curvilinear grid hydrodynamic model: CH3D.
In:(M.L. Spaulding, ed.) Proceedings of the Estuarine and Coastal Modeling Conference.
ASCE, Newport, R.I., pp.40-49.
Sheng, Y.P. (1994). Modeling hydrodynamics and water quality dynamics in shallow waters.
In Proceedings of the first International Symposium on Ecology and Engineering, Taman
Negara, Malaysia. University of Western Australia and Malaysian Technical University.
Sheng, Y.P. (1995). On Modeling circulation in Florida Bay. In American Geophysical
Union Conference on Circulation in Intra- American Seas, Pueto Rico. American Geophysical
Union.
Sheng, Y.P. (1998). Circulation in the Charlotte Harbor estuarine system. Proceedings of the
Charlotte Harbor public conference and technical symposium, Technical Report No. 98-02:
91-110.
Sheng, Y.P. (2000). A framework for integrated modeling of coupled hydrodynamic-
sedimentary-ecological processes. In Estuarine and Coastal Modeling, Volume 6, pages 350-
362. American Society of Civil Engineers.
Sheng, Y.P., Ahn, K.M., and Choi, J.K. (1990). Wind-wave hindcasting and estimation of
bottom sheer stress in Lake Okeechobee. Technical Report UFL/COEL-93/02 1 , Coastal and
Oceanographic Engineering Department, University of Florida, Gainesville, Florida.
Sheng, Y.P., Chen, X.J., Schofield, S., and Yassuda, E. (1993). Hydrodynamics, sediment
and phosphorous dynamics in Lake Okeechobee during an episodic event. Technical Report,
Coastal and Oceanographic Engineering Department, University of Florida, Gainesville,
Florida.
Sheng, Y.P., and Chiu, S.S. (1986). Tropical cyclone generated currents. In Proceedings of
the 20th International Conference on Coastal Engineering, pages 737-751. American Society
of Civil Engineers.
Sheng, Y.P., Christian, D., and Kim, T.Y. (2002c). A IRL light attenuation model. Indian
River Lagoon Pollutant Load Reduction (IRLPLR) Model Development Project. Final Report
Volume II (A 3-D Water Quality Model) to St. Johns River Water Management District.
Coastal and Oceanographic Engineering Department, University of Florida, Gainesville,
Florida.
322
Sheng, Y.P., Davis, J.R., and Park, K. (2003) Development of parallel computing version of
CH3D: Parallel CH3D User's Manual. Amendment to the three dimensional water quality
model of the Charlotte Harbor estuarine system for the South Florida Water Management
District.
Sheng, Y.P., Davis, J.R., Sun, D.,Qiu, C, Park, K., Kim, T.Y., and Zhang, Y. (2001).
Application of an integrated modeling system of estuarine and coastal ecosystem to Indian
River Lagoon, Florida. In Estuarine and Coastal Modeling, Session 4B. American Society
of Civil Engineers, pp. 329-343
Sheng, Y.P., and Lick, W. (1979). Resuspension and transport of sediments in a shallow
lake. Journal of Geophysical Research, V. 84, pp. 1809-1826.
Sheng, Y.P. and Park, K. (2000). A 3-D Circulation and Salinity Model of Charlotte Harbor
Estuarine System. Phase I Final Report to Southwest Florida Water Management Department
(SFWMD), University of Florida, Gainesville, Florida.
Sheng, Y.P. and Park, K. (2001). Modeling the Effects of Caloosahatchee Discharges on
Circulation and Salinity in Charlotte Harbor. Interim Report Southwest Florida Water
Management Department (SFWMD), University of Florida, Gainesville, Florida.
Sheng, Y.P. and Park, K. (2002). 3-dimensional Water Quality model for Charlotte Harbor
estuarine system. Phase HI Final Report Southwest Florida Water Management Department
(SFWMD), University of Florida, Gainesville, Florida.
Sheng, Y.P., Qiu, Q., Park, K., and Kim, T.Y. (2002b). A 3-D water quality model. Indian
River Lagoon Pollutant Load Reduction (IRLPLR) Model Development Project. Final Report
Volume II (A 3-D Water Quality Model) to St. Johns River Water Management District.
Coastal and Oceanographic Engineering Department, University of Florida, Gainesville,
Florida.
Sheng, Y.P., Sun, D., and Zhang, Y. (2002a). A 3-D IRL suspended sediment transport
model. Indian River Lagoon Pollutant Load Reduction (IRLPLR) Model Development
Project. Final Report to St. Johns River Water Management District. Coastal and
Oceanographic Engineering Department, University of Florida, Gainesville, Florida.
Sheng, Y.P., and Villaret,C. (1989). Modeling the effect of suspended sediment stratification
on bottom exchange processes. Journal of Geophysical Research - Oceans, 94(C10): 14229-
14444.
Simon, N.S. (1989). Nitrogen cycling between sediment and the shallow-water column in
the transition zone of the Potomac River and Estuary. U. The role of wind-driven
resuspension an adsorbed ammonium. Estuarine, Coastal and Shelf Science 28:531-547.
Smith, R.C. and Baker, K. (1981). Optical properties of the clearest waters. Applied Optics,
20(2): 177-184.
323
Smith, S. D. (1980). Wind stress and heat flux over the ocean in gale force winds. Journal
of Physical Oceanography, 10, 709-726
Smith, S. D. (1989). Coefficients for sea surface wind stress, heat flux, and wind profiles as
a function of wind speed and temperature. Journal of Geophysics Research, 93, 15467-15472
Steele, J. A. (1974). The structure of marine ecosystem. Oxford, Blackwell: p74-91.
Steele, J.M. (1965) Notes on some theoretical problems in production ecology. In:(C.R.
Goldman ed.) Primary Production in Aquatic Environments. University of California Press.
Berkeley
Steinberger, N., and Hondzo, M. (1999). Experimental study of the hydrodynamic control
of diffusional mass transfer at the sediment-water interface. Journal of Environmental
Engineering, 125(2): 192-200
Stoker, Y.E. (1986). Water quality of the Charlotte Harbor estuarine system, Florida,
November 1982 through October 1984. U.S. Geological Survey Open-File Report 85-563:
213.
Stoker, Y.E. (1992). Salinity distribution and variation with fresh-water inflow and tide, and
potential changes in salinity due to altered freshwater inflow in the Charlotte Harbor
estuarine system, Florida. U.S. Geological Survey Water-Resources Investigations Report
92-4062: 30.
Stull, R. B. (1988). An Introduction to Boundary Layer Meteorology. Kluwer Academic
Publishers, Netherlands, 660pp.
Sun, D. and Sheng, Y.P. (2001). Modeling suspended sediment transport under combined
wave current interactions in Indian River Lagoon. PhD dissertation, University of Florida,
Gainesville, Florida.
Taylor, P., Gulev, S., Barnier, B.,Bradiey, F.,Charlock, T.,Gleckler, P.,Kubota, M.,
Kutsuwada, M.,Legler, D., Lindau, R., Rothrock, D., Schulz, J.da-Silva, A., Sterl, A., and
White, G. (2000). Intercomparison and validation of ocean-atmosphere energy flux fields.
Final Report of the joint WCRP/SCOR working group on air-sea fluxes, WCRP- 112, WMO/
TD-No. 1036, 303 S
Thomann, R.V. (1992). Section 14.1. in Technical guidance manual for performing waste
load allocations. Book III: Estuaries-Part 4, Critical review of coastal embayment and
estuarine waste load allocation modeling. U.S. EPA Technical Report EPA-823-R-92-005.
Washington, D.C.
Thomann, R. V., and Fitzpatrick, J.F. (1982). Calculation and verification of a amthematical
model of the eutrophication of the Potomac Estuary. Prepared for the Department of
Environmental Services, Government of the Distric of Columbia.
324
Thompson, J.F. (1982). Numerical Grid Generation, chapter General Curvilinear Coordinate
Systems, pages 1-30. Elsevier, Amsterdam.
Thompson, J.F., Warsi, Z.U.A. and Mastin, C.W. (1985). Numerical Grid Generation:
Foundations and Applications. North-Holland. New York
Weast (1977). Handbook of chemistry and physics, 58th ed., Table F-200
Witkowski, J. and Jaffe, P.R. (1987). Kinetic and equlibrium formulations in modeling of
hydrophobic organic chemicals in surface and ground waters. System Analysis in Water
Quality Management (M.B. Beck, ed.), Advences in Water Pollution Control- Aseries of
Conferences Sponsored by IAWPRC.
Wolfe, S.H., and Drew R. D., eds. (1990). An ecological characterization of the Tampa Bay
watershed. U.S. Fish Wildl. Serv. Biol. Rep. 90: 20.
Yassuda,E.A. andSheng, Y.P. (1996). Integrated modeling of Tampa Bay Estuarine System.
PhD dissertation, University of Florida, Gainesville, Florida.
Zison, S., Mills, W.B., Deimer, D., and Chen, C.W. (1978). Rates, constants and kinetic
formulations in surface water quality modeling. Prepared by Tetra Tech, Inc., Lafayette, CA
for Environmental Research Laboratory, U.S. Environmental Protection Agency, Athen, GA.
EPA-600/3-78-105:335.
BIOGRAPHICAL SKETCH
Kijin Park was born in Pusan, the largest harbor of Korea, on the 25th of October. He
received the Bachelor of Science degree and M.S. degree in department of Marine Science
at the Pusan National University in February, 1991 and 1994, respectively. After finished
M.S. degree, he had worked as research scientist at Korea Oceanography Research &
Development Institute (KORDI) and Korea Meteorological Administration (KMA) until
June, 1997. The following summer he left for Gainesville, Florida , and began his graduate
studies in coastal and oceanographic engineering at the University of Florida.
325
I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.
Y. Pe^fSneng, Chair^
Professor of Civil and Coastal Engineering
I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.
Robert G. Dean
Graduate Research Professor of Civil and
Coastal Engineering
I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.
Robert J. Thiek£
Assistance Professor of Civil and Coastal
Engineering
I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of PhilcrStoph\
Louis H. Motz
Associate Professor of Civil and Coastal
Engineering
I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosopl;
K. Ramesh Reddy
Graduate Research Professor of Soil and
Water Science
This dissertation was submitted to the Graduate Faculty of the College of
Engineering and to the Graduate School and was accepted as partial fulfillment of the
requirements for the degree of Doctor of Philosophy.
) Ajgw^l t/j^-t^-yv-cW^
August, 2004
Pramod P. Khargonekar
Dean, College of Engineering
Kenneth J. Gerhardt
Interim Dean, Graduate School
,PA3&>
UNIVERSITY OF FLORIDA
3 1262 08554 5670
Live Music Archive
Librivox Free Audio
Metropolitan Museum
Cleveland Museum of Art
Internet Arcade
Console Living Room
Books to Borrow
Open Library
TV News
Understanding 9/11