SIMULATION AND ANALYSIS OF SEDIMENT RESUSPENSION OBSERVED IN OLD TAMPA BAY, FLORIDA
David H. Schoellhamer and
Y. Peter Sheng
REPORT DOCUMENTATION PAGE
. .3. Recipients Accessiono.
4. Title and Subtitle 5. Report Date
Simulation and Analysis of Sediment Resuspension April 1993
observed in Old Tampa Bay, Florida 6.
7. h&utor(s) S. Performing Orgenization Report No.
David H. Schoellhamer and Y. Peter Sheng UFL/COEL-TR/091
9. Performing Organization Bais and Address 10. Pr ject/Task/ork Unit No.
University of Florida
Obastal & Oceanographic Engineering 11. Contract or Grant Mo.
336 Weil Hall
Gainesville, FL 32611 1y Report
12. Sponsoring Orgenizetion Nam end Address
15. Supplementary Notes
Estuarine sediments may limit light availability for photosynthesis, supply nutrients to the water column, and affect the fate of contaminants in estuaries, including Old Tampa Bay, Florida. These adverse impacts are enhanced when bottom sediments are resuspended into the water column. Sediment resuspension was
intermittently monitored in Old Tampa Bay from 1988 to 1990. The data indicates that net sediment resuspension is caused by depth-transitional wind-waves that are generated by strong and sustained winds associated with storm systems and not tidal currents. The bottom roughness regime in Old Tampa Bay is transitional between the smooth and rough limits.
A vertical one-dimensional numerical model was modified and used to help interpret the Old Tampa Bay sediment resuspension data. The model was modified to simulate low Reynolds number flows, multiple wave frequencies, variable bottom roughness regimes, and spatially-averaged grain shear stress. The model successfully simulated steady flow profiles with laminar, transitional, and turbulent regions and turbulent dissipation in the marine surface layer. Simulated shear stresses are in agreement with observed critical conditions for sediment motion for combined wave and current motion on the continental shelf. Wave spectra observed in Old Tampa Bay and simulations of the sediment resuspension data indicate that formation of large
(continued on back)
17. Originator's ey Words 18. Availability Statement
Sediment Resuspension, Numerical Simulation, Tampa Bay
19. U. S. Security Classif. of the Report 20. U. S. Security Classit. of This Page 21. No. of Pages 22. Price
Unclassified Unclassified 262
SIMULATION AND ANALYSIS OF SEDIMENT RESUSPENSION
OBSERVED IN OLD TAMPA BAY, FLORIDA
David H. Schoellhamer and
Y. Peter Sheng
Coastal & Oceanographic Engineering Department
University of Florida Gainesville, FL 32611
TABLE OF CONTENTS
LIST OF TABLES..................................................... vi
LIST OF FIGURES................................................... viii
KEY TO SYMBOLS..................................................... x
1 INTRODUCTION................................................. 1
Significance of Estuarine Sediments......................... 4
Sediment Transport Processes................................. 8
Initiation of Motion of Bed Sediments .......................8
Bed Load Transport........................................ 15
Erosion and Bottom Shear Stress........................... 16
Suspended Load Transport.................................. 24
Flocculation and Aggregation.............................. 26
Settling and Deposition................................... 27
Other Bed and Near-bed Factors............................ 31
Field Studies Related to Estuarine Sediment Transport ... 35
Numerical Models Applicable to
Estuarine Sediment Transport.............................. 43
Relation of this Research to Previous Studies ................49
2 SEDIMENT RESUSPENSION DATA COLLECTION METHODOLOGY ...........52
Study Area.................................................. 52
Sediment Resuspension Monitoring Sites...................... 54
Old Tampa Bay Instrumentation Platform...................... 56
Calibration and Output of Electromagnetic Current Meters 63
Response Threshold and Biological Interference of
OBS Sensors............................................... 66
Water Sample Collection..................................... 68
Instrument Deployment Strategies............................ 69
3 SEDIMENT RESUSPENSION DATA AND ANALYSIS..................... 70
March 1990 Storm............................................ 73
November 1990 Storm......................................... 84
Tropical Storm Keith........................................ 90
Tropical Storm Marco........................................ 94
Implications for Numerical Modeling......................... 96
4 NUMERICAL MODEL........................... .................. 98
Momentum and Transport Equations............................ 102
Turbulence Closure........................................... 106
Nondimensional Equations.................................... 112
Steady State Conditions..................................... 116
Wave-Induced Pressure Gradients............................. 117
Bottom Shear Stress and Erosion............................. 119
Suspended-Sediment Stratification........................... 122
5 NUMERICAL SIMULATIONS OF THE MARINE SURFACE LAYER
AND CRITICAL SHEAR STRESSES ON CONTINENTAL SHELVES ........123
Simulation of Turbulence in the Marine Surface Layer ........123 Critical Shear Stresses Observed on Continental Shelves ... 126
6 OLD TAMPA BAY NUMERICAL SIMULATION RESULTS.................. 138
Steady Flow Simulation...................................... 138
Reproduction of Energy Spectra of Observed Currents .........142 Simulated Shear Stresses.................................... 144
Old Tampa Bay Suspended-Sediment Simulation Procedure .......154
Old Tampa Bay November 1990 Suspended-Sediment
Calibration Simulation.................................... 161
Old Tampa Bay March 1990 Suspended-Sediment
Validation Simulation..................................... 166
Old Tampa Bay March 1990 Suspended-Sediment
Improved Simulation....................................... 171
Old Tampa Bay November 1990 Sensitivity Simulations .........175
7 SUMMARY AND CONCLUSIONS..................................... 181
A OCM1D FINITE-DIFFERENCED EQUATIONS AND TURBULENCE
CLOSURE ALGORITHMS........................................ 188
Finite-Differenced Equations for Momentum and
Suspended Sediment........................................ 188
Turbulence Closure Algorithms............................... 192
BIOGRAPHICAL SKETCH................................................ 215
LIST OF TABLES
Table 1-1, Previous studies that are related to this research
and include an estuarine environment, wind waves,
field measurements of suspended-sediment
concentrations, or a numerical vertical sediment
transport model........................................ 50
3-1, Old Tampa Bay platform instrumentation deployments 71
3-2, Old Tampa Bay platform data, March 8, 1990 ..............75
3-3, Correlation coefficients for various wave properties
and bottom shear stresses with suspended-solids
concentrations at the Old Tampa Bay platform,
March 1990............................................. 78
3-4, Old Tampa Bay platform data, November 30
December 1, 1990....................................... 86
3-5, Correlation coefficients for various wave properties
and bottom shear stresses with suspended-solids
concentrations at the Old Tampa Bay platform,
November 1990.......................................... 88
4-1, Modifications made to the 1986 version of OCM1D
(Sheng 1986) for this research......................... 101
4-2, Comparison of the equilibrium closure and TKE
closure methods for turbulent transport ................113
5-1, Mean current speed, maximum wave orbital velocity,
wave period, and critical shear stresses from a
wave-current model (Drake and Cacchione 1986) and
the OCM1D model........................................ 128
5-2, Mean current speed, maximum wave orbital velocity,
wave period, angle between the mean current and
waves, and critical shear stresses from a
wave-current model (Larsen et al. 1981) and
the OCM1D model........................................ 133
6-1, Total and grain shear stress distribution and
critical stresses, 1800 hours March 8, 1990 ............153
6-2, Old Tampa Bay suspended-sediment simulations ...........157
6-3, Values of the calibration coefficients determined by
other studies and determined for the November 1990 calibration simulation and the March 1990 improved
simulation ........................................... 173
6-4, Mean percent differences of simulated suspended-solids
concentrations 70 and 183 cm above the bed caused by
20% changes in the calibration coefficients,
November 1990 sensitivity simulations ................ 176
LIST OF FIGURES
Figure 1-1, Shields critical shear stress diagram ..................12
1-2, Roughness function B in terms of Reynolds number,
from Schlichting (1969)........... .................... 19
2-1, Old Tampa Bay study area.............................. 53
2-2, Size distribution of bottom sediment at the
Old Tampa Bay platform................................ 57
2-3, Old Tampa Bay platform................................ 58
2-4, Suspended-solids concentration and median OBS
sensor output 70 cm above the bed at the Old Tampa
Bay platform, March 1990.............................. 67
3-1, Measured and calculated quantities at the Old Tampa
Bay platform, March 1990.............................. 74
3-2, Measured and calculated quantities at the Old Tampa
Bay platform, November 1990........................... 85
3-3, Mean water depth, mean current speed, and maximum
bottom orbital velocity at the Old Tampa Bay
platform during tropical storm Keith,
November 1988......................................... 91
3-4, Wind speed, wind vector azimuth, and
suspended-solids concentration at the Old Tampa Bay
platform during tropical storm Marco, October 1990.. 95
5-1, Measured, simulated, and theoretical turbulent
dissipation in the marine surface layer, measurements
reported by Soloviev et. al (1988).................... 124
5-2, Extended Shields diagram for continental shelf
data, shear stresses from wave-current models (WC)
and OCM1D............................................. 130
6-1, Computational grid for 45 layers, 1.15 neighboring
layer height ratio, and a 399 cm domain height ........140
6-2, Comparison of Reichardt and simulated velocity
6-3, Comparison of turbulence macroscale from the
dynamic equation and the integral constraints .........143
6-4, Raw energy spectra computed from measured and
simulated velocities, 1500 hours November 30, 1990.. 145
6-5, Spatial convergence of total and
grain shear stress.................................... 147
6-6, Maximum grain shear stress calculated by OCM1D and
from Engelund's experimental results ..................149
6-7, Simulated total and grain shear stresses,
1800 hours March 8, 1990......... ..................... 150
6-8, Simulated grain shear stress,
1800 hours March 8, 1990.............................. 152
6-9, Maximum total bottom shear stress from the
OCM1D model and the Grant and Madsen (1979)
wave-current model, November 1990..................... 155
6-10, Simulated and measured suspended-solids
concentrations 70 cm above the bed,
November 1990 calibration simulation .................162
6-11, Simulated and measured suspended-solids
concentrations 183 cm above the bed,
November 1990 calibration simulation .................163
6-12, Simulated and measured suspended-solids
concentrations 24 cm above the bed,
March 1990 validation and improved simulations .......168
6-13, Simulated and measured suspended-solids
concentrations 70 cm above the bed,
March 1990 validation and improved simulations .......169
6-14, Simulated and measured suspended-solids
concentrations 183 cm above the bed,
March 1990 validation and improved simulations .......170
6-15, Sensitivity of November 1990 calibration simulation
results to erosion rate exponent q................... 177
6-16, Sensitivity of November 1990 calibration simulation
results to erosion rate coefficient a ................178
6-17, Sensitivity of November 1990 calibration simulation
results to settling velocity w5...................... 180
A-1, Coordinate axes definition and grid structure for
program OCM1D......................................... 189
KEY TO SYMBOLS
A invariant constant for turbulence algorithm (0.75, eqn. 4-26)
Ab orbital amplitude just above the boundary layer (eqn. 1-14)
A eddy viscosity (eqn. 1-16)
nondimensional eddy viscosity (eqn. 4-42) a wave amplitude (eqn. 1-12)
small reference elevation above the bed (eqn. 1-18)
Fourier series coefficient (eqn. 2-3)
invariant constant for turbulence algorithm (3, eqn. 4-23) a corrected Fourier series coefficient (eqn. 2-9)
B roughness function (eqn. 1-5)
b intercept of line that fits a segment of fig. 1-2 (eqn. 1-6)
Fourier series coefficient (eqn. 2-3)
invariant constant for turbulence algorithm (0.125, eqn. 4-23) b corrected Fourier series coefficient (eqn. 2-10)
C well-mixed suspended-sediment concentration (eqn. 1-19)
C coefficient for integral constraint on A (eqn. 4-32)
c mean suspended-sediment concentration (eqn. 1-15)
magnitude in frequency domain (eqn. 2-4)
nondimensional suspended-sediment concentration (eqn. 4-42)
ca suspended-sediment concentration at elevation a (eqn. 1-18) cc corrected magnitude in frequency domain (eqn. 2-7) cm suspended-sediment mass concentration (eqn. 4-62) cu suspended-sediment concentration at upper boundary (eqn. 4-19)
nondimensional c (eqn. 4-52)
D rate of deposition (eqn. 1-19)
d particle diameter for which n percent of sediments are finer
d sediment particle diameter (eqn. 1-1)
E erosion rate (eqn. 1-3)
Erosion rate (equ. 1-3)
nondimensional erosion rate (eqn. 4-42) Ez Ekman number (eqn. 4-43) F Froude number (eqn. 4-43)
neighboring layer height ratio (eqn. 5-2) f friction factor (eqn. 1-12)
Coriolis coefficient (eqn. 4-4)
G filter gain for current meter, function of w (eqn. 2-1)
g gravitational acceleration (fig. 1-1)
H elevation of model domain (eqn. 4-42)
nondimensional elevation of model domain (eqn. 4-42) h water depth (eqn. 1-12)
i counter and exponent (eqn. 5-2)
K pressure transfer function (eqn. 4-60)
K eddy diffusivity (eqn. 1-16)
nondimensional eddy diffusivity (eqn. 4-42) k wave number (eqn. 1-12)
molecular diffusion (eqn. 4-26) kb bottom roughness (eqn. 1-14) ks height of bottom roughness elements (eqn. 1-5) M erosion rate constant (eqn. 1-3)
N number of data points for fast Fourier transform (eqn. 2-3)
Brunt-Vaisala frequency (eqn. 4-31)
number of layers (eqn. 5-2) n counter (eqn. 2-3)
P Prandtl number (eqn. 4-43)
p pressure (eqn. 4-2)
p pressure at the water surface (eqn. 4-6)
nondimensional pressure at the water surface (eqn. 4-42) Q nondimensional quantity (eqn. A-8)
q turbulent fluctuating velocity (eqn. 4-23)
nondimensional turbulent fluctuating velocity (eqn. 4-42) R roughness Reynolds number u ks/v (fig. 1-2)
Re Reynolds number (eqn. 4-43) Ri Richardson number (eqn. A-11) Rx horizontal Rossby number (eqn. 4-43) R vertical Rossby number (eqn. 4-43)
R, boundary Reynolds number U*ds/v (eqn. 1-1) RC electronic filter time constant for current meter (eqn. 2-1) s slope of line that fits a segment of fig. 1-2 (eqn. 1-6)
invariant constant for turbulence algorithm (1.8, eqn. 4-27) T wave period (table 5-1)
t time (eqn. 1-15)
nondimensional time (eqn. 4-42)
U1/10 mean of the highest 1/10 ub (table 5-2) U100 mean current speed 100 cm above bed (table 5-1) U maximum wave orbital velocity 20 cm above bed (table 5-1)
U, shear velocity (eqn. 1-1) u horizontal velocity at elevation z (eqn. 1-5)
mean velocity of sediment particles (eqn. 1-15)
mean velocity component (eqn. 4-1)
nondimensional mean velocity component (eqn. 4-42) ub maximum bottom orbital velocity (eqn. 1-11) u, shear velocity (eqn. 1-5) v mean horizontal velocity component (eqn. 4-4)
nondimensional mean velocity component (eqn. 4-42)
v invariant constant for turbulence algorithm (0.3, eqn. 4-23)
w mean vertical velocity component (eqn. 4-4)
w s terminal settling velocity (eqn. 1-17)
nondimensional settling velocity (eqn. 4-42) x coordinate axis (eqn. 1-15)
nondimensional horizontal coordinate axis (eqn. 4-42) y horizontal coordinate axis (eqn. 4-4)
nondimensional horizontal coordinate axis (eqn. 4-42) Z Rouse number ws/(Psu,) (eqn. 1-18)
z elevation above bed (eqn. 1-5)
vertical coordinate axis (eqn. 1-17)
vertical coordinate (eqn. 4-4, fig. A-1)
zv vertical coordinate of velocity measurement (eqn. 4-61)
z nondimensional quantity u.z/v (eqn. 1-10)
z 0 elevation with zero velocity (eqn. 1-9)
z nondimensional quantity uz o/v (eqn. 1-10)
a erosion rate coefficient (eqn. 1-4)
invariant constant for turbulence algorithm (0.75, eqn. 4-26) p K/A, the inverse of the turbulent Schmidt number (eqn. 1-16) Y specific weight of fluid (eqn. 1-2)
Ys specific weight of sediment (eqn. 1-2) At time interval of data (eqn. 2-3)
nondimensional simulation time step (eqn. A-1)
simulation time step (eqn. 4-59) Az1 height of bottom layer (eqn. 5-2) Aak nondimensional layer elevation (eqn. A-1) Ac-k nondimensional distance between grid points k and k-I (eqn. A-1) Aa+k nondimensional distance between grid points k and k-i (eqn. A-1) Ao angular frequency increment 2w/(NAt) (eqn. 2-3)
6 Kronecker delta (eqn. 4-23)
6 distance for integral constraint on A (eqn. 4-32)
S phase (eqn. 2-4)
alternating tensor (eqn. 4-2)
energy dissipation rate (eqn. 5-1) e c corrected phase (eqn. 2-8)
nondimensional quantity (eqn. 1-13)
S erosion rate exponent (eqn. 1-4) e production term (eqn. 4-3)
nondimensional production term (eqn. 4-42) X von Karman's constant (eqn. 1-5)
A turbulence macroscale (eqn. 4-23)
nondimensional turbulence macroscale (eqn. 4-42) A Taylor microscale (eqn. 4-28)
v kinematic viscosity (eqn. 1-1)
nondimensional kinematic viscosity (eqn. 4-42)
surface displacement from the mean water depth (eqn. 4-6)
nondimensional surface displacement (eqn. 4-42) p fluid density (eqn. 1-1)
nondimensional fluid density (eqn. 4-42) po reference fluid density (eqn. 4-2) Ps sediment density (eqn. 4-62) Pw water density (eqn. 4-62) a nondimensional vertical coordinate (eqn. 4-42)
r bottom shear stress (eqn. 1-3)
Tb total bottom shear stress (table 5-1) c critical shear stress for erosion (eqn. 1-3)
cr critical shear stress for deposition (eqn. 1-19) min minimum r that maintains sediment in suspension (eqn. 1-19)
0 bottom shear stress (eqn. 1-2)
vs grain shear stress (table 5-1) w maximum bottom shear stress in oscillatory flow (eqn. 1-11)
r x component of wind shear stress at free surface (eqn. 4-14)
nondimensional x component of wind shear stress (eqn. 4-47) r y component of wind shear stress at free surface (eqn. 4-15)
WY nondimensional y component of wind shear stress (eqn. 4-48)
7 x component of total bottom shear stress (eqn. 4-20)
nondimensional x component of total shear stress (eqn. 4-53) Sy component of total bottom shear stress (eqn. 4-21) y nondimensional y component of total shear stress (eqn. 4-54) r, dimensionless shear stress (Shields parameter, eqn 1-2)
filter phase delay for current meter, a function of w (eqn. 2-2)
angle between the mean current and wave direction (table 5-2)
angular velocity of the Earth (eqn. 4-2)
w angular wave frequency (eqn. 1-12)
angular frequency of periodic input signal to filter (eqn. 2-1) Subscripts:
k layer number (eqn. A-l)
r reference value (eqn. 4-42)
n time step (eqn. A-l)
* nondimensional value (eqn. 4-42)
S turbulent fluctuation
Estuarine sediments may limit light availability for
photosynthesis, supply nutrients to the water column, and affect the fate of contaminants in estuaries, including Old Tampa Bay, Florida. These adverse impacts are enhanced when bottom sediments are resuspended into the water column. Sediment resuspension was intermittently monitored in Old Tampa Bay from 1988 to 1990. The data indicates that net sediment resuspension is caused by depth-transitional wind-waves that are generated by strong and sustained winds associated with storm systems and not tidal currents. The bottom roughness regime in Old Tampa Bay is transitional between the smooth and rough limits.
A vertical one-dimensional numerical model was modified and used to help interpret the Old Tampa Bay sediment resuspension data. The model was modified to simulate low Reynolds number flows, multiple wave frequencies, variable bottom roughness regimes, and spatially-averaged grain shear stress. The model successfully simulated steady flow profiles with laminar, transitional, and turbulent regions and turbulent dissipation in the marine surface layer. Simulated shear stresses are in agreement with observed critical conditions for sediment motion for combined wave and current motion on the continental shelf. Wave spectra observed in Old Tampa Bay and simulations of the sediment resuspension data indicate that formation of large aggregates is an important process that controls settling in Old Tampa Bay. A simple aggregation algorithm
was added to the model and the calibrated coefficients are reasonable compared to values from other studies. Only the finer particles are probably intermittently transported as bed load. Simulation results indicate that the bottom sediments were more erodible in March 1990 than November 1990. Reduced biological binding of the fine bed sediments and increased storm activity may increase bottom sediment erodibility in March 1990.
Estuaries are transition zones between riverine and marine
environments. Potential sources of sediment particles for an estuary include rivers, net sediment flux from the marine environment, overland runoff, and anthropogenic point sources. Sediment particles are commonly trapped and deposited in the deeper parts of an estuary. The bed sediment affects the overall health of an estuary in several ways. Bed sediment, especially fine sediment, can be resuspended up into the water column where it may reduce the amount of light penetrating the water column, may act as a source for constituents adsorbed onto the sediment, and may be transported to undesirable locations. The reduction of light in the water column may adversely affect biological communities. Adsorbed constituents that can be released to the water column during suspension and possibly while on the bed include nutrients which may contribute to eutrophication of the estuary, heavy metals, pesticides, and organic carbons that may decrease the productivity of the estuary. Resuspended sediment may be transported throughout the estuary and spread the adverse effects and possibly become deposited in undesirable locations, such as shipping channels, turning basins, and marinas. Thus, the overall health of an estuarine environment is partially dependent upon the resuspension, transport, and deposition of sediment.
Numerical models can be used to study and predict sediment dynamics in an estuary. An accurate numerical model must include
algorithms that represent significant hydrodynamic and sediment transport processes which may be identified from comprehensive field data. Governmental regulators could use a sediment model to help predict the effect of proposed anthropogenic alterations to an estuary on light attenuation, transport and fate of toxic substances, and sedimentation. Potentially adverse alterations include increased wastewater discharge, increased stormwater runoff, dredging, dredge material disposal, and wetland destruction. Because sediment particles are negatively buoyant and settle, accurate sediment models must consider vertical sediment dynamics. Two obstacles, however, limit application of sediment models. Field data are needed to calibrate and validate estuarine sediment models, and the bottom boundary conditions for the sediment must be specified.
Improved understanding and simulation of estuarine sediment
processes are dependent upon reliable field data. The importance of the processes that potentially control sediment dynamics must be understood and included in a realistic model. For example, if windwaves are an important sediment resuspension mechanism, then a model that ignores water motions at wind-wave frequencies can not realistically simulate estuarine sediment transport. Laboratory studies can be used to improve understanding of sediment transport process, but field conditions are often difficult to recreate in the laboratory. Reconstruction of realistic sediment beds in a laboratory is difficult, especially when the natural sediments are biologically active. A field data collection program has far less control than found in a laboratory, but the data and insights gained in the field are directly applicable to the estuary being studied. Calibration is the selection of model parameters that permit the model to accurately
simulate field data. Once selected, validation may be performed on an independent data set to prove that the model algorithms and model parameters can be used to make reliable predictions. Field data is therefore required to develop a predictive model in which managers, scientists, and the public can have confidence.
Sediment transport processes at the interface of the water column and bottom sediments must be accurately simulated in a numerical model because the bed is an omnipresent potential source and sink of suspended sediment. A particle resting on the bed will move (erode) when the lift force generated by the hydrodynamics is sufficiently large. If a particle moves by rolling, sliding, or saltating (jumping), then it contributes to the bed load. A mobilized particle may also become suspended (or resuspended) in the water column. Interaction of suspended particles may be caused by salt flocculation of clay minerals and formation of large aggregates of organically bound inorganic particles. Deposition is the process that returns particles to the bed. The sedimentary processes of initiation of particle motion, bed load, and deposition all occur at or near the interface of the sediment bed and the water column. Other near-bed factors which may affect hydrodynamics and sediment transport include bed forms, armoring, porous beds, suspended-sediment stratification, and biological activity.
The purpose of this research was to improve the understanding and numerical modeling of sediment resuspension and the vertical transport of resuspended sediment in the shallow estuarine environment. Field data and a numerical model were used to accomplish this task. From 1988 to 1990, the author collected sediment resuspension data during potential periods of resuspension in Old Tampa Bay, a shallow estuary
on the west-central coast of Florida. These field data were analyzed by the author to determine the sediment resuspension mechanisms in Old Tampa Bay. The author modified and used a vertical one-dimensional model to simulate the Old Tampa Bay data and other data. The numerical model was used as a tool to help analyze the Old Tampa Bay data and to help determine significant sediment transport processes in Old Tampa Bay.
The remainder of this chapter discusses existing literature on the significance of estuarine sediments, sediment transport processes, field studies related to estuarine sediment transport, numerical models applicable to estuarine sediment transport, and the relation between previous studies and this research. The data collection methodology is discussed in chapter 2, and the data are presented and analyzed in chapter 3. The numerical model is described in chapter 4. Numerical simulations of the marine surface layer and critical shear stresses on continental shelves were conducted to test modifications made to the model during this research, and these simulation results are presented and discussed in chapter 5. Simulations of suspended-solids concentrations in Old Tampa Bay are presented and discussed in chapter
6. Conclusions of this study are summarized in chapter 7.
Significance of Estuarine Sediments
Bottom sediments are an omnipresent factor that affect the water quality and biological productivity of an estuary. Potential sources of sediment particles for an estuary include rivers, net sediment flux from the marine environment, overland runoff, and anthropogenic point sources. Sediment particles are commonly trapped and deposited in the deeper or vegetated parts of the estuary. Bed sediments provide the substrate for benthic organisms, seagrasses, and marshes, and chemical
exchange processes occur between the bed sediments and the water column. Bed sediments, especially fine sediments, can be resuspended up into the water column where they may 1) increase nutrient concentrations, 2) transport trace metals, 3) influence the mortality, life cycle, food supply, and photosynthesis of estuarine species, and 4) may move and settle in ports or marshes.
Bed sediments chemically interact with the overlying water column and benthic biological communities. Diffusive fluxes between the bed sediments and water column commonly remove dissolved oxygen from the water column (Hinton and Whittemore 1991, Svensson and Rahm 1991) and recycle nutrients to the water column (Callender and Hammond 1982, Hammond et al. 1985, Simon 1988, Ullman and Aller 1989). In addition, sediment geochemistry in seagrass beds determines the limiting nutrient for seagrass growth (Short 1987).
The release of nutrients from sediments to the water column is
enhanced during resuspension events. Increased nutrient concentrations in the water column that were caused by resuspension events have been observed by Gabrielson and Lukatelich (1985) during wind related sediment resuspension events in the Peel-Harvey estuarine system in Australia, by Fanning et al. (1982) during storms on the continental shelf of the Gulf of Mexico, and by Schwing et al. (1990) after destabilization of bottom sediments by a seiche in Monterey Bay, California. Grant and Bathmann (1987) found that bacterial mats deposit sulfur on surficial bottom sediments and that resuspension is an important mechanism for returning sulfur to the water column. Phytoplankton and organic detritus resting on the bottom can also be resuspended and impact estuarine productivity (Roman and Tenore 1978, Gabrielson and Lukatelich 1985). The supply of Radon 222, a tracer, to
the water column from resuspension flux and diffusion flux is about the same in the Hudson River estuary (Hammond et al. 1977). Simon (1989) estimates that one resuspension event in the Potomac River that lasts minutes can add as much ammonium to the water column as the diffusive flux can in 5 to 1000 days, depending upon the site. Laboratory experiments show that typical water column concentrations of particulate nutrients would double in a few hours during a resuspension event (Wainright 1990) and that biological growth is increased when resuspended material is added to microcosms (Wainright 1987).
In addition to nutrients, trace metals and other contaminants may be adsorbed to sediment particles and these contaminants are detrimental to the biological health of many estuaries. For example, metals are partitioned between adsorbed and dissolved phases, so the transport of metals is related to sediment transport (Dolan and Bierman 1982, Li et al. 1984, Horowitz 1985). In South San Francisco Bay, availability of trace metals may be a factor that limits growth of some phytoplankton species, sorption processes influence dissolved concentrations of metals, and sorption processes vary among specific metals (Kuwabara et al. 1989). Sediment concentrations of trace metals, PCBs, pesticides, or polynuclear aromatic hydrocarbons exceed the median concentration associated with biological effects in estuaries in Alaska, California, Connecticut, Florida, Hawaii, Maryland, Massachusetts, New Jersey, New York, Oregon, Texas, and Washington (Long and Morgan 1990).
The mortality, food supply, and life cycle of some estuarine
species may be affected by suspended sediments and sediment transport processes. For species restricted to the benthos, mortality may be increased by resuspension or burial and sediment transport may regulate
the food supply to both suspension and deposit feeders (Nowell et al. 1987). Eggs, cysts, and spores of many zooplankton and phytoplankton species reside in bottom sediments and erosion may inject them into the water column where they may hatch. Circumstantial evidence indicates that this process may contribute to red tide outbreaks (Nowell et al. 1987). Sellner et al. (1987) found that increased suspended-sediment concentrations reduced the survival rate of newborn larval copepods in Chesapeake Bay. In addition, development and reproduction of survivors was inhibited.
Suspended sediments reduce the sunlight available for
photosynthesis. Smaller particles are more efficient light attenuators (Baker and Lavelle 1984, Campbell and Spinard 1987). McPherson and Miller (1987) found that non-chlorophyll suspended material is the most important component of light attenuation in Charlotte Harbor, Florida. Inorganic suspended material is the dominant cause of light attenuation in several New Zealand estuaries (Vant 1990) and sediment resuspension by tidal currents and wind waves is an important cause of attenuation (Vant 1991). In the coastal waters of northwest Africa, light attenuation is greatest in nearshore waters where sediment concentrations are greatest, compared to offshore waters, and nearshore light attenuation reduces phytoplankton growth (Smith 1982). The reduction of light in the water column reduces seagrass photosynthesis and the maximum depth at which seagrasses can grow (Dennison 1987).
Resuspended sediments may move throughout the estuary, depending upon the circulation, and, in addition to possibly spreading the adverse effects already mentioned, may deposit in ports or marshes. Large man-made basins that serve as ports and marinas are commonly depositional environments that require costly maintenance dredging
(Granat 1987, Kobayashi 1987, Headland 1991). Estuarine sediments may also deposit in marshes where vegetation and benthic algae impede water motion and resuspension and enhance deposition, formation, and maintenance of an important habitat (Ward et al. 1984, Krone 1985, Huh et al. 1991).
Sediment Transport Processes
Sediment transport processes differ somewhat depending on whether the sediment is noncohesive or cohesive. Noncohesive sediment particles do not interact electro-chemically with other particles, and cohesive sediment particles interact electro-chemically with other cohesive particles. Sediments with a diameter larger than 20 Am are generally noncohesive. Thus, gravel, sand, and coarse silts are noncohesive. Cohesion increases as particle size decreases below 20 um (Migniot 1968), and clay minerals are generally cohesive. Transport of noncohesive sediments is controlled by the processes of initiation of particle motion, bed load transport, suspended load transport, and deposition. Cohesive sediments differ in that they are not transported as bed load, and interparticle electro-chemical forces may cause flocculation in brackish estuarine waters. Both cohesive and noncohesive suspended inorganic particles may adhere to large organically bound aggregates that can deposit rapidly in an estuary. Bed and near-bed factors that may affect hydrodynamics and sediment transport include bed forms, bed armoring, suspended-sediment stratification, fluid-mud, porous beds, and biological activity. Initiation of Motion of Bed Sediments
The horizontal transport of noncohesive sediment as bed load and suspended bed-material load is dependent upon the initiation of motion of stationary particles in the sediment bed. Particles that roll,
slide, or saltate along or near the bed are part of the bed load, and particles that are lifted into suspension are part of the suspended load. Initiation of particle motion has been reviewed by the Task Committee on Preparation of Sedimentation Manual (1966), Vanoni (1975), Miller et al. (1977), Simons and Senturk (1977), and Lavelle and Mofjeld (1987a, 1987b).
A noncohesive particle lying on a sediment bed, for which fluid is flowing above, will be acted upon by the hydrodynamic forces of lift and drag in addition to gravity and normal forces from adjacent touching particles. Vanoni (1975, pp. 92-93), Simons and Senturk (1977, pp. 400-407), Yalin (1977), and Wiberg and Smith (1987) describe the forces acting on a bed particle in detail. The formulation of the hydrodynamic forces will differ depending on whether the flow over the particle is laminar or turbulent and whether the bed is composed of uniform or heterogeneous particles. If the hydrodynamic forces exceed a threshold or critical value, then the particle will move. Particle motion may also be initiated by organisms that disturb the bottom sediments and by trawling in coastal waters (Churchill 1989).
Because the exact geometry, size, and shape of every sediment particle in a bed can not be determined, practical analysis of the problem of initiation of particle motion requires assuming that the bed is a continuum of particles instead of a large quantity of discrete particles. Noncohesive sediment particles are usually assumed to be spherical. The particle sizes (measured by sediment particle diameter d s) that are present in an actual bed will vary vertically and horizontally, but it is commonly represented as one or more sizes for which a certain percentage N of the bed material is finer (denoted as
dN). The physical properties of the sediment particles at a fixed location also will vary with time as particles are transported.
In addition to sediment particle properties, the flow field will vary spatially and temporally. The shear force exerted by the flow on the bed is commonly used to indicate the magnitude of the hydrodynamic force on the bed. Because most natural flows are turbulent and turbulent flows fluctuate in space and time, the bottom shear stress at a fixed point will vary about the mean bottom shear stress. The bottom shear stress will also vary spatially, especially if bed forms are present. For example, for three-dimensional ripples, Ikeda and Asaeda (1983) found that sediment is eroded from the side slopes of longitudinally trailing ridges by lee side eddies and that sediment entrainment is correlated with intermittent bursts of the lee side eddy.
If the bed is assumed to be a continuum of particles, the inception of particle motion is a stochastic rather than a deterministic process because the particle sizes and bottom shear stress vary spatially and temporally (He and Han 1982). Usually, the existence of particle motion is determined by assuming a threshold shear stress or threshold velocity. Definitions of threshold have been categorized by Lavelle and Mofjeld (1987a) as those based on sediment flux in a flume, visual flume observations (Kramer 1935, White 1970, Mantz 1977), erosion rate experiments for cohesive sediments (Partheniades 1965, Ariathurai and Arulanandan 1978, Sheng and Lick 1979, Kelly and Gularte 1981, Parchure and Mehta 1985), and field measurements in marine environments (Sternberg 1971, Wimbush and Lesht 1979, Lesht et al. 1980, Larsen et al. 1981). Because of the stochastic nature of the inception of particle motion problem, however,
some particle motion will still occur below the threshold values (Einstein 1941, 1966, Taylor and Vanoni 1972, Vanoni 1975, Christensen 1981, Lavelle and Mofjeld 1987a). The concept of threshold is useful for practical problems (Simons and Senturk 1977 pp. 417-487, Blaisdell et al. 1981, Blaisdell 1988) and when ability to observe particle motion is limited in the field (Sternberg 1971, Wimbush and Lesht 1979, Lesht et al. 1980). Threshold criteria should only be applied with the knowledge that initiation of particle motion is a stochastic process.
The most common threshold criterion is probably that presented in the Shields diagram as modified by Rouse (fig. 1-1) (Vanoni 1975 p. 96, Simons and Senturk 1977 p. 410). The Shields threshold criterion was determined by extrapolating measured transport rates of laboratory experiments to the point of zero transport for fully developed turbulent flows, noncohesive sediments, and flat beds. The abscissa is the boundary Reynolds number
R. U. d s/ V (1-1)
in which is the shear velocity for which U. (,o/p) where r is
the bottom shear stress and p is the fluid density, and v is the kinematic fluid viscosity. The curve for R. less than two was extrapolated by Shields and is not based upon data. The ordinate is a dimensionless shear stress or Shields parameter
* (- ) ds (1-2)
in which -ys is the specific weight of the sediment particle and 7 is the specific weight of the fluid. The line on figure 1-1 indicates the critical condition for sediment motion. If (R*,T*) is below the line, then there is no sediment motion, and if (R*,r*) is above the line,
then there is sediment motion.
ALU ( l)gd,
0.1 -- -0 00
0.4 0.6 1.0
6 8 10 20 40 80100 200
BOUNDARY REYNOLDS NUMBER, R.- Figure 1-1, Shields critical shear stress diagram, from Vanoni (1975).
Several modifications to the Shields diagram have been proposed because the bottom shear stress is included on both axes of the diagram, so an iterative procedure must be used to determine the critical shear stress for a given particle. The Task Committee on Preparation of Sedimentation Manual (1966) added the quantity d s' 0.5
to the diagram in which g is the acceleration of gravity. Yalin (1977), Bonnefille (see Vollmers 1987) and Gessler (1971) regrouped the dimensionless variables to make the diagram easier to use.
The Shields diagram also has been modified to account for bed forms and small boundary Reynolds numbers. Gessler (1971) adjusted Shields diagram because some of Shields' flume experiments formed ripples and small dunes that increased the critical shear values by 10 percent. Inman (1963) shows a second curve for rippled beds. Shields diagram was extended for values of R, less than 2 by Miller et al. (1977) (mostly with data from White (1970)) with an envelope encompassing the data scatter and by Mantz (1977) with a power function that lies inside the envelope (Larsen et al. 1981). The extended Shields diagrams are applicable to noncohesive fine sands and silts.
Although the Shields diagram was developed for unidirectional flows, it also may be applicable to the estuarine and marine environments. Field observations in oscillatory flow environments have been in good agreement with the Shields criterion (Davies 1985). Madsen and Grant (1977) stated that Shields diagram is applicable to locations with both waves and currents and good agreement with data in such environments was observed by Drake and Cacchione (1986). Field measurements in Puget Sound by Sternberg (1971) were in good agreement
with the Shields diagram modified to account for ripples presented by Inman (1963). Larsen et al. (1981) found that the extended Shields diagram was in good agreement with observed threshold grain motion for oscillatory flows on a continental shelf.
Other threshold relations have been developed for noncohesive
sediments. Lane (1955) used field data to develop curves of critical shear stress vs. mean particle diameter, which give higher critical shear stresses than Shields' diagram. Sundborg (1956) developed a threshold criterion based on particle size and mean velocity 1 meter above the bed, and Inman (1949, 1963) developed a criterion that is dependent on particle size and shear velocity, both of which agree with Puget Sound data (Sternberg 1971). Wiberg and Smith (1987) derived a critical shear stress equation for beds with uniform particle size that corresponded closely to Shields' diagram. In addition, they also derived a critical shear stress equation for heterogeneous beds that was in good agreement with experimental data.
For cohesive sediment beds, the consolidation of the bed is an
important factor that helps determine whether and how the bed sediment will move. When initially deposited, cohesive sediment beds are unconsolidated high concentration suspensions (fluid-mud) and have little shear strength. Consolidation is a time dependent function of the overbearing pressure, particle size, and of the clay mineralogy that dewaters, compresses, and strengthens the bed (Meade 1966, Terzaghi and Peck 1967 p. 84, Hayter 1986). Therefore, there is no general threshold of motion criterion for cohesive sediments such as the Shields diagram for noncohesive sediments. Critical shear stresses for various cohesive sediments and consolidation states have been determined in the laboratory (Partheniades 1965, Ariathurai and
Arulanandan 1978, Sheng and Lick 1979, Thorn and Parsons 1980, Kelly and Gularte 1981, Mehta et al. 1982) and the field (Gust and Morris 1989). Three modes of initial motion of cohesive sediments are reentrainment of unconsolidated high concentration suspensions, surface erosion of individual particles and flocs, and bulk erosion (also called mass erosion) which is the sudden failure of the upper part of the bed (Krone 1986, Mehta et al. 1989a). Bed Load Transport
If a particle moves by rolling, sliding, or saltating (jumping), then it contributes to the bed load. Determining whether a particle that is not in contact with the bed is contributing to the bed load or suspended load can be difficult (Einstein 1950, Bagnold 1966, Murphy and Aguirre 1985). Bed load usually is composed of sand-sized and coarser particles. Fine sediments (silts and clays) are usually immediately suspended upon initiation of motion and are not transported as bed load.
Calculation and measurement of bed load are difficult. Reviews of equations for calculating bed load have been presented by Vanoni (1975, pp. 168-172), Simons and Senturk (1977, pp. 508-543), Yalin (1977), Gomez and Church (1989) (for gravel beds only), Ludwick (1989), and Stevens and Yang (1989). Bed load transport equations were compared by van Rijn (1984a), who concluded that predicted bed load transport rates are accurate only within a factor of 2. Gomez and Church (1989) assessed bed load equations for gravel beds and determined that none of the reviewed equations consistently performed well. Carson (1987) evaluated several factors used to estimate bed load in alluvial channels and determined that the grain component of the bed shear stress (shear stress corrected for sidewalls and bed forms) is a better
predictor of bed load than mean velocity, stream power, and unit stream power. The bed load transport rate is related to the grain component of the bed shear stress so this quantity should be used in bed load formulas when bed forms are present (Wiberg and Smith 1989). Field measurements of bed load are discussed by Emmett (1980), Edwards and Glysson (1988), and Ludwick (1989), but bed load is difficult to measure because any device placed near the bed may disturb the flow and the rate of bed load transport.
Erosion and Bottom Shear Stress
The rate at which particle motion is initiated is an important
quantity for studies and numerical models of sediment transport. van Rijn (1984b) developed a sediment pick-up function (mass per unit area per unit time) by utilizing a mechanical device (a sediment lift) at the bottom of a flume to supply erodible noncohesive particles and compared the developed sediment pick-up function to the sediment pickup functions by Einstein (1950), Yalin (1977), Nagakawa and Tsujimoto (1980), de Ruiter (1982), and Fernandez-Luque (see van Rijn 1984b). Bed load and suspended load transport rates calculated with the developed sediment pick-up function compared well with field and laboratory data (van Rijn 1986b). Noncohesive particles that are picked up are initially saltating and will either contribute to the bed load or suspended load (Murphy and Aguirre 1985).
Quantification of the erosion of cohesive sediments is difficult because of the lack of understanding of erosion mechanisms and the numerous factors involved. Thus, the erosion relationships that have been developed are simple relationships that contain coefficients that must be determined in the laboratory or field. Assuming that the rate
of surface erosion is proportional to the nondimensional excess shear gives (Ariathurai and Arulanandan 1978)
E -M (I.](1-3)
in which E is the surface erosion rate in mass per unit area per unit time, M is an erosion rate constant defined as 'the increase in the rate of erosion for an increase in the interface fluid shear by an amount equal to the critical shear stress of that soil' (Ariathurai and Arulanandan 1978) that has a range of values from 0.003 gcm' minI to
0.03 g cm min r is the bottom shear stress, and r is the critical shear stress for erosion. For shear stresses less than the critical value, no surface erosion occurs. Equation 1-3 was developed using data from placed beds with uniform shear strength which is not representative of cohesive beds in nature (Mehta et al. 1982). Erosion functions by Mehta et al. (1982) and Parchure and Mehta (1985) were determined using more realistic laboratory sediment bed conditions. Equation 1-3 has been applied in numerical models of cohesive sedimentation (Ariathurai and Krone 1976, Thomas and McAnally 1985, Sheng et al. 1990b, Uncles and Stephens 1989). Because equation 1-3 is an empirical erosion formula, it is not limited to cohesive sediments, and it has been used to simulate transport of noncohesive suspended sediments in the lower Mississippi River (Schoellhamer and Curwick 1986).
A general erosion equation is determined by setting the erosion rate equal to a power of the excess shear stress
E = a Ir?7 (1-4)
in which a and q are constants that are determined by calibration.
Values of a have been found to range from 1.9xlO to 3.7xi0 for r
in dynes/cm2, and n has been found to range from 0.23 to 10 (Lavelle et al. 1984). Lavelle et al. (1984) used equation 1-4 to simulate erosion as a stochastic process (Lavelle and Mofjeld 1987a).
The erosion rate is dependent upon the shear stress applied to the bed by the flowing water and the bottom shear stress from the near-bed velocity. Schlichting (1969) gave velocity profile equations for turbulent flow in the near-wall region for hydraulically smooth, rough and transitional cases. Given a measured or simulated velocity profile, these equations can be used to calculate the bottom shear stress. In general, the velocity distribution is
u =1 n(z/ks) + B (1-5)
1.1 K~ s
in which u. is the shear velocity for which 7 = p u*, the velocity u is at an elevation z above the bed, x is von Karman's constant (0.4), k
is the height of the bottom roughness elements, and B is a roughness function that has the form
u. k u, k
B=slog + b ln s + b (1-6)
V 2.3 1
in which u*ks/v is a roughness Reynolds number (R), and s and b are the slope and intercept of a line that fits a segment of the data presented in figure 1-2 (Schlichting 1969 fig. 20.21), which is a plot of B vs. log R developed from laboratory data. The boundary is smooth for R < 5, rough for R > 70, and transitional for intermediate values of R. For a smooth wall, s 5.75 and b 5.5, and equations 1-5 and 1-6 can be written as
u__ = 1 In (9.03 z u* / v) (1-7)
For a rough wall s 0.0 and b = 8.5 and equations 1-5 and 1-6 can be written as
u_- I ln (30 z / k (1-8)
a trns~ionComnletely rough
2 0 a a o z I Z Z 2
I ____Figure 1-2, Roughness func trastion B in terms of Reynolds number, from
S I ':I I
0.2 04 b" as Z Z J.2 6 LB 2. 2' 2, 2.8 28 .10 32
Figure 1-2, Roughness function B in terms of Reynolds number, from Schl.ichting (1969).
for which the elevation with zero velocity is zo=k s/30. Equations 1-5 and 1-6 can be combined to derive a general expression for the elevation with zero velocity
z k R-'s/2.3 e b (1-9)
For transitional flow regimes, figure 1-2 and equations 1-5 and 1-6 must be applied in an iterative manner.
If the velocity used to calculate the bottom shear stress is from the near-bed region of a hydraulically smooth (R < 5) or transitional flow (5 < R < 70), the velocity profile may not be logarithmic near the bed and a different velocity profile equation is applicable. Reichardt (see Wiberg and Smith 1987) developed a velocity distribution equation that gives a linear profile for elevations much smaller than the top of the viscous sublayer, a logarithmic profile for elevations much larger than the top of the viscous sublayer, and a smooth and accurate transition between the two regions. This velocity distribution is
u = [ in(l+z +)
.6+ + -03z+
(ln(z+) + ln(K)) (1 e-z+/1 z )I33z+
/1. 11. e 0
+ + U Zo/V
in which the nondimensional quantities z = uz/v and z0
The bottom roughness is dependent upon grain size in the bed and bed forms. The bottom roughness element that controls the resistance in the region of the water column adjacent to the bed is the grain roughness and above this region the roughness is controlled by larger roughness elements such as ripples, dunes, or surface irregularities caused by bioturbation (Smith and McLean 1977). The grain roughness is normally related to the bed sediment size distribution.
Two types of shear stresses, the total shear stress and the grain shear stress, can be considered. The total shear stress is the shear stress exerted on the flow by all of the bottom roughness elements. The grain shear stress is the spatially-averaged (over a bed form wavelength) shear stress exerted on the sediment particles in the bed by the flow. The grain shear stress is less than the total shear stress. Sediment motion is dependent upon the grain shear stress (Vanoni 1977, McLean 1991).
If the height and wavelength of the bed forms are constant, then the matching elevation at which the velocities from the two regions are equal can be calculated based on the bed form geometry (Smith and McLean 1977). Smith and McLean (1977) measured velocity profiles at several positions on dunes in the Columbia River and showed that the spatially-averaged (or "zero-order") velocity profile could be constructed from equation 1-5. The zero-order velocity profile averages near-bed velocity variations caused by the bed forms and the resulting total shear stress and grain shear stress are spatiallyaveraged values. Expressions for the bottom roughness due to bed forms have been developed by van Rijn (1984c) for steady flow in alluvial channels and by Grant and Madsen (1982) for oscillatory flow.
For an oscillatory flow, a simple expression for the maximum bottom shear stress is (Jonnson 1967) 7' 'Of(1-11) w 2 ub bl
in which f is a friction factor and
-b w cosh(kh) (-2
is the maximum bottom orbital velocity from linear wave theory, where a is the wave amplitude, k is the wave number, w is the angular wave frequency, and h is the water depth. Shallow-water waves are assumed
to be present for kh < w/10 and deep-water waves are present for kh > r (Dean and Dalrymple 1984). Equations and diagrams for the friction factor f as a function of the wave properties and bottom roughness have been determined empirically (Jonsson 1967, Kamphius 1975). For waves in the absence of a mean current and for rough turbulent flow, Grant and Madsen (1979, 1982) determined that the friction factor is given by
f 0.08 / [Ker2(2F) + Kei2(2F )] (1-13)
in which Ker and Kei are Kelvin functions of zero order and
=30 u (1-14)
in which kb is the bottom roughness. The friction factor appears on both sides of equation 1-13, so the equation must be solved iteratively. Equation 1-13 is valid for rough turbulent flow for which Ab/kb > 1 where Ab=Ub/w is the orbital amplitude just above the boundary layer. Bottom stress in oscillatory boundary layers (e.g. Jonsson and Carlsen 1976) has also been computed with turbulent boundary layer models (Sheng 1984, Sheng and Villaret 1989).
A potentially important resuspension mechanism in estuaries is the nonlinear interaction of a wave field and a mean current that can increase the shear stress on the bed to a value greater than the sum of the wave only and current only shear stresses. Grant and Madsen (1979) developed a model to estimate the bed shear stress when waves and current are present. The model is based on the assumption of rough turbulent flow that is wave dominated. They also developed a model that includes ripple formation and the effect of ripples on the bed shear stress (Grant and Madsen 1982). Weaknesses of the Grant and Madsen models include the introduction of a fictitious reference velocity at an unknown level, a rather arbitrarily estimated thickness
of the wave boundary layer, and the model being valid only for wave dominant cases (Christoffersen and Jonsson 1985). The Grant and Madsen models also assume that 1) the thickness of the logarithmic layer is constant, which is not correct when waves are present (Sheng 1984), 2) the wave field can be represented by a single wave period and wave height, and 3) the eddy viscosity is linear and time invariant. Cacchione et al. (1987) found that the shear stress and bottom roughness estimated by the moveable bed model were in good agreement with estimates from measured velocity profiles, but no bottom photographs were available to check the estimated bed form geometry. Drake et al. (1992) took bottom photographs and found that the moveable bed model overestimated the size of bottom ripples. Larsen et al. (1981) present a simpler solution for the model formulation by Grant and Madsen (1979) that is applicable to smooth, transitional, and rough bottom roughness regimes and the shear stress at the observed threshold of motion of noncohesive sediments observed in the field was in good agreement with Shields diagram. The results of the Larsen et al. (1981) model and the Grant and Madsen (1979) model are virtually identical for rough bottoms. Simpler wave-current models that compare well with laboratory data have also been developed by Christoffersen and Jonsson (1985) for wave and current dominated cases and by Sleath (1991) for wave-dominated cases and rough beds. All of the above models consider wave-averaged bottom stresses by invoking some sort of a priori parameterization of wave-current interaction.
Wave-current interaction in bottom boundary layers has also been studied without such a priori parameterization. Sheng (1984) used a Reynolds stress turbulence model to simulate the detailed dynamics of the boundary layer over the wave cycle by using a small time step
(1/100 of the wave cycle) and specifying the mean and orbital currents at the outer edge of the boundary layer. The results were then averaged over the wave cycle to produce wave-averaged stresses. Sheng (1984) found that the model of Grant and Madsen (1979) generally overestimated the wave-averaged stress and apparent roughness height. Suspended Load Transport
Suspended sediment is transported by the flow in the water column. The Reynolds time-averaged equation for three-dimensional sediment transport, written in tensor notation, is (Vanoni 1975)
8c ac au'c'
at + ax (1-15)
in which c is the mean suspended-sediment concentration, u is the mean velocity of the sediment particles, c' is the turbulent fluctuation of sediment concentration about the mean, u' is the turbulent fluctuating velocity of sediment particles, t is time, and x is the coordinate axis. The first term in equation 1-15 represents the time rate of change of sediment concentration, the second term represents the advection and settling of particles, and the term on the right hand side represents the turbulent dispersion of particles. The second order correlation of velocity fluctuation and concentration fluctuation is often represented as the product of the mean concentration gradient and an eddy diffusivity Kv such that (Vanoni 1975)
u c' - K ac (1-16)
1 v. ax.
The eddy diffusivity is often assumed to be proportional to the eddy viscosity Av (Kv=OA v) which is often assumed to be related to mean flow variables (Fischer et al. 1979) or to have a particular distribution (Vanoni 1975, Fischer et al. 1979). Eddy diffusivity also can be
calculated with an advanced turbulence closure algorithm (Sheng 1986a, Celik and Rodi 1988, Sheng and Villaret 1989).
An analytic expression for the vertical distribution of suspended sediment in an open channel can be derived from equations 1-15 and 116. Assuming that lateral and longitudinal variations are small and that the mean vertical water velocity is small, equation 1-15 reduces to the vertical conservation of mass equation for sediment
t az (wsc w'c') (1-17)
in which w is the terminal settling velocity of sediment particles and z is the vertical coordinate axis (Vanoni 1975). The first term within the parenthesis is the settling flux of sediment and the second term is the vertical flux of sediment (usually upward) caused by turbulence. For steady flow and a parabolic distribution of the eddy diffusivity, equations 1-16 and 1-17 can be used to derive the Rouse equation
( h-z a )Z
in which z is the elevation above the bed, c a is the suspended-sediment concentration at a usually small reference elevation a above the bed, and the exponent Z is the Rouse number equal to ws /(ficu*) (Vanoni 1975). Equation 1-18 gives a suspended sediment distribution that is greatest near the bed, as is expected due to the negative buoyancy of sediment particles. In addition, the vertical gradient of suspendedsediment concentration is greatest near the bed. The water velocity is small near the bed compared to the vertically-averaged velocity, so the dissimilar distributions of suspended sediment and velocity must be considered when calculating the suspended load in a river (Schoellhamer 1986, McLean 1991).
Flocculation and Aggregation
Salt flocculation of clay minerals and formation of large
organically bound aggregates of inorganic particles may occur in the water column, especially in estuaries. Clay minerals transported to estuaries by rivers encounter high cation concentrations that reduce repulsive forces that prevent flocculation in freshwater. Particle cohesion begins at salinities of 0.6 to 2.4 ppt, depending on the clay mineralogy, and increases with salinity up to about 10 ppt, although the rate of increase is small for salinities greater than 3 ppt (Krone 1962, Mehta 1986). Seawater has salinity of about 34 ppt. Flocculation has been observed for clay concentrations as low as 50 mg/L (Ozturgut and Lavelle 1986). As a floc grows, the density, settling velocity, and shear strength of the floc decrease (Krone 1986). Turbulence may break up relatively weak flocs (Krank 1984). Meade (1972) states that salt flocculation has been overemphasized in the literature while biological agglomeration by filter feeding organisms has not received enough attention.
Suspended material in estuaries and oceans is commonly found in the form of large aggregates (length scale 0.5 mm or larger), often called marine snow (Wells and Shanks 1987). In the ocean, large aggregates are primarily biogenic material (Fowler and Knauer 1986), but estuarine aggregates are primarily inorganic particles attached to organic material (Eisma 1986). The primary collision mechanism that forms large aggregates is differential settling, not Brownian motion or fluid shear (Hawley 1982, Eisma 1986). Aggregation by organisms may also be important in estuaries (Meade 1972, Krank 1984, Eisma 1986). Large aggregates have been observed to remain intact in currents as much as 50 cm/s in estuaries (Wells 1989), but large aggregates tend to
break apart when sampled (Shanks and Trent 1980, Krank 1984, Eisma 1986, Fowler and Knauer 1986). Sampling procedures are limited to in situ settling columns (Shanks and Trent 1980, Gibbs 1985, Fowler and Knauer 1986) and photography (Krank 1984, Eisma 1986, Wells and Shanks 1987, Wells 1989). Eisma (1986) and Dyer (1989) state that the size of the aggregates is limited by the turbulence microscale, which is the size of the smallest turbulent eddies. The density of large aggregates decreases with increasing size and the settling velocity and porosity increases with increasing size (McCave 1975, Hawley 1982, Gibbs 1985). Settling and Deposition
Particles that are part of the bed or suspended load may settle through the water column and deposit on the bed. A noncohesive particle falling in quiescent fluid is affected by the forces of gravity, buoyancy, and drag. The terminal settling velocity for a sphere in quiescent fluid is given by Stokes law and Rubey's equation, which are presented by Vanoni (1975) and Simons and Senturk (1977). A nonspherical particle, which is common in nature, may have a different terminal settling velocity. A shape factor defined by Alger and Simons (1968) can be used to predict settling velocity and settling behavior of nonspherical particles and shells (Mehta et al. 1980). Terminal settling velocity equations are for a single particle, but high concentrations of noncohesive particles, which are most likely to occur near the bed, may hinder settling and decrease the terminal settling velocity (Vanoni 1975, Simons and Senturk 1977). The size of suspended particles tends to decrease with elevation above the bed, so assuming a single suspended particle size with a single settling velocity may not be appropriate (McLean 1991).
Deposition of suspended sediments in natural hydrologic systems is complicated by turbulence. The effective settling velocity of a particle settling in a vertically oscillating flow will be less than the terminal settling velocity for the particle in quiescent fluid because of nonlinear modification of the drag force (Hwang 1990) or vortex trapping (Nielson 1984). As a particle settles toward the bed, the turbulence intensity generally increases and, thus, the probability increases that a turbulent eddy will carry the particle either upward or downward. The turbulence intensity will decrease very near the bed and vanish in the viscous sublayer adjacent to the bed. If a particle passes through the high turbulence zone near the bed, it can then deposit on the bed. Li and Shen (1975) and Bechteler and Farber (1985) presented random walk models that stochastically simulate particle settling. Turbulence, however, is not an independent variable. Density stratification by suspended sediments can dampen turbulence (Sheng and Villaret 1989) and reduce the transport capacity of the flow (McLean 1991).
Because turbulent intensity and the vertical gradient of sediment concentration are greatest near the bed, near-bed hydrodynamic processes determine deposition rates. Sheng (1986c) derived a deposition velocity formula by considering the resistances that sediment particles experience in various near-bed layers including the logarithmic layer, vegetation canopy, laminar sublayer, and biochemical effects. Numerical suspended-sediment transport models that solve equation 1-15 generally have finer vertical grid resolution closer to the bed and represent the bottom boundary condition either by setting the net upward sediment flux equal to an erosion rate minus a deposition rate (Sheng and Lick 1979, Schoellhamer 1988, Sheng et al.
1990a), by estimating a near-bed sediment concentration (van Rijn 1986a, Celik and Rodi 1988, Schoellhamer 1988), or by setting a net erosion or deposition flux which is dependent on the bottom shear stress (Thomas and McAnally, 1985).
Estimates of sediment deposition rates in estuaries have been made with sediment traps and acoustic devices. Interpretation of sediment trap data is difficult, but sediment traps have been used to estimate deposition rates (Oviatt and Nixon 1975, Gabrielson and Lukatelich 1985). Bedford et al. (1987) used an acoustic transceiver to measure near-bed suspended-sediment concentration profiles and calculated deposition and erosion fluxes over a 3.5 hour period and a tidal cycle at one site in Long Island Sound.
Large aggregates control settling and are an important mechanism for transporting material to the bottom of oceans (McCave 1975, Shanks and Trent 1980, Hawley 1982, Fowler and Knauer 1986) and estuaries (Wells and Shanks 1987, Dyer 1989). Fowler and Knauer (1986) give a minimum range of large aggregate settling velocities of 1 to 1000 m/day. Shanks and Trent (1980) measured settling velocities of large aggregates in Monterey Bay, California, and the northeastern Atlantic Ocean and found that the range of settling velocities was 43 to 95 m/day.
The settling velocity of depositing cohesive sediments is
dependent upon the sediment concentration. Krone (1962) measured settling velocities of San Francisco Bay sediment in still water and determined that the median settling velocity was proportional to the concentration to the 4/3 power. This is a typical result for settling column experiments with cohesive sediments (Mehta 1986), probably because differential settling is the dominant collision mechanism
(Farley and Morel 1986). In estuaries, however, collisions caused by velocity gradients are most important because they form the strongest aggregates (Krone 1986). Ross (1988, also in Mehta 1989) conducted a settling column experiment and found that the settling velocity of Tampa Bay mud increased as the concentration increased up to about 1 g/L, the settling velocity was about constant (about 0.32 mm/s or 27 in/day) for concentrations from about 1 to 10 g/L, and the settling velocity decreased as the concentration increased above 10 g/L due to hindered settling.
Deposition of flocs of cohesive sediments may be prevented by the turbulent boundary layer above the bed in which flocs may be broken apart and lifted up into the water column. Krone (1962) conducted deposition experiments in a recirculating flume from which he determined that the rate of deposition of cohesive sediment from a vertically mixed flow is
D=-C w s(1 TI),c r< r (1-19)
in which C is the vertically well-mixed suspended-sediment concentration and rTc is the critical shear stress for deposition that must be determined by analyzing time series of concentration and shear stress measured in the laboratory or field. Laboratory experiments indicate that r cris less than 7 r, the critical shear stress for erosion. If r > r cr' no deposition occurs, and if r < r r' equation l19 will eventually deposit all of the sediment in suspension. Laboratory experiments by Mehta and Partheniades (1975), however, showed that a constant fraction of the original suspension will be maintained in suspension indefinitely for r cr > 7 > r mnwhere r mnis a minimum shear stress below which all of the suspended sediment will deposit. Equation 1-19 is an empirical formula for deposition that
does not include the near-bed hydrodynamic processes that determine deposition rates. Equation 1-19 has been used to simulate noncohesive suspended-sediment transport in the lower Mississippi River (Schoellhamer and Curwick 1986) in addition to cohesive sediment transport (Ariathurai and Krone 1976, Thomas and McAnally 1985, Uncles and Stephens 1989).
Other Bed and Near-bed Factors
Many bed and near-bed factors may affect hydrodynamics and
sediment transport. These factors include bed forms, bed armoring, suspended-sediment stratification, fluid-mud, and porous beds. Some of these factors (bed forms, bed armoring, suspended-sediment stratification, and fluid-mud) are the result of sediment transport and affect the hydrodynamics, which, in turn, affect the sediment transport. Biological activity also affects sediment transport.
One consequence of noncohesive sediment transport can be the
formation of bed forms in riverine (Kennedy 1969, Vanoni 1975, Simons and Senturk 1977, Yalin 1977) and coastal (Boothroyd 1985) environments. Bed forms such as ripples and dunes do not occur in cohesive sediment beds. Bed forms increase the bottom roughness of open channels compared to flat bed conditions in which the only roughness elements are the bed sediment particles. Because some of the total bottom shear stress is caused by the bed forms, the shear stress applied to the particles will decrease, so the total (grain and form) critical shear stress for initiation of particle motion will increase (Vanoni 1975, McLean 1991). On the upstream face of a dune, the grain shear increases from zero at the reattachment point to a maximum at the crest (Shen et al. 1990). van Rijn (1984c) used flume and field data to develop and verify a methodology to predict bed form dimensions and
effective hydraulic roughness for steady flow in alluvial channels. For oscillating flows, Grant and Madsen (1982) present empirical relations for bed ripple geometry and an expression for the bottom roughness. Drake et al. (1992), however, found that the Grant and Madsen empirical relations overestimated ripple height on the Northern California continental shelf.
An armored or paved bed has finer particles removed from the bed surface so that only coarser noncohesive particles remain at the surface thus preventing erosion of finer particles below (Simons and Senturk 1977, Sutherland 1987). For equilibrium conditions, Jain (1990) defines an armored bed as having a coarse surface layer that is inactive (or immobile) and a paved bed as having a coarse surface layer that is actively eroding and depositing (or mobile). For nonequilibrium or degrading conditions, the particle-size distribution of the surficial bottom sediments coarsens and the erosion rate decreases with time. Therefore, near-bed sediment transport may be dependent upon the armoring process, especially in alluvial channels. Dawdy and Vanoni (1986) review several bed armoring and pavement studies and predictive algorithms. More recently, bed armoring algorithms have been presented by Karim and Holly (1986) and Park and Jain (1987).
Near-bed suspended-sediment stratification will affect the
hydrodynamics and, therefore, affect sediment transport processes at the bed. As previously mentioned, high near-bed sediment concentrations may hinder settling of particles. Several authors have found that sediment-laden flows reduced near-bed turbulence intensity or bottom shear stress in steady and uniform open-channel flows (Vanoni and Nomicos 1960, Itakura and Kishi 1980, Lau 1983, Julien and Lan
1988, Xingkui and Ning 1989). West and Oduyemi (1989) measured turbulence and near-bed density stratification in an estuary and found that density stratification damped turbulence. Flume experiments by Vanoni and Nomicos (1960), however, indicated that the effect of bed forms on the total bottom shear stress is much greater than the effect of suspended sediment. Parker and Coleman (1986) found that suspended sediments may either increase or decrease near-bed turbulence intensity and bottom shear stress (discussed by Julien and Lan (1988)). Near-bed stratification in sediment-laden flows may reduce near-bed turbulent shear stresses and, therefore, reduce erosion rates (Sheng and Villaret 1989) and the sediment-transport capacity of the flow (McLean 1991). Costa and Mehta (1990) collected hydrodynamic and suspended sediment data in a high energy coastal environment and found a hysteresis in the relationship between suspended-sediment concentration and the flow velocity, with higher concentrations for accelerating flows, due to near-bed density stratification.
An extreme case of near-bed suspended-sediment stratification is the formation of an unconsolidated fluid-mud by cohesive sediments above a consolidated bed in estuarine and near-shore environments. In this situation, the water column is composed of two layers--a thin dense non-Newtonian fluid underlying a Newtonian fluid. The fluid-mud may interact with the overlying fluid. Surface waves cause mud waves to form in the fluid-mud and the fluid-mud acts to attenuate (dampen) surface waves (Suhayda 1986, Wells and Kemp 1986, Mehta 1991, Jiang and Mehta 1992). Suhayda, Mehta, and Jiang and Mehta developed models to simulate the interaction of the two fluids. Wells and Kemp also point out that the formation of mud waves can produce significant reentrainment of the fluid-mud.
The bed is commonly assumed to be an impermeable boundary but
water exchange between the bed and water column may advect sediment and ripples may enhance water exchange at the bed/water column interface. In an alluvial river water may exchange between the river and pore space of the alluvium (Bencala et al. 1984) and this water exchange may transport fine sediments by advection to and from the relatively coarse alluvium (Jobson and Carey 1989). Thus, advection may be an important transport process at the bed/water interface for fine, noncohesive sediments in an alluvial channel. For oscillatory flows over porous beds in coastal environments, the total water exchange across the bed/water column interface averaged over the wave period is significantly greater for rippled beds than for flat beds (Shum 1992).
Benthic biological communities may influence sediment-transport processes, especially in biologically abundant estuaries. Erosion rates may be decreased by microbial films and benthic algae that stabilize sandy sediments (Gabrielson and Lukatelich 1985, Grant et al. 1986, Meadows et al. 1990) and cohesive sediments (Montague 1986). Tracks made by motile bivalves, however, reduced the critical shear velocity of fine sandy sediments in a laboratory flume by 20 percent (Nowell et al. 1981). Fecal mounds from polychaetes and a depositfeeding bivalve in a laboratory flume had a higher critical shear velocity than ambient cohesionless sediments and were transported as bed load (Nowell et al. 1981). Sediments that have been bound by secretions from benthic invertebrates may settle more slowly than unbound particles when resuspended (Meadows et al. 1990). Seagrasses reduce bottom orbital velocities of wind-waves and thus reduce erosion and increase deposition (Ward et al. 1984). Biological effects frequently are ignored in sediment-transport studies because biological
activity is difficult to quantify in the field, difficult to reproduce in a laboratory, and difficult to incorporate into a numerical model.
Field Studies Related to Estuarine Sediment Transport
Field studies of estuarine sediment transport are often conducted to determine the mechanisms that control sediment transport processes, such as resuspension and horizontal transport. Data is collected by in situ instrumentation, water sample collection and analysis, and/or bottom sediment sampling and observation. Statistical analysis, time series analysis, and numerical process models are used to analyze the data. Field studies provide less control than laboratory studies, but they are more applicable because of the difficulty recreating natural conditions in the laboratory. For numerical studies, an initial or concurrent field study may indicate the processes that must be included in the numerical model in order to accurately simulate sediment transport and a field study will provide data for calibration and validation of the numerical model. Sediment resuspension mechanisms are often the focus of estuarine field studies because the bottom of the estuary is an omnipresent potential source of suspended sediment and many potential sediment resuspension mechanisms are present in the estuarine environment. The potential sediment resuspension mechanisms include tidal currents, wind waves, wave/current interactions, seiches, trawling, and vessel traffic. Many of these potential mechanisms are also present on the continental shelf and in large lakes, so some relevant field studies in these environments will also be mentioned in this section.
The mean current in estuaries and on the continental shelf may
resuspend bottom sediments. Wimbush and Lesht (1979) deployed a tripod with current meters and a bottom camera at a site with a bottom
sediment of medium sands in the deep (710 meters) Florida Straits and estimated the critical velocity for ripple formation. Lavelle et al. (1984) deployed a current meter and transmissometer 5 meters above a fine sediment bottom in 200 meters of water in Puget Sound. They observed that tidal currents resuspended bottom sediments and that the erosion rate could be described with the shear stress power law given by equation 1-4. Both of these studies were conducted in deep water where wave motion was not observed. Bohlen (1987) deployed an instrument array that included an electromagnetic current meter and transmissometers for monitoring suspended-solids concentrations in 12 meters of water in Chesapeake Bay. Spectral analysis of the suspendedsolids concentration showed variation at tidal frequencies, including a spring/neap variation. Roman and Tenore (1978) collected and analyzed water samples over several tidal cycles in Buzzards Bay, Massachusetts, and observed resuspension of organic carbon and chlorophyll-a at a site with a muddy bottom in 13 meters of water by tidal currents that was significant enough to potentially affect estuarine productivity. They did not address resuspension by wind waves, possibly because the data may contain a fair weather bias due to the manual collection of water samples and the lack of in situ instrumentation. Costa and Mehta (1990) observed resuspension by tidal currents in Hangshou Bay, China, a high energy tidal environment.
Wind waves may resuspend bottom sediments in relatively shallow water. Anderson (1972) collected water samples in a tidal lagoon with fine sediments and found that wave heights of only a few centimeters resuspended bottom sediments in water depths of 40 to 150 cm. Water samples and sediment traps were used by Ward et al. (1984) to determine that shallow water sites (less than 2 meters) in Chesapeake Bay with
seagrasses attenuate wave motion and reduce resuspension compared to sites without seagrasses. Gabrielson and Lukatelich (1985) found that sedimentation rates calculated from sediment traps deployed in the Peel-Harvey estuarine system were temporally correlated with the strength and duration of wind events and spatially correlated with wind direction and fetch. Kenney (1985) deployed sediment traps in Lake Manitoba at a depth of 4.2 meters and found that resuspension of bottom sediments during wind events stratified the water column. Lavelle et al. (1978) deployed a current meter and turbidimeter 1 meter above the medium sand bed of Long Island Sound at an average depth of 10.5 meters. Wind waves during a storm resuspended bottom sediments and tidal currents did not cause resuspension. An empirical relationship between suspended sediment concentration and wave orbital velocity in Long Island Sound was developed by Lesht et al. (1980). In addition to resuspension by tidal currents in Chesapeake Bay discussed previously, root-mean-squared velocity fluctuations and suspended-solids concentrations increased when the wind was blowing up the longitudinal axis of Chesapeake Bay (Bohlen 1987). Davies (1985) deployed a tetrahedra containing electromagnetic current meters and a video camera in 4 to 10 meters of water in Start Bay, England, and found that the observed threshold of sediment motion in oscillatory flow was in good agreement with Shields diagram (fig. 1-1). Sheng et al. (1990b) collected synoptic suspended sediment data and deployed anemometers, pressure transducers, electromagnetic current meters, and optical backscatterance suspended-solids sensors from several platforms and piles over two one-month periods in Lake Okeechobee, Florida. It was found that wind waves associated with the diurnal lake breeze resuspended fine sediments and this field data were used to calibrate
and validate three-dimensional and vertical one-dimensional numerical sediment transport models (Sheng et al. 1990b, Sheng et al. 1992). Mehta (1991) also collected data and applied a vertical one-dimensional model to study fluid-mud and sediment resuspension by wind waves in Lake Okeechobee. Perjup (1986) deployed an instrument tower in about 2 meters of water in the Ho Bugt estuary, Denmark, from which water velocity, suspended-solids concentration, salinity, depth, wind speed, and wind direction were measured. Perjup found that the suspendedsolids concentration did not correlate with mean water velocity but was correlated to parameters containing wind speed and onshore wind direction and inversely correlated with salinity, which was wellcorrelated with water depth. These results indicate that resuspension was caused by onshore wind waves, especially at low tide when the nearbed orbital wave motion can be greatest.
As mentioned previously, the nonlinear interaction of waves and current can significantly increase the bottom shear stress and thus it may be an important sediment resuspension mechanism. Drake and Cacchione (1986) used data from Norton Sound, Alaska, and the northern California continental shelf to calculate the grain shear stress with a modified version of the fixed bed wave/current interaction model of Grant and Madsen (1979). The observed resuspension threshold was in good agreement with Shield's diagram (fig. 1-1). Cacchione et al. (1987) and Drake et al. (1992) compared field data collected on the northern California continental shelf (water depth 85 meters) and results of the moveable bed wave/current interaction model of Grant and Madsen (1982) and found good agreement for bottom shear stress but not bed form geometry. The high bottom stresses caused by wave/current interaction during winter storms were a major factor controlling the
distribution of surficial sediment on the northern California continental shelf. Measurements were made with a tripod containing electromagnetic current meters, a pressure transducer, a nephelometer (for measuring light scattering which can be calibrated to suspended solids), and a camera (Cacchione and Drake 1979).
Estuaries contain abundant fisheries and they are convenient sites for ports, so trawling and commercial vessels are potential sediment resuspension mechanisms. Churchill (1989) reviewed several sets of hydrodynamic and sediment data from the middle Atlantic Bight and determined that sediment resuspension by trawling is a significant source of suspended sediments. Schoellhamer (1991b) found that during the departure and arrival of a cruise ship at the Port of St. Petersburg the suspended-solids concentration increased almost an order of magnitude due to the maneuvering of the vessel. The resuspended sediments settled within 1 to 2 hours once the cruise ship either docked or departed.
In partially-mixed and well-mixed estuaries, the maximum
suspended-solids concentration is often present in brackish waters and is greater than concentrations found elsewhere in the estuary (Uncles and Stephens 1989). This feature is called a turbidity maximum and field data collection programs and numerical models have been used to determine the controlling mechanisms. Uncles and Stephens (1989) collected water samples and data from an optical suspended-solids sensor to measure salinity and suspended-solids concentrations during twice monthly sampling trips that proceeded up the Tamar estuary, England. These data were used with a longitudinal one-dimensional model to determine that the location of the turbidity maximum is associated with the location of the freshwater/saltwater interface and
that local resuspension, and perhaps gravitational circulation and stratification, determined suspended-solids concentrations in the turbidity maximum. West and Oduyemi (1989) deployed electromagnetic current meters and suspended-solids sensors on a bed frame in the Tamar estuary and observed that near-bed stratification by suspended sediment reduced the turbulence intensity. Hamblin (1989) collected vertical profiles of velocity, salinity, and suspended-solids concentrations from a vessel anchored at several stations near the turbidity maximum in the upper Saint Lawrence estuary. These data showed that local resuspension by the mean current controlled the suspended-solids concentrations at the measurement sites and that a landward flux of sediment near the bed is maintained by ebb-flood asymmetry and a reduction in vertical mixing during flood tide caused by the salt wedge. A vertical one-dimensional sediment transport model was developed with these data. Grabemann and Krause (1989) analyzed time series of long-term salinity, current meter, and optical transmittance sensor data collected in the Weser River estuary to determine that tidally controlled resuspension and deposition is the dominant process in the turbidity maximum and that gravitational circulation is a longterm source and sink of particles. An intensive two-week experiment on the Weser River estuary included the deployment of instrumentation to measure velocity, salinity, and suspended-solids concentration at several locations and similar measurements were taken from vessels (Lang et al. 1989). These data showed that there was a close relationship between suspended-sediment concentration and near-bed velocity gradient and stratification and the data was used to develop a three-dimensional numerical model.
Most of the field studies discussed so far have focused on tidal transport processes and vertical sediment transport processes, but field studies have also been conducted to quantify subtidal horizontal sediment transport. Powell et al. (1989) collected water samples in South San Francisco Bay and observed that during spring runoff fresh water and suspended sediment from the Sacramento River entered South San Francisco Bay. Wells and Kim (1991) used vibracores, surficial sediment data, and vertical profiles of velocity, salinity, and suspended sediment concentration collected during monthly sampling trips from several sites in the Neuse River estuary, North Carolina, to describe long-term sedimentation patterns. In Puget Sound, Baker (1984) collected suspended-solids concentration data with a transmissometer at several sites during several sampling trips to help determine that gravitational circulation and surface and bottom sources of particles control the distribution and transport of suspended solids.
Sample collection for the preceding studies was generally
performed by collecting water samples and/or collecting pressure, velocity, and suspended solids concentration data from conventional sensors. Alternative methods of data collection have been utilized to collect data on estuarine sediment transport processes. Rubin and Mc~ulloch (1979) used side scan sonar to determine the shape of bedforms in central San Francisco Bay, which indicate near-bed circulation patterns and sediment transport directions. Bedford et al. (1987) used an acoustic transceiver to sample the sediment concentration in 1 cm bins in the bottom 70 cm of the water column over a 3.5 hour period and a tidal cycle at a site in Long Island Sound. This device provides much better resolution of the near bed sediment
concentration profile than is available from optical instrumentation, but calibration is difficult due to sensitivity to the particle size distribution. Resuspension fluxes were calculated and were found to be best correlated with the squared velocity fluctuation (horizontal and vertical) due to wind waves and turbulence about the mean velocity. Portable flumes that are deployed on the bottom of the estuary have been used to study the erosion of in situ sediments under controlled conditions. For example, Young and Southard (1978) deployed a sea floor flume in Buzzards Bay, Massachusetts, and observed that the in situ critical shear velocity was one-half the value found in laboratory experiments due to bioturbation. Laboratory and sea flume values for the critical shear velocity also differed in Puget Sound (Gust and Morris 1989). To estimate the residence time of fine particles introduced at the water surface in Puget Sound (11-16 days), Lavelle et al. (1991) used vertical profiles of radioactive isotope activities and a sorption model.
Due to practical limitations, it is very difficult to collect
synoptic samples throughout a large water body by conventional means, but remote sensing from aircraft and satellites can be used to synoptically determine suspended-solids concentrations near the water surface in large water bodies. The advantage of remote sensing is that a measurement of an entire estuary can be made instantly but the disadvantages are that remote measurements must be calibrated with suspended-solids concentrations, sampling times and locations are limited by the satellite orbit, clouds and other weather may degrade or prevent satellite observation, and resolution may be limited. Sheng and Lick (1979) used remote-sensing data and field data to produce the near-surface suspended-sediment concentrations in the western basin of
Lake Erie, which were then used to provide initial conditions and validation for a numerical sediment transport model of Lake Erie. Huh et al. (1991) used remote sensing to help determine that storms 1) produce landward sediment transport along coastal Louisiana and 2) help build marshes. In Mobile Bay, remote sensing has shown rapid changes in sediment concentrations due to high river inflow and wind-induced sediment resuspension (Stumpf 1991).
Numerical Models Applicable to Estuarine Sediment Transport
Numerical models can be used to simulate or predict estuarine
sediment transport, resuspension events, sedimentation rates, adsorbed constituent transport, and light availability. Transport processes are dependent upon hydrodynamics, so sediment transport models require an accurate hydrodynamic model. Deposition, erosion, and density stratification caused by suspended sediments may affect the hydrodynamics, so a coupled hydrodynamic and sediment transport model may be required. Many numerical models of suspended sediment transport have been developed for steady riverine flows (for example van Rijn 1986a, Celik and Rodi 1988, Schoellhamer 1988), but they are not applicable to estuaries because of the unsteady motions of tidal waves, seiches, and wind waves. Some riverine models are stochastic or random walk models that use Lagrangian particles to represent suspended sediment (Alonso 1981, Bechteler and Farber 1985). The random walk approach may be applicable to estuarine sediment transport. Mehta et al. (1989b) reviewed estuarine applications of primarily cohesive sediment transport models, and they discussed simulation of the bed and zero-, one-, two-, and three-dimensional models. A similar outline will be followed herein. Some estuarine sediment transport processes are also present on the continental shelf and in large lakes, so some
relevant numerical models for these environments will also be discussed in this section.
The properties of the sediment bed may vary with time and depth below the interface with the water column, especially for cohesive sediments. Bed properties such as density and shear strength may vary with distance below the top of the sediment bed, the elevation of which may vary during a tidal cycle. In order for a numerical sediment transport model to account for these temporal and spatial variations, the bed could be divided into layers with different properties, and new layers could be added during deposition and existing layers could be removed by erosion. Properties such as layer density, thickness, and shear strength can vary temporally and spatially in the simulated estuarine bed (Thomas and McAnally 1985, Hayter 1986, Sheng 1991).
If the spatial variation of suspended-sediment concentration can be ignored, then only the temporal variation of suspended-sediment concentrations needs to be considered. This type of modeling is referred to as zero-dimensional and is equivalent to assuming that the study area is well-mixed. Krone (1985) used a zero-dimensional model to simulate and predict deposition in a marsh. Amos and Tee (1989) simulated the Cumberland Basin in the Bay of Fundy as a well-mixed water body in order to calculate sediment fluxes at the mouth of the Basin. Because the distribution of sediment sources in an estuary is likely to be nonuniform and because the settling property of sediment increases concentrations deeper in the water column, the assumption that the spatial variation of suspended sediment is negligible is generally poor.
The longitudinal variation of suspended sediment in an estuary can be simulated with horizontal one-dimensional models. Cross sectional
variations are averaged transversely and vertically, so if the suspended-sediment concentration varies significantly in the cross section, one-dimensional model may not be applicable. Uncles and Stephens (1989) used a longitudinal one-dimensional model to describe the magnitude and location of the turbidity maximum in the Tamar estuary. Equations 1-3 and 1-19 were used to simulate deposition and erosion and the coefficients in the equations were selected by calibration with measured suspended-sediment concentrations. Hayter et al. (1985) predicted shoaling rates in the Hooghly River estuary, India, with a longitudinal one-dimensional model that was calibrated with measured channel dredged material volumes.
The vertical profile of suspended sediment at a given location in an estuary can be simulated with a vertical one-dimensional model. These models are applicable when horizontal gradients of suspendedsediment concentration can be neglected. Weisman et al. (1987) simulated a depositional tidal lagoon with a series of vertical layers for which vertical dispersion was neglected and the simulated shoaling rate was reasonable compared to shoaling rates estimated with radioactive isotopes. Teeter (1986) developed a vertical transport model that uses a Richardson number dependent parabolic eddy diffusivity to include the effect of density stratification. Hamblin (1989) used this model to simulate vertical mixing of suspended sediment at a site in the upper St. Lawrence estuary were suspendedsediment concentrations were observed to depend upon local resuspension. Field data were used to determine the erosion function and particle settling velocity. Costa and Mehta (1990) also applied a Richardson number dependent model to simulate vertical sediment transport in Hangzhou Bay, China. Steady state vertical profiles of
near-bed suspended sediment in the Florida Straits were estimated with a model by Adams and Weatherly (1981) that used three sediment size classes. Velocity data but no suspended-sediment concentration data were collected to calibrate the model. Sheng and Villaret (1989) used a vertical one-dimensional model (OGMiD, which is presented in detail and utilized later) with a simplified second-order turbulence closure model to determine vertical profiles of velocity and suspended-sediment concentration and the erosion rate of bottom sediments for laboratory experiments. They found that near-bed stratification by high suspended-sediment concentrations reduces turbulence intensity and erosion rates. The same basic model was used to determine erosion rates of sediments from measured suspended-sediment concentrations from Lake Okeechobee and was successfully applied to simulate the dynamics of the vertical structure of suspended-sediment concentration over several 3-day and 1-week periods (Sheng et al. 1990b, Sheng 1991, Sheng et al. 1992).
Vertical hydrodynamic models have also been developed for
estuarine applications. OGM1D has been used to simulate storm-induced currents in Grand Bank (Sheng 1986b) and in the Atlantic Ocean during the passage of hurricane Josephine (Sheng and Chiu 1986). Davies et al. (1988) used a one-dimensional vertical hydrodynamic model to simulate the interaction of waves and a mean current. This model achieves turbulence closure with a dynamic equation for turbulence energy and turbulence scaling relations for the mixing length and eddy viscosity, but results were not compared to data. A vertical hydrodynamic model with complete second order turbulence closure model (i.e., Reynolds stress model) was used by Sheng (1984) to simulate the
development and evolution of the logarithmic wave boundary layer for laboratory and field data.
The horizontal transport of sediment in an estuary has been
simulated with depth-averaged two-dimensional models. Ariathurai and Krone (1976) developed a two-dimensional finite element model for simulating cohesive sediment transport in estuaries. Erosion and deposition were simulated with equations 1-3 and 1-19. This model is included in the U.S. Army Corps of Engineers TABS-2 modeling system for estuarine hydrodynamics and sedimentation (Thomas and McAnally 1985) which has been applied to several estuaries (Heltzel 1985, Granat 1987, Hauck 1991). Heltzel (1985) and Granat (1987) used physical model results for the simulated hydrodynamic boundary conditions and the validation of simulated hydrodynamics and measured shoaling rates (instead of measured suspended-sediment concentrations) were used to calibrate sediment simulations. For San Francisco Bay (Hauck 1991), hydrodynamic simulations were in agreement with measured data but sediment simulations did not successfully reproduce field data, possibly because simulated settling velocities for cohesive sediment were not dependent upon the concentration, poor simulation of wave action and related shear stress, and only one grain size for cohesive sediments was simulated.
Depth-averaging may not be appropriate for estuarine hydrodynamic and sediment transport models. Density stratification and gravitational circulation are three-dimensional flow features in estuaries that can not be simulated by a depth-averaged model. Sediment resuspension may be dependent upon the bottom shear stress but depth-averaged models do not calculate the vertical velocity profile and therefore empirical relations must be used to calculate the bottom
shear stress instead of equation 1-5. Because of the settling property of sediment, sediment concentrations are usually much higher near the bed than up in the water column, so depth-averaging may not be appropriate. For example, in an open channel, the near-bed region contains the largest suspended-sediment concentrations but the smallest velocities, so the depth-averaged longitudinal velocity of suspended sediment is less than the depth-averaged water velocity (Schoellhamer 1986). Downing et al. (1985) combined an analytic vertical onedimensional sediment model with linear eddy viscosity and diffusivity, the Grant and Madsen (1979) wave-current model, and a horizontal twodimensional sediment transport model and obtained reasonable simulation results based on a comparison with field data from the Sagavonirktok River Delta in Alaska.
Three-dimensional models permit vertical discretization for simulation of vertical suspended sediment profiles and density stratification in addition to vertically variable horizontal transport by tidal currents, wind-induced circulation, and gravitational circulation. Sheng and Lick (1979) used vertical two- and threedimensional circulation and sediment transport models and a wavehindcasting model, in addition to remote-sensing data, to simulate wind-wave sediment resuspension and transport in Lake Erie. Laboratory experiments were used to determine settling velocity, critical shear stress, and erosion rates which were a bilinear function of the excess shear stress. Wang et al. (1987) used a three-dimensional hydrodynamic and sediment transport model to simulate sediment transport in Kachemak Bay, Alaska, but field data were not included in the study. A threedimensional hydrodynamic and sediment transport model of the turbidity maximum in the Weser estuary, Germany, by Lang et al. (1989) was able
to reproduce some of the features present in measured suspendedsediment concentrations. Hayter and Pakala (1989) applied a threedimensional model of estuarine hydrodynamics, sediment transport, and contaminant transport to the Sampit River in South Carolina but field data was not available to validate the model. Sheng (1991) and Sheng et al. (1992) collected field data and successfully applied a threedimensional model of hydrodynamics, fine sediment transport, and contaminant transport to study phosphorus dynamics in Lake Okeechobee, Florida. A preliminary simulation of three-dimensional sediment transport in Tampa Bay was performed by Sheng et al. (1992).
Relation of this Research to Previous Studies
The purpose of this research was to improve the understanding and numerical modeling of sediment resuspension and the vertical transport of resuspended sediment in the shallow estuarine environment and this was accomplished by combining three important elements of previous studies that have not been previously combined. These three elements are: 1) resuspension by wind-waves in a shallow estuary, 2) field measurements of suspended-sediment concentrations, and 3) a vertical sediment transport model. Estuarine sediments and their transport are important because they may reduce the amount of light penetrating the water column, may act as a source for adsorbed constituents, and may be transported to undesirable locations. Estuaries are also biologically active, and this activity may affect sediment transport. Wind waves are an important resuspension mechanism in many estuaries (table 1-1), and they are shown to be an important resuspension mechanism in Old Tampa Bay in chapter 3. Field measurements of sediment concentrations are needed to understand estuarine sediment transport processes and to develop accurate numerical models. Resuspension from the bed and
Table 1-1, Previous studies that are related to this research and include an estuarine environment, wind waves, field measurements of suspended-sediment concentrations, or a numerical vertical sediment transport model.
Estuarine Wind Field Vertical Environment waves conc. sed. model
Adams and Weatherly (1981) N N N Y
Amos and Tee (1989) Y N Y N
Anderson (1972) Y Y Y N
Baker (1984) Y N Y N
Bedford et al. (1987) Y N Y N
Bohlen (1987) Y N Y N
Cacchione et al. (1987) N Y Y N
Churchill (1989) N N Y N
Costa and Mehta (1990) Y N Y Y
Davies (1985) Y Y N N
Davies et al. (1988) Y Y N N
Downing et al. (1985) Y Y Y N
Drake and Cacchione (1986) N Y Y N
Drake et al. (1992) N Y Y N
Gabrielson and Lukatelich (1985) Y Y N N
Hamblin (1989) Y N Y Y
Hauck (1991) Y Y Y N
Hayter and Pakala (1989) Y N N Y
Kenney (1985) N Y N N
Lang et al. (1989) Y N Y Y
Lavelle et al. (1978) Y Y Y N
Lavelle et al. (1984) Y N Y N
Lavelle et al. (1991) Y N Y Y
Lesht et al. (1980) Y Y Y N
Mehta (1991) N Y Y Y
Perjup (1986) Y Y Y N
Powell et al. (1989) Y N Y N
Roman and Tenore (1978) Y N Y N
Schoellhamer (1991b) Y N Y N
Sheng (1991) N Y Y Y
Sheng and Lick (1979) N Y Y Y
Sheng and Villaret (1989) N N N Y
Sheng et al. (1990b) N Y Y Y
Sheng et al. (1992) N Y Y Y
Stumpf (1991) Y Y N N
Teeter (1986) N N N Y
Uncles and Stephens (1989) Y N Y N
Wang et al. (1987) Y N N Y
Ward et al. (1984) Y Y Y N
Weisman et al. (1987) Y N N Y
Wells and Kim (1991) Y N Y N
West and Oduyemi (1989) Y N Y N
settling of suspended sediment exemplify the importance of simulating the vertical axis in a numerical sediment transport model.
None of the field and numerical model studies described previously include estuarine wind-wave resuspension, field measurements of suspended-sediment concentration, and a vertical sediment transport model (table 1-1). Studies that include almost all of the elements include those by Hamblin (1989), Costa and Mehta (1990), Sheng et al. (1990b), Mehta (1991), and Sheng et al. (1992). Hamblin studied resuspension by tidal currents near a turbidity maximum, Costa and Mehta studied resuspension by tidal currents in a high tidal energy environment, and Mehta and Sheng and his colleagues studied sediment transport in a large shallow lake. This research used field measurements of suspended-sediment concentration and a vertical sediment transport model to study sediment resuspension by wind waves in an estuary. The numerical model was used as a tool to help analyze the Old Tampa Bay data and to help determine significant sediment transport processes in Old Tampa Bay.
SEDIMENT RESUSPENSION DATA COLLECTION METHODOLOGY
In 1987, the U.S. Geological Survey began a study to determine the effect of fine sediment resuspension on light attenuation in Tampa Bay and to determine the mechanisms that cause resuspension of fine sediments. Light attenuation in the waters of Tampa Bay may adversely affect benthic organisms, seagrasses, and fish and other marine communities that are dependent upon the seagrasses. Resuspension of sediment on the bottom of the bay may contribute to light attenuation, and the mechanisms that cause sediment resuspension in Tampa Bay had not previously been studied. The author was the project chief for the study and his duties included project administration, project planning, data collection, data analysis, presentation of project results at meetings, and report preparation.
During the U.S. Geological Survey study, pressure, water velocity, and suspended-solids concentration data were collected in Old Tampa Bay, a subembayment of Tampa Bay, in order to observe sediment resuspension events and to determine the hydrodynamic mechanisms that cause sediment resuspension. In this research, sediment resuspension data collected from Old Tampa Bay were simulated with a numerical model of vertical one-dimensional hydrodynamics and sediment transport, and the model results were analyzed.
Tampa Bay is located on the west-central coast of Florida as shown in figure 2-1. The estuary is Y-shaped, areally large (about 1000
Figure 2-1, Old Tampa Bay study area.
km2), shallow (average depth 3.6 meters), vertically well-mixed, microtidal (spring tide range about 1 meter), and warm (temperature range from about 14 to 31 C in 1988 and 1989) (Goodwin 1987, Boler 1990). The northwest subembayment is called Old Tampa Bay and the northeast subembayment is called Hillsborough Bay. These subembayments are of most concern ecologically because seagrass loss is more common and water-quality is probably more impacted by point and non-point nutrient loading and by reduced tidal flushing (Lewis et al. 1985, Goodwin 1987). The cities of Tampa, St. Petersburg, and Clearwater are adjacent to the bay. The subtropical weather includes almost daily thunderstorms during the summer, occasional storms from winter cold fronts, and the possibility of tropical storms primarily during the fall. The river inflow is small compared to the volume of the bay, and the riverine flushing time, the bay volume divided by the combined river discharge, is about 2 years (Goodwin 1987). Bottom sediments in Old Tampa Bay are generally silty very fine sands in the deeper water (4 meters) and fine sands in shallow water (less than 2 meters) near the shoreline. Goodell and Gorsline (1961) found clay minerals only in isolated portions of Old Tampa Bay, and the sedimentology has not changed significantly since their study (Schoellhamer 1991a).
Sediment Resuspension Monitoring Sites
State variables in an estuary, such as water velocity, salinity, and suspended-solids concentration, vary spatially and temporally. Unfortunately, it is neither technically or economically feasible to obtain complete spatial and temporal coverage when measuring these variables, so a limited data collection network must be designed that is representative of a large portion of the estuary. Therefore, representative sites for resuspension monitoring were selected.
The most important site selection criterion was that a potential site be at the center of a large area of homogeneous sediment. A potential problem with monitoring sediment resuspension at one site is the possibility that horizontal advection may transport suspended sediment to the site and the resulting increase in suspended-solids concentration may be mistaken for local resuspension. Selection of a site in the center of a large homogeneous area of bed sediments reduces the possibility that more erodible sediments will be transported to the site and insures that the site is representative of a large fraction of the bay bottom. Careful data analysis is required to identify the source of an increase in suspended-solids concentration, and specific data are discussed in chapter 3. Other criteria were 1) nearly uniform residual currents over the homogeneous sediment bed, based upon a depth-averaged two-dimensional barotropic hydrodynamic model by Goodwin (1987), 2) site location far from ship channels for safety and homogeneity of bottom sediments and currents, and 3) a secluded location to reduce vandalism.
The size classification of the bottom sediments in upper Tampa Bay (north of a line that extends approximately east from St. Petersburg) were determined with a fathometer and grab samples in 1987 and 1988 (Schoellhamer 1991a). Finer sediments, which are more easily suspended, are generally found in the deeper parts of the bay. The fine inorganic bed sediments are commonly in the form of fecal pellets and organically bound aggregates (Ross 1975). Coarser sediments found closer to shore in shallower water may experience more wave activity, however, and also are likely to be resuspended. Thus, selection of a deeper site with finer sediments and a shallow site with coarser sediments in Old Tampa Bay was desired.
In Old Tampa Bay, typical deep- and shallow- water sediment resuspension monitoring sites were selected. A deep-water site (average depth about 4 meters) was located in the approximate center of a large area of silty-fine sand at latitude 27'57'01" N and longitude 82*37'55" W. The particle size distribution at this site is shown in figure 2-2, and it has changed little from the 1950's to the 1980's (Goodell and Gorsline 1961, Taylor and Saloman 1969, Schoellhamer 1991a). The mean particle diameter is 127 Am, and 16% of the material is fine material (particle diameter less than 63 ym). The density of the bottom sediments is 2.68 g/cm3 and 2.7%, by weight, of the bottom sediments are organic. A shallow-water site (average depth about 2 meters) was selected 3.0 km south-southwest of the deep-water site in an area of fine sands on the estuarine shoal at latitude 27o55'30' N and longitude 82'38'331 W. The mean particle diameter is 152 pm, and the material is 1.5% fine sediment. A submersible instrument package was used to collect hydrodynamic and sediment resuspension data at the shallow-water site (Schoellhamer 1990), but only a single point velocity was measured so the bottom roughness could not be determined. Therefore, the data collected at the shallow-water site is not wellsuited for numerical modeling and only data collected at the platform were utilized in this research.
Old Tampa Bay Instrumentation Platform
In June 1988, a platform was constructed at the Old Tampa Bay deep-water site in order to support sediment resuspension monitoring instrumentation (fig. 2-3). The platform consists of three vertical pilings that are the apexes of a 3.7 m equilateral triangle (Schoellhamer 1990, Levesque and Schoellhamer in press). The pilings support a triangular galvanized expanded-steel deck approximately 2
I u'J k i i i i il
60 40 20 f 1 1 1 1 1 I | a t I IaI a a a I
GRAIN DIAMETER, IN MICRONS
Figure 2-2, Size distribution of bottom sediment at the Old Tampa Bay
Figure 2-3, Old Tampa Bay platform (Levesque and Schoellhamer,
manuscript in review).
meters (average) above the water surface. Water column instrument sensors were mounted on movable horizontal aluminum arms that were 61cm-long and extended perpendicular to a vertical 610-cm-long, 10-cmdiameter aluminum pipe that was fixed to the center of the steel deck. The entire pipe structure resembled an inverted tree.
Several types of sensors were deployed from the platform. A
biaxial electromagnetic current meter was mounted at the end of each horizontal arm, and an optical backscatterance (OBS) suspended-solids sensor was mounted at the midpoint of each arm. In addition to a current meter and an OBS sensor, one of the horizontal arms also supported a pressure transducer for measuring water depth and wave activity. Wind velocity was measured with a cup anemometer and a wind vane that were mounted at the top of one platform pile, located about 3 meters above the steel deck. The sensor electronics, data recorder, and associated power supplies were housed in an aluminum shelter mounted at one corner of the steel deck. An underwater camera for taking bottom photographs was not deployed because visibility was usually insufficient, especially during resuspension events.
Resuspension monitoring instrumentation consisted of Marsh
McBirney Model 512 biaxial electromagnetic current meters, Downing and Associates Instruments Model OBS-lP backscatterance sensors, and two types of pressure transducers. The biaxial current meters have a 5.1cm-diameter sphere attached very near the end of a 20.3-cm-long metal rod. The biaxial electromagnetic current meters measure water velocity using the Faraday principle of electromagnetic induction, where a conductor (water) moving in a magnetic field (induced by the current sensor) produces a voltage that is proportional to the water velocity. The OBS sensors are thumb-size, and they have an optical window at the
relative position of the thumbnail (Downing et al. 1981, Downing 1983). The optical window is used to transmit an infrared pulse of light that scatters or reflects off particles in the water up to a distance of about 10 to 20 cm at angles up to l40* in front of the window. Some of this scattered or reflected light returns to the optical window where a receiver converts the backscattered light to an output voltage. For well-sorted suspended material, the output voltage is proportional to the suspended-solids concentration and turb idity in the water column. The calibration of the OBS output to suspended-solids concentration varies depending on the size and optical properties of the suspended solids, so the OBS sensors must be calibrated either in the field or in a laboratory with the same suspended material as is found in the field. A laboratory evaluation by Ludwig and Hanes (1990) concluded that instrument response to suspended mud was linear up to a concentration of 3,000 to 4,000 mg/L, and they recommended that OBS sensors not be used for the measurement of suspended sand in areas that concurrently experience suspended mud. Originally, a Geokon vibrating-wire pressure transducer was used at the platform-site and was eventually replaced with a Druck strain-gage transducer for improved reliability and increased sensitivity. Data acquisition, data storage, and sensor timing were controlled by a Campbell Scientific CR10 data logger. A 5minute burst sample of the current meter outputs, OBS sensor outputs, and the pressure transducer output was collected every hour during deployments at the platform-site. The burst sample consisted of 1second data of all sensor outputs for the duration of the 5-minute sampling interval. The burst sample was temporarily stored in the data logger and following the end of the burst sample collection the data was sent to an external data storage module.
The instruments were submerged continuously when initially
deployed in August 1988. Data was transmitted from the platform in Old Tampa Bay to the U.S. Geological Survey office in Tampa by a modem and a cellular phone. Every night, the platform data logger would turn on a cellular telephone. At the same time, a shore-based personal computer would automatically call the cellular phone through a modem, establish communications with the data logger through a modem on the platform, and issue commands for the data logger to transmit data that were then stored by the shore-based personal computer. Approximately 2 hours were required to transmit data from 24 burst samples. Transmission time and power requirements of the cellular phone required the connection of a deep-cycle 12 volt battery to the solar panel on the platform, which was accomplished in mid-October 1988. If the weather conditions were poor during data transmission, the transmission would fail and cause a loss of data. Both the platform and shore-based modems used an error checking protocol to help insure accurate data transmission.
In October 1988, analyses of the available data indicated that the OBS sensors had fouled, and when the sensors were cleaned by SCUBA equipped divers, large amounts of marine growth were observed on all of the sensors. The output from the OBS sensors began to increase as the sensors fouled, usually about 24 to 48 hours after cleaning, and the current meters fouled in about seven days. The OBS sensors were coated with an antifoulant for optical surfaces (Spinard 1987) that only prevented barnacle growth on the optical surface. The cause of the fouling was probably an algal slime that would grow on the face of the sensors and affect their optical properties. In late 1988, when the instruments were submerged continuously, cleaning dives were conducted
about every two weeks. Therefore, the OBS sensors were fouled most of the time, and only data collected within 24 to 48 hours of cleaning were reliable.
The sensor fouling caused by the continuously submerged system proved to be impractical, and a modification was required, so the vertical pipe that supported the instruments was attached to an A-frame and pulley system in December 1988. The vertical pipe was suspended from an A-frame steel-pipe structure secured to the steel deck that allowed the vertical pipe to be raised above the water surface for sensor cleaning and storage. Daily servicing visits to clean the sensors were usually made when the sensors were deployed, so the cellular phone and modem were removed from the platform and the data storage module was exchanged daily during the servicing visits.
Flow around the platform pilings is a potential cause of sediment resuspension that could affect suspended-solids concentrations at the platform, so several steps were taken to reduce this possibility and to determine that any local scour caused by the platform did not significantly affect concentrations at the platform. Barnacles were removed from the platform pilings on February 22, 1990, to reduce their effect on the flow. Scour holes were not observed by divers at the bases of the pilings, possibly due to bioturbation. The bottom of the aluminum pipe that supported the instruments was about 20 cm above the bed to reduce the possibility that it would cause local scour. Four sets of water samples collected at the platform and 750 and 1500 meters south-southwest of the platform on November 30, 1989, July 13, 1990, October 12, 1990, and November 30, 1990, indicate that the average concentration of suspended-solids at the platform was 6.9 mg/L greater than the other sites. In March 1990, however, two water samples were
collected at the platform at the same time on 8 different occasions, and the average concentration difference between concurrently collected sample pairs was 7.1 mg/L. Thus, the higher observed platform concentration is equivalent to the sample concentration variation. Although only a few data points are available, these results indicate that any local scour caused by the platform did not significantly affect suspended-solids concentrations at the platform.
Calibration and Output of-Electromagnetic Current Meters
The relationship between the output voltages of the
electromagnetic current meters and the water velocity must be known. The electromagnetic current meters have two separate output voltages, one for each velocity component, that are linearly related to the water velocity components. Linear calibration equations are used to convert output voltages from the meters to water velocities. After construction and following any repairs, the manufacturer calibrated the meters by adjusting the output voltages to match specifications. About annually, the current meter calibrations were checked by the U.S. Geological Survey hydraulics laboratory at the Stennis Space Center in Mississippi. The calibration check generally agreed within 10% of the manufacturers stated calibration values. All calibrations were for steady flows. If a U.S. Geological Survey calibration was available, then it was used; otherwise, the manufacturer's calibration was used.
A potential limitation of the electromagnetic current meters used in this study is a reduction in the meter's output response to short period water waves, such as wind-waves with 2 to 4 second periods (frequencies 0.25 to 0.5 Hz). The current meter output response to short period water waves is reduced by an electronic filter network that is used to suppress a 60 Hertz carrier signal that is inherent in
the current meter design. At wind-wave frequencies of interest to this study, the gain (output voltage) of the meters is reduced by the electronic filter so that actual velocities are greater than the recorded values.
The recorded velocities can be corrected for the electronic filtering (Guza 1988). The output filter is an active RC (resistor/capacitor) 2-pole filter, 6dB per octave rolloff, with a time constant RC = 0.94 seconds for the meters used in this study. The filter reduces the magnitude of output signal and the reduction increases as the frequency of the input signal increases. For a periodic input signal with angular frequency w, the gain of the filter is
G(w) [1 + (wRC)2] (2-1)
The filter also causes a phase delay of the output signal. The phase delay, a negative number in radians, is
O(w) = tan-1 [1/(wRC)] n/2 (2-2)
Equations 2-1 and 2-2 can be applied to correct the recorded data in the frequency domain. The time series is converted to the frequency domain via the fast Fourier transform (FFT), resulting in the series
a(1) + 2 Z a(n) cos[(n-l) Aw t] + b(n) sin[(n-l) Aw t] =
Z a(n) cos[(n-l) Aw t] + b(n) sin[(n-l) Aw t] +
Z a(n) cos[(n-N-l) Aw t] + b(n) sin[(n-N-l) Aw t] (2-3)
for which N is the number of data points, a power of 2, the angular frequency increment Aw-2r/(NAt), At is the time interval of the data, and the Fourier series coefficients a(2)-a(N), b(2)=-b(N), a(3)-a(N-l), b(3)--b(N-l), . ., a(n)=a(N-n+2), b(n)--b(N-n+2), .,
a(N/2)=a(N/2+2), b(N/2)=-b(N/2+2). The left hand side of the equality is how the Fourier series is commonly presented and the right hand side is how the FFT algorithm represents the Fourier series, which is visually more complicated but computationally more efficient. Each discrete frequency w is represented by a sine and cosine term that can also be written as
a(n) cos(wt) + b(n) sin(wt) = c(n) cos(wt+e(n)) (2-4)
in which the magnitude is
c(n) = ( a(n)2 + b(n)2 ) (2-5)
and the phase is
c(n) = tan1 (b(n)/a(n)) (2-6)
The corrected magnitude at the frequency w is
c (n) = c(n)/G(w) (2-7)
and the corrected phase is
SC(n) = e(n) + O(w) (2-8)
The corrected Fourier series coefficients are
a c(n) C (n) cos[c(n)] (2-9)
b c(n) = c c(n) sin[c (n)] (2-10)
The inverse FFT is then applied to the corrected Fourier coefficients to determine the corrected velocity time series. Guza and Thornton (1980) found that the significant wave height from pressure and velocity sensors agreed within 20%, and similar agreement occurred between pressure and corrected velocity data from platform sensors at the same elevation in November 1990. Raw wave spectra from pressure and corrected velocity data were also in good agreement, so the corrected velocities seem to be reasonable.
Response Threshold and Biological Interference of OBS Sensors
Ambient suspended-solids concentrations were often below the
response threshold of the OBS sensors, and biological interference with the sensors was a potential problem (Schoellhamer, manuscript to be published in Marine Geology). The sediment load of the rivers that flow into Tampa Bay is small and Tampa Bay is microtidal, so the ambient suspended-solids concentrations are also small, about 10 to 50 mg/L. Due to the response threshold of OBS sensors, accurate interpretation of OBS data may be difficult except during episodic events that resuspend bottom sediments. Backscatterance from phytoplankton may be detected when suspended-solids are at ambient concentrations. The OBS sensor electronics were factory adjusted in mid-1989 to improve their sensitivity, which diminished but did not eliminate these problems. Laboratory calibrations of the OBS sensors with bottom sediments from the platform site did not agree with the suspended-solids concentrations of collected water samples, so the sensors were calibrated with water sample data. The standard error of an OBS sensor in Old Tampa Bay is 5.8 mg/L, based upon a set of 21 OBS measurements and water samples collected from an elevation 70.1 cm above the bed by an automatic water sampler during a storm in March 1990 (fig. 2-4). Shallow depths, high water temperatures, and eutrophic conditions encouraged biological growth on the OBS sensors that sometimes increased their output voltages and invalidated the data, so daily cleanings were desirable.
Fish would sometimes interfere with the OBS sensors. During dives and instrument cleaning, it was noted that fish would sometimes be congregated around the instrumentation and occasionally would swim past the OBS sensors. The infrared light pulse would reflect off the fish
40 L 20
35 40 45 50 55 60 65
MEDIAN OBS OUTPUT, IN MILLIVOLTS
Figure 2-4, Suspended-solids concentration and median OBS sensor output
70 cm above the bed at the Old Tampa Bay platform, March 1990 (Schoellhamer, manuscript to be published in Marine
and produce a high spike (short-duration increase in output voltage) during the burst sample collection. Usually no more than a few spikes would occur during a burst, but they were large enough to significantly affect the resulting mean value for the burst, so the median value proved to be a simple and more appropriate measure of the OBS burst average than the mean value. Regular sensor maintenance and careful data analyses to identify spikes and fouling were used to minimize abnormalities in OBS data.
Water Sample Collection
Water samples were collected manually and automatically at the platform. The water samples were usually analyzed by the methods of Fishman and Friedman (1989) to determine specific conductance, turbidity, and concentrations of suspended-solids, volatile suspendedsolids, and dissolved chloride. At the Old Tampa Bay platform, point water samples used for the calibration of the OBS sensors were collected each day from each OBS sensor depth using a peristaltic pump connected to tygon tubing that was attached at the end of a long pole that was lowered to the desired depth.
A continuous water sample collection technique was required for
accurate suspended solids monitoring during storm events and nighttime. An automatic water sampler was secured in one corner of the steel deck of the Old Tampa Bay platform, beginning with deployments in March 1990. A SIGMAMOTOR Model 6601 automatic water sampler, connected to nylon-reinforced teflon-tubing, collected an OBS calibration point sample every hour at one OBS depth for the duration of most instrument deployments. The water sampler was set to sample during the sensor ontime.
Instrument Deployment Strategies
Sediment resuspension monitoring instrumentation was deployed intermittently in Old Tampa Bay from 1988 to 1990. As mentioned previously, instrumentation was continuously deployed from the Old Tampa Bay platform in fall 1988. Analyses of the data indicated that bottom sediment resuspension did not occur with normal or spring tidal currents. Waves generated by strong winds were determined to be the most likely sediment resuspension mechanism in Old Tampa Bay. In Florida, the typical sources for strong winds are winter storm systems, tropical storms, and summer thunderstorms. Therefore, in late 1989 and 1990, instrumentation was deployed in Old Tampa Bay before the anticipated arrival of selected meteorological events. The automatic water sampler was set up at the platform, the instruments were tested, and the vertical pipe that supported the instruments was lowered into the water and secured to the steel deck at the beginning of each deployment. The submersible instrument package was also deployed at the shallow-site shortly after the platform instrumentation was deployed. Weather permitting, daily servicing trips were made to clean sensors, retrieve data, and collect water samples. The vertical pipe was secured out of-the water, and the submersible instrument package was recovered several days after being deployed.
SEDIMENT RESUSPENSION DATA AND ANALYSIS
Data collected intermittently from 1988 to 1990 in Old Tampa Bay
shows that sediment resuspension coincided with wind-waves generated by strong sustained winds associated with storm systems (Schoellhamer 1990, Schoellhamer and Levesque 1991, Schoellhamer manuscript in review). Tidal currents were too weak to resuspend measurable quantities of sediment at the Old Tampa Bay platform, but some bottom sediment motion probably did occur because of the stochastic nature of the process (Lavelle and Mofjeld 1987a). Suspended-solids concentrations returned to ambient values within several (4 to 8) hours as wave activity diminished.
Sediment resuspension data collected at the Old Tampa Bay platform during storms in March 1990 and November 1990 were suitable for numerical simulation, and these data and the analysis of these data by Schoellhamer (manuscript in review) are presented in this chapter. Instrumentation deployments at the Old Tampa Bay platform are summarized in table 3-1. Platform data collected in March 1990 and November 1990 were suitable for simulation because net sediment resuspension occurred and both hydrodynamic and suspended-solids concentration data were collected successfully. Several data sets were collected during which the OBS sensors could not be calibrated, sediment resuspension was not observed, or instruments malfunctioned.
Data collected during two tropical storms unfortunately were not suitable for numerical simulation, but these data and the analysis of
Table 3-1.--Old Tampa Bay platform instrumentation deployments.
Sept 20-21, 1989 Nov 28-30, 1989 March 8-10, 1990 July 11-13, 1990 Oct 9-12, 1990 Nov 28
Dec 3, 1990
Suspended-solids concentration data available
Net sediment resuspension
these data by Schoellhamer (manuscript in review) are presented and compared to the March and November 1990 data. Hydrodynamic data, but no suspended-solids concentration data, were collected successfully during tropical storm Keith November 21-24, 1988. No hydrodynamic data were collected successfully during tropical storm Marco in October 1990. The limited data collected during tropical storms Keith and Marco indicate that tropical storms can resuspend more sediment than winter storms can resuspend.
The bottom roughness at the Old Tampa Bay platform was determined with equation 1-5 and velocity profiles collected during several instrumentation deployments. Velocity profiles that were measured during periods of small wave motion and during relatively strong flood and ebb tides were used. Velocity data collected 183 cm above the bed were not used because the values usually were not logarithmic compared to velocities closer to the bed. Data collected before and immediately after sediment resuspension events did not indicate that bottom roughness changed significantly. Bottom photographs taken by divers were not useful due to poor visibility. Divers observed that the bed was nearly flat with some undulations, possibly from bioturbation, and that there were no regular bed forms. Thus, bed load transport at the platform probably was not significant, except possibly during major sediment resuspension events. The bottom sediment included 16% fine material, and it is possible that this was the material that was observed in suspension and that the sandy material either did not move or did not create ripples that significantly affected the data. The analysis of the velocity data produced an optimal bottom roughness equal to 0.3 cm, and the bottom roughness regime was usually transitional between the rough and smooth limits. The total bottom
roughness is composed of contributions from form drag and grain roughness, but only the roughness associated with the particles determines particle motion (Vanoni 1975, McLean 1991). Spatiallyaveraged grain shear stress can be calculated by collecting velocity data within the flow layer adjacent to the bed that is influenced by grain roughness or by applying empirical relationships based on ripple geometry (Smith and McLean 1977). For this study, however, velocity data could not be collected close enough to the bed to recognize different bottom roughness scales, and no regular ripples were observed. Therefore, in this chapter, the total shear stress calculated with the total bottom roughness was used to determine the sediment resuspension mechanisms at the platform. The spatiallyaveraged grain shear stress was estimated by the numerical model that is presented in chapter 4. For a given particle diameter, a calculated total shear stress that is greater than a critical shear stress (such as Shields critical shear stress, fig. 1-1) may not indicate that motion will occur because not all of the total shear stress is acting on the grains and biological activity may increase the critical shear stress.
March 1990 Storm
Data were collected at the Old Tampa Bay platform after a cold front moved through the Tampa Bay area on March 8, 1990. The high pressure system behind the front generated 8- to 9-meter-per-second sustained northeasterly winds from 1100 to 2100 hours on March 8 (fig. 3-1, table 3-2). Wave activity increased as a result of the sustained northeasterly winds, providing favorable conditions for sediment resuspension. Operational equipment at the platform consisted of electromagnetic current meters at elevations of 70 and 183 cm above the
ORBITAL VELOCITY, IN
CENTIMETERS PER SECOND
o I I
-, L) o M p)
AMPLITUDE. IN CENTIMETERS
I I I
BOTTOM SHEAR STRESS FOR WAVES ONLY, IN DYNES/CM'f
-1--I--a a J -o
IN METERS PER SECOND
40 R( 0 Is cR 40 0 I I I I I
I I i I I I BOTTOM SHEAR STRESS FOR CURRENT ONLY, IN DYNES/CM2
o0 ro uI eI e
MEAN CURRENT SPEED, IN CENTIMETERS PER SECOND
0 U' 0 UR C
0I I 0
CONCENTRATION, IN MG/L
I I I Im
a MAXIMUM BOTTOM SHEAR
STRESS FOR CURRENT
AND WAVES, IN DYNES/CM2
Table 3-2.--Old Tampa Bay platform data, March 8, 1990.
1400 402.4 1500 410.2 1600 409.8 1700 405.7 1800 399.1 1900 395.4 2000 398.1 2100 405.4 2200 413.9 2300 420.3
Mean current (cm/s) at elev 70 cm 183 cm
3.8 4.7 1.4 1.3 3.4 4.8 8.3 9.6 8.9 9.5 2.6 2.7 0.6 4.2 8.7 10.0 7.8 8.7 5.7 7.0
9.2 8.1 8.4 8.6 8.5 8.6 8.1 8.3 6.0 5.8
38.2 30.8 30.0 26.4 27.7 30.8
21.6 26.2 24.8 14.0
2.46/2.43 2.78/2.53 2.64/2.50 2.69/2.52 2.46/2.52 2.75/2.54 2.15/2.41 2.31/2.42 2.27/2.35 2.19/2.58
Suspended-solids conc. (mg/L) 24 cm 70 cm 183 cm
54.0 42.0 44.0 66.0 47.9 44.0 58.0 47.9 49.1 74.0 51.5 47.8 98.0 68.2 59.3 86.0 67.0 63.2 71.9 62.2 52.9 53.9 45.6 44.0 44.0 43.2 40.1 42.0 42.0 36.3
Wave periods are maximum energy of the surface amplitude spectrum/zero upcrossing period of the squared bottom orbital velocity spectrum.
bed and OBS sensors at 24, 70, and 183 cm above the bed. Water samples were collected from 1500 March 8 to 1100 March 9 at a position 70 cm above the bed using the automatic water sampler. In addition, discrete water samples were manually collected during instrument on-time once each day throughout the instrument deployment (March 8 10) at the elevations of the OBS sensors and analyzed for suspended-solids concentration. The discrete samples were used to calibrate the output of the OBS sensors to suspended-solids concentration, and the automatic samples validated the calibration of the OBS sensor 70 cm above the bed. The suspended-solids concentrations (fig. 3-1) peaked at 1800 hours March 8, then decreased rapidly. No data was successfully recorded before 1400 hours on March 8 because of a power supply problem, but the suspended-solids concentrations for the first platform measurement at 1400 hours were slightly greater than the upper limit of the observed ambient concentrations (20 to 40 mg/L), which indicates that only the initial resuspension was missed.
The temporal variation in the bottom shear stresses calculated
from the measured mean current with equation 1-5 did not correspond to the temporal variation of the measured suspended-solids concentrations. The mean current speeds measured 70 and 183 cm and the bottom shear stresses calculated from the mean current speed 70 cm above the bed are shown in figure 3-1. 'A bottom roughness of 0.3 cm was used. A small ebb tide from 1600 hours to 1900 hours March 8 and a small flood time from 1900 hours March 8 to 0100 hours March 9 had maximum speeds of about 9 cm/s and increased the mean current bottom shear stress to about 0.15 dynes/cm 2, but these increases are not correlated with the observed suspended-solids concentrations. A strong ebb tide at the platform during a period of relatively little wave activity from 0200
to 0900 hours March 9 had maximum speeds of 12 to 16 cm/s and increased the mean current bottom shear stress to 0.28 dynes/cm2, but apparently did not resuspend bottom sediments. This total (grain and form) shear stress is smaller than the Shields critical shear stress for the platform sediment. The mean current during the morning of March 9 was relatively large for the platform site, and the lack of a corresponding increase in suspended-solids concentration indicates that the tidal currents did not generate enough shear stress to resuspend bottom sediments at the platform site. The suspended-solids concentrations did not correlate with mean current bottom shear stress (r--0.4, table 3-3).
Horizontal advection is not a likely cause of the observed
increase in suspended-solids concentration because the tidal excursion was within the large area of homogeneous sediments that surrounded the platform and sediments further upcurrent were probably not resuspended. The tidal excursion of the small ebb tide from 1600 to 1900 hours (about 750 meters) was within the large area of silty very fine sands that surrounded the platform. The ebb tidal flow was from the northwest to the southeast at the platform, so the most likely source of resuspended sediment for transport to the platform during an ebb tide was resuspension in relatively shallow water 1500 meters northwest of the platform and within 500 meters of the southern side of the Courtney Campbell Causeway. The wind was from the northeast, however, so wind waves were not approaching the southern side of the causeway, and resuspension was unlikely.
Wave properties were calculated using spectral analysis. Usually, buoy acceleration or pressure data are used to calculate the energy spectrum, but the pressure transducer was only partially responding to
Table 3-3.--Correlation coefficients for various wave properties and bottom shear stresses with suspended-solids concentrations at the Old Tampa Bay platform, March 1990.
Shear stress: mean current only Shear stress: wave only (Kamphius 1975) Shear stress: wave only (Grant and Madsen 1979 and 1982) Shear stress wave-current (Grant and Madsen 1979) Wave amplitude Bottom orbital velocity Square of bottom orbital velocity
Elevation of measured
24 cm 70 cm 183 cm
-0.29 -0.41 -0.52
0.74 0.61 0.70
0.59 0.59 0.66
0.67 0.78 0.70 0.66
the changes in pressure from wave activity, so the velocity component pairs measured by the lowest current meter were used to calculate the energy spectra after correcting the raw data for the electronic output filter of the electromagnetic current meters as described in chapter 2. The wave energy was located almost exclusively at wave periods from 2 to 3 seconds, and the maximum energy period of the surface amplitude spectrum was selected to represent the wave period. The zeroupcrossing period of the squared bottom orbital velocity spectrum (square root of the second moment divided by the zero moment), however, is probably more indicative of wave periods that affect the bottom shear stress. Table 3-2 indicates that these periods are similar, and the interpretations that result from this analysis are not affected by this difference. For narrow banded spectra such as these, the significant wave amplitude is twice the square root of the area under the wave spectrum (Ochi 1990). Significant wave amplitudes calculated with data from the current meter 70 cm above the bed (fig. 3-1) corresponded with the wind speed and decreased after 2200 hours March
8. The wave amplitude was somewhat correlated with the suspendedsolids concentrations (r=0.7, table 3-3). The waves during the storm (1400 to 2100 hours) were transitional between the deep-water and shallow-water limits, but as the wind diminished the waves became deepwater waves. The maximum orbital particle velocities calculated from linear wave theory based upon the significant wave amplitude were close to measured values, so the calculated wave properties appeared to be reasonable. The bottom orbital velocity (fig. 3-1) decreased in conjunction with the wind speed. Note that only a single wave amplitude and period are considered, whereas realistically, there are many periods and amplitudes present in the wave field. Thus, the
results of the spectral analysis were used to provide approximate wave data in a consistent manner.
The bottom orbital velocities corresponded with the observed
suspended-solids concentrations. The bottom orbital velocities were
9.8 to 16 cm/s until 1900 hours March 8 and during this time the suspended-solids concentrations increased from slightly greater than ambient values to the maximum values measured during the deployment. After 1900 hours March 8, the waves were deep-water waves, the bottom orbital velocities were less than 7 cm/s, and the suspended-solids concentrations decreased to ambient values in the early morning of March 9. Thus, resuspension seems to have occurred during the period of greatest wave activity, and the resuspended sediments settled as the wave action diminished. The bottom orbital velocity and squared bottom orbital velocity were somewhat correlated with suspended-solids concentrations (r=0.7, table 3-3).
The calculated bottom orbital velocity and estimated maximum bottom shear stress were more dependent on the water depth and wave period than the wave amplitude. The bottom orbital velocity increases with increasing wave amplitude, increasing wave period, and decreasing mean water depth. A sensitivity analysis was performed to investigate the relative importance of wave amplitude, wave period, and mean water depth on the bottom orbital velocity calculation for this data set. Typical storm values for this data set are a significant wave amplitude of 30 cm, a maximum energy period of 2.6 seconds, and a water depth of 400 cm. A 10% increase in wave amplitude, a 10% increase in wave period, and a 10% decrease in water depth, increase the bottom orbital velocity 10, 29, and 25%, respectively. The bottom orbital velocity during resuspension at the Old Tampa Bay platform was more sensitive to
wave period and mean water depth than wave amplitude because the waves are depth transitional. The maximum bottom shear stress is proportional to the square of the maximum bottom orbital velocity (eqn.
1-1,so an error in the estimated bottom orbital velocity may severely degrade the estimated bottom shear stress. For example, assuming that the wave friction factor is unchanged, if a 10% overestimate of the wave period produces a 29% overestimate of the maximum bottom orbital velocity, then the maximum bottom shear stress will be overestimated by 66%.
The maximum bottom shear stresses estimated for wave motion only are much greater than the bottom shear stresses estimated for the mean current only and correspond to the suspended-solids concentrations. Equation 1-11 and friction coefficients determined empirically (Kamphius 1975) and theoretically (eqn. 1-13, Grant and Madsen 1979, 1982) were used to estimate the maximum bottom shear stress for the observed bottom roughness (fig. 3-1). The estimated bottom shear stress considering wave motion only is much greater than the estimated bottom shear stress considering the mean current only. The greatest wave shear stress occurred during the period of sediment resuspension on March 8 and the wave shear stress is somewhat correlated with suspended-solids concentration (r=0.7, table 3-3).
Poor knowledge of the behavior of the wave friction factor for regime transitional waves may account for the differences between the results of the two methods for calculating the wave only bottom shear stress (fig. 3-1). The waves were transitional between the laminar (smooth bottom) and fully turbulent (rough bottom) flow regimes. Kamphius (1975) states that the data used to determine the friction factors for regime transitional waves are poorly ordered and that the
resulting values should be used with caution. Grant and Madsen (1979) state that their approach is applicable to regime transitional waves but fully rough turbulent flow has previously been assumed (Drake and Gacchione 1986, Drake et al. 1992, Gacchione et al. 1987, Grant and Madsen 1979 and 1982, Signell et al. 1990).
The maximum wave-current bottom shear stresses were estimated with the Grant and Madsen (1979) model (fig. 3-1). Estimated maximum bottom shear stresses were greatest from 1400 to 1900 hours during which time the suspended-solids concentration increased. This qualitative behavior and the correlation coefficient with suspended-solids concentration for the wave-current bottom shear stress (r=0.7, table 33) are virtually identical to that for the maximum bottom shear stresses estimated considering wave motion only.
Compared to the bottom shear stress estimated by the Grant and
Madsen model for wave motion only (eqns. 1-11 and 1-13) and the sum of this wave shear stress and the mean current shear stress (eqn. 1-5), consideration of wave-current interaction slightly increases the estimated bottom shear stress, but this increase is smaller than the uncertainty associated with the wave friction factor. During the period of greatest wave activity from 1400 to 1900 hours, the maximum bottom shear stresses estimated with the wave-current model were 9% greater than those estimated considering waves only (eqns. 1-11 and 113) and 6% greater than the sum of the mean current (eqn. 1-5) and wave only shear stresses. The maximum wave bottom shear stress during this period calculated with the friction factor diagram by Kamphius (1975) is 44% greater than the shear stress calculated with the friction factor of equation 1-13. Therefore, for this data set, the estimated maximum bottom shear stress seems to be more sensitive to the selected
estimation procedure than the possible effect of wave-current interaction.
Because the waves at the platform are depth transitional, the
bottom orbital velocity and estimated wave bottom shear stress are also sensitive to the wave period and water.depth and errors in these quantities can produce large errors in the wave bottom shear stress, as discussed previously. A similar sensitivity analysis of the wavecurrent model using a water depth of 400 cm, wave period of 2.6 seconds, wave amplitude of 30 cm, an angle between the wave and current of 150 degrees, a 0.3 cm bottom roughness, and a mean velocity of 10 cm/s at an elevation 70 cm above the bed indicates that a 10% increase in wave amplitude, a 10% increase in wave period, and a 10% decrease in water depth, increases the maximum wave-current bottom shear stress 14, 39, and 37%, respectively. The maximum wave-current bottom shear stress was relatively insensitive to 10% changes in mean velocity (4%), angle (0.4%), and bottom roughness (4%). Selection of a representative wave period from a measured wave spectrum is probably the most likely source of inaccuracy.
The sediment resuspension observed on March 8, 1990, at the Old Tampa Bay platform was caused by increased wave motion associated with strong and sustained northeasterly wind. The bottom shear stresses estimated by considering the mean current only were much less than the maximum bottom shear stresses estimated by considering wave motion only. Wave-current interaction may have contributed to the bottom shear stress, but this difference is not as significant as the differences associated with the selected wave period and the selected procedure used to calculate the wave friction factor. The period of the largest estimated wave and wave-current shear stresses corresponds