UFL/COEL 2001/007
TRAPPING ORGANICRICH FINE SEDIMENT IN AN ESTUARY
by
Neil Kamal Ganju
Thesis
2001
TRAPPING ORGANICRICH FINE SEDIMENT IN AN ESTUARY
By
NEIL KAMAL GANJU
A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN
PARTIAL FULFILLMENT OF THE REQUIREMENTS
FOR THE DEGREE OF MASTER OF SCIENCE
UNIVERSITY OF FLORIDA
2001
ACKNOWLEDGMENT
I would like to thank Dr. Ashish Mehta for his guidance in my education and
research, as well as the entire Coastal and Oceanographic Engineering Program faculty.
Also deserving praise for their assistance are Dr. John Jaeger, Dr. D. Max Sheppard,
Helen Twedell, Kim Hunt, Vernon Sparkman, Viktor Adams, Jim Joiner, Sidney
Schofield, Jaime MacMahan, and Justin Davis, as well as Leonid Parshukov for
analyzing ADCP and tidal data.
Fernando Marvin should receive recognition for the original development of the
models used in this study and for his support throughout the process. Diana Loomis
deserves kudos for her graphical wizardry and emotional support.
My mother and father merit unlimited praise for providing me with mind, body,
and soul, as do my sisters for helping me develop my mind and soul.
Last, but not least, I would like to thank the eternal and undying Self for
providing the basis for the Universe and everything contained within.
TABLE OF CONTENTS
page
ACKNOW LEDGM ENT............................................................................................... ii
LIST OF TABLES....................................................................................................... vi
LIST OF FIGURES ......................................................................................................... viii
LIST OF SYM BOLS ................................................................................................... xi
ABSTRACT...................................................................................................................... xv
1 INTRODUCTION ..................................................................................................... 1
1.1 Problem Statement............................................................................................ 1
1.2 Problem Approach ............................................................................................ 3
1.3 Outline of Chapters ........................................................................................... 3
2 TRAP EFFICIENCY M ODELING .................................................. ....................... 4
2.1 Introduction....................................................................................................... 4
2.2 Flow M odeling.................................................................................................. 4
2.2.1 Governing Equations ....................................... .............. ......................... 4
2.2.2 M odel Operation ......................................... ................ ............................ 6
2.2.3 Flow Boundary Conditions ................................................................... 6
2.2.4 Flow M odel Input/Output Parameters ....................................... ............. 6
2.3 Sediment Transport M odeling ................................................. ....................... 7
2.3.1 Governing AdvectionDiffusion Equation................................. ............. 7
2.3.2 Erosion Flux.............................................................................................. 7
2.3.3 Deposition Flux......................................................................................... 8
2.3.4 Suspended Sediment Boundary Conditions............................... ............. 9
2.3.5 Sediment M odel Input/Output Parameters................................ .............. 9
2.4 Sedimentation, Sediment Trap, and Trap Efficiency......................................... 10
2.4.1 Sedimentation ......................................................................................... 10
2.4.2 Definition of Trap ...................................................... ........................... 10
2.4.3 Definition of Trap Efficiency..................................................................... 10
2.4.4 Calculation of Trap Efficiency............................ ........................................ 11
2.4.5 Calculation of Trap Efficiency as a Function of Organic Content ............... 11
3 FIELD AND LABORATORY INFORMATION...................................... .......... 12
3.1 Site D escription............................................................................................... 12
3.2 Tributary Flow s............................................................................................... 18
3 .3 T ides...................................................................................................................... 2 1
3.4 C urrents........................................................................................................... 22
3.5 Sedim entation ................................................................................................. 24
3.6 Sediment Rating Curves ............................................................................... 25
3.7 Sediment Composition.................................................................................. 27
3.7.1 Organic Fraction Composition.............................................................. 27
3.7.2 Inorganic Fraction Composition............................................. ........... 28
3.8 Sedim ent Erodibility ....................................................................................... 30
3.9 Settling V velocity ............................................................................................ 32
4 MODEL CALIBRATION ..................................................................................... 35
4.1 Introduction..................................................................................................... 35
4.2 G rid D evelopm ent........................................................................................... 36
4.3 Flow Model Calibration................................................................................ 38
4.3.1 Initial and Boundary Conditions............................................. ........... 38
4.3.2 Comparison of Measured and Simulated Discharges............................... 44
4.3.3 Neap Tide Simulation ........................................................................... 46
4.3.4 Comparison with Analytical Solution for a Cooscillating Tide .................. 49
4.3.5 Tributary Boundary Conditions............................................. ........... .... 51
4.4 Sediment Transport Model Calibration ........................................... .......... 52
4.4.1 Sediment Bed Properties............................................................................. 52
4.4.2 Suspended Sediment Boundary Conditions............................ ............ .. 53
4.4.3 Governing Transport Equation .............................................. ............ 54
4.4.4 Erosion Function Calibration.............................................................. 55
4.4.5 Deposition Function Calibration.......................................................... 55
4.4.6 Sensitivity Analysis for Longitudinal Dispersion Constant ..................... 57
4.4.7 Comparison with Analytical Solution for AdvectionDiffusion................. 58
4.4.8 Comparison with Flume Data for Deposition Under Turbulent Flow.......... 61
5 SEDIMENTATION AND TRAP EFFICIENCY.................................... ........... 63
5.1 Historic Sedimentation Rates........................................................................ 63
5.2 Sediment Trap and Trap Efficiency Calculation ................................................ 63
5.2.1 Trap Efficiency as a Function of C18 Canal Discharge ......................... 63
5.2.2 Trap Efficiency as a Function of C18 Canal Sediment Concentration....... 63
5.2.3 Trap Efficiency as a Function of Organic Content................................... 63
6 CON CLU SION S...................................................................................................... 63
6.1 Sum m ary ......................................................................................................... 63
6.2 C onclusions..................................................................................................... 63
6.3 Recom m endations for Future W ork................................................. .............. 63
REFEREN CES ................................................................................................................. 63
BIO G RA PH ICA L SK ETCH ............................................................................................ 63
LIST OF TABLES
Table page
3.1 Basin areas in the Loxahatchee River estuary watershed....................................... 15
3.2 Chronology of hydrologyrelated events in the Loxahatchee River watershed, 1928
19 8 0 ........................................................................................................................... 18
3.3 Median, high, and maximum flows in three tributaries.......................................... 20
3.4 Spring/neap tidal ranges and phase lags for three gauges....................................... 22
3.5 Selected transect discharges and times of discharge, 08/30/00 ............................ 24
3.6 Shoaling rates in three locations of the estuary .................................... ........... 25
3.7 Median and high flow concentration data and coefficients for Equation 3.3 ........... 26
3.8 Likely organisms in the Loxahatchee River and corresponding organic materials.. 28
3.9 Percent fines and percent organic content for combined samples ........................ 29
3.10 Density measurements of composite sample ....................................... ........... .. 30
3.11 PES test consolidation times, erosion rate constants, shear strengths .................. 32
3.12 Settling column testing parameters................................................................... 33
4.1 Fourier coefficients for tidal forcing at Jupiter Inlet............................. ............ 41
4.2 Comparison of measured (08/30/00) and simulated discharges at two transects ..... 46
4.3 Fourier coefficients for neap tidal forcing at Jupiter Inlet flow boundary ............ 47
4.4 Input sediment concentrations at the tributaries ............................................... 54
5.1 C18 discharges and input concentrations for determination of sedimentation rate as
a function of discharge........................................................................................ 63
5.2 Sedimentation rate in the C18 canal................................................. ............. 63
5.3 Flows and suspended sediment concentrations for trap efficiency as a function of C
18 canal discharge..................................................................................................... 63
5.4 Removal ratio for three C18 discharges, with two different trap locations............. 63
5.5 Simulation parameters for trap efficiency as a function of input sediment
concentration to the C18 canal (Qc18=1.7 m3/s)................................................... 63
5.6 Values used for coefficient a, dry density, and granular density, for simulation of
removal ratio as a function of organic content....................................... ....... 63
LIST OF FIGURES
Figure pae
3.1 Loxahatchee River estuary and tributaries........................................................ 12
3.2 Dredging plans for C18 canal, 1956............................................................. 14
3.3 Loxahatchee River estuary watershed basins, estuarine limits, and central
em baym ent.......................................................................................................... 15
3.4 Normalized salinity in the Northwest Fork vs. normalized freshwater discharge to
the N orthw est Fork. ............................................................................................. 17
3.5a Cumulative frequency distribution of tributary flow, Northwest Fork ................... 19
3.5b Cumulative frequency distribution of tributary flow, North Fork........................ 19
3.5c Cumulative frequency distribution of tributary flow, Southwest Fork..................... 20
3.6 Tributary flow over a one month period, 10/10/8111/10/81............................... 21
3.7 Locations of tide gauges (UFG#), current transects (T#), grab samples (GS#),
suspended sediment data locations (SSD). ........................................ ........... .. 22
3.8 Sample records of tidal measurements at three locations (09/14/0009/15/00)....... 23
3.9 Poling depth bed thicknesses in the C18 canal. .................................. ............ 25
3.10 Erosion rate vs. shear stress for beds with three pretest consolidation times. ........ 31
3.11 Erosion rate constant vs. shear strength............................................. ........... .... 32
3.12 Settling velocity vs. concentration for composite fine sample ............................. 34
4.1 Bathymetry of Loxahatchee River estuary, as used in the flow/sediment transport
m o d el......................................................................................................................... 3 7
4.2 Computational grid, with flow boundaries in black cells, simulated tide gauge and
current transect locations. ...................................................... ........................... 38
4.3 Midtide elevation at UFG1 vs. average wind speed from two offshore buoys....... 40
4.4a Tidal data from UFG1, 09/14/0010/13/00. Raw data ........................................... 42
4.4b Tidal data from UFG1, 09/14/0010/13/00. After midtide trend is removed. ........ 42
4.5 Selection of Manning's n throughout model domain. ............................................ 44
4.6a UFG1 measured tide vs. model result, 09/20/00................................. ........... ... 45
4.6b UFG2 measured tide vs. model result, 09/20/00................................. ........... ... 45
4.6c UFG3 measured tide vs. model result, 09/20/00................................. ........... ... 46
4.7a UFG1 measured neap tide vs. model result, 10/02/00............................................ 47
4.7b UFG2 measured neap tide vs. model result, 10/02/00............................................ 48
4.7c UFG3 measured neap tide vs. model result, 10/02/00............................................ 48
4.8 Definition sketch for tide entering a channel with a reflecting wall..................... 49
4.9 Comparison of Equation 4.3 and model results for C18 canal............................ 51
4.10 Variation of granular, bulk, and dry densities with organic content using data from
three Florida locations and the Loxahatchee River............................... ........... 53
4.1 la Variation of coefficient a (Equation 2.16) with organic content ........................ 56
4.1 lb Variation of settling velocity (Equation 2.16, with C=0.5 kg/m3) with organic
content.................................................................................................................................. 56
4.12 Sedimentation rate in C18 canal as a function of longitudinal dispersion constant,
K L .............................................................................................................................. 5 8
4.13 Definition sketch for substance concentration in uniform 1D channel, with barrier
separating zones of constant concentration and zero concentration at t=0............. 59
4.14 Comparison of analytical solution and numerical model prediction for concentration
in a 1D channel, at t=750 s and 1500 s............................................................... 62
4.15 Comparison of flume data (Krone, 1962) with model result for deposition under
turbulent flow ...................................................................................................... 63
5.1 Sedimentation rate 480 m downstream of the S46 structure in the C18 canal as a
function of canal discharge. .................................................... .......................... 63
5.2 Sedimentation rate as a function of input suspended sediment concentration in the C
18 canal, under a discharge of 1.7 m3/s. ........................................... ............. ... 63
5.3 Sedimentation rate as a function of discharge in the C18 canal, with input
suspended sediment concentration of 0.014 kg/m3............................... ........... 63
5.4 Schematic diagram of a suspended sediment particle subjected to constant flow
velocity and settling velocity. .................................................. ......................... 63
5.5 Historical flow record for C18 canal, from 03/01/8101/18/91........................... 63
5.6 Assumed C18 canal flow record obtained by prorating the measured record for the
Northwest Fork for the period 03/01/8101/18/91............................................... 63
5.7 Cumulative deposition over tenyear period (03/01/8101/18/91) for existing
regulated versus assumed unregulated C18 canal discharge............................... 63
5.8 Portion of the computational grid. .................................................... ........ ..... 63
5.9 Removal ratio in the presence of trap as a function of C18 canal discharge.......... 63
5.10 Schematic diagram of a particle passing over a trap at three different velocities..... 63
5.11 Removal ratio as a function of input suspended sediment concentration to the C18
can al.......................................................................................................................... 6 3
5.12 Difference between influent and effluent loads (trapped load) as a function of input
suspended sediment concentration to the C18 canal............................................. 63
5.13 Removal ratio as a function of organic content at a constant discharge of 1.7 m3/s. 63
5.14 Sedimentation (shoaling) rate in trap as a function of organic content at a constant
discharge of 1.7 m3/s, assuming uniform consolidation......................................... 63
5.15 Mass deposited in the C18 canal (upstream of the trap, Figure 5.8) as a function of
organic content.................................................................................................... 63
5.16 Influent (qi) and effluent (qe) loads vs. organic content, and the difference (qiqe).. 63
6.1 Experimental test pit configuration....................................... ............................ 63
LIST OF SYMBOLS
area
C
Co
Cz
Dx, Dxx, Dxy, Dyx, Dyy
H
He
Ho
K
KL, KT
L
M
Oc
P
Pe
Q
Qd
Qe
Qf
R
sediment concentration (kg/m3)
frictionless wave celerity related to Equation 4.3
Chdzy discharge coefficient
horizontal dispersion coefficients
depth (m)
equilibrium bed elevation related to Equation 6.1
datum bed elevation related to Equation 6.1
sedimentation coefficient related to Equation 6.1
longitudinal and transverse dispersion constants, respectively
channel length related to Equation 4.3
number of time steps in one ebb tidal period
organic content (%)
tidal prism (m3)
Peclet number
discharge (m3/s)
depositional flux (kg/m2s)
erosional flux (kg/m2s)
freshwater discharge (m3/s)
sediment removal ratio
R2 correlation coefficient
Rave ebbtide averaged removal ratio
S sediment source/sink term
Sa salinity (ppt)
SR sedimentation rate (m/d)
T tidal period (s)
U depthaveraged, xdirection velocity (m/s)
Ua,b,bt,c,ct flow velocity related to Figures 5.4 and 5.10
V depthaveraged, ydirection velocity (m/s)
Ws settling velocity (m/s)
Wsf free settling velocity (m/s)
a empirical coefficient related to Equation 2.16
an Fourier coefficients related to Equation 4.2
as empirical coefficient related to Equation 3.2
b empirical coefficient related to Equation 2.16
bs empirical coefficient related to Equation 3.2
g acceleration due to gravity (m/s2)
h water depth
Ah deposit thickness
ht trap depth related to Figure 5.10
k damped wave number
ko frictionless wave number
m empirical coefficient related to Equation 2.16
n Manning's flow resistance coefficient
ns sediment bed porosity
nw empirical coefficient related to Equation 2.16
p probability of deposition
q sediment load (kg/s)
qe influent sediment load (kg/s)
qi effluent sediment load (kg/s)
t time
At time step
Ub bottom, xdirection velocity
Vb bottom, ydirection velocity
we sediment water content
x horizontal coordinate
xs pretrap channel length related to Figure 5.10
xt trap length related to Figure 5.10
y horizontal coordinate
z vertical coordinate
a empirical coefficient related to Equation 5.1
as empirical coefficient related to Equation 2.13
P empirical coefficient related to Equation 5.1
Ps empirical coefficient related to Equation 2.13
Y bottom frictiondependent coefficient
EN erosion rate constant (kg/m2s)
eNO limiting erosion rate constant (kgN/s)
Ti water surface elevation
rIo water surface elevation correction related to Equation 4.1
rTHT water surface elevation at high tide related to Equation 4.1
TL water surface elevation at channel opening related to Equation 4.3
TILT water surface elevation at low tide related to Equation 4.1
Xs empirical coefficient related to Equation 3.5
ut damping coefficient related to Equation 4.3
v eddy viscosity
Pb sediment bulk density (kg/m3)
pd bottom sediment dry density (kg/m3)
ps sediment granular density (kg/m3)
pw water density (kg/m3)
C wave frequency related to Equation 4.3
Tb bed shear stress (Pa)
Td critical shear stress for deposition (Pa)
Ts bed shear strength (Pa)
D solids volume fraction
De limiting solids volume fraction
Xs empirical coefficient related to Equation 3.5
Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Science
TRAPPING ORGANICRICH FINE SEDIMENT IN AN ESTUARY
By
Neil Kamal Ganju
August 2001
Chairman: Ashish J. Mehta
Major Department: Civil and Coastal Engineering
Trenchtraps are frequently implemented to capture incoming sediment before it
can reach the main body of an estuary. Given the presence of organic matter in Florida's
humic estuarine environment, the effect of organic content on the efficiency of an
estuarine trap scheme is the focus of this study. The candidate location, the Loxahatchee
River estuary, is fed by tributaries which contribute organicrich fine sediment. Flow and
sediment transport models were utilized and calibrated using collected field and literature
data. The calibrated models were first used to simulate historic sedimentation rates in the
flowregulated C18 canal, one of the main tributaries. A relation was developed for
sedimentation rate as a function of discharge in this canal. This relation was applied to a
tenyear flow record, to yield a total sedimentation of 0.15 m. To determine what effect
flow regulation has on sedimentation, a hypothetical unregulated C18 canal flow record
was constructed using data from another feeder tributary. In this case the simulated
sedimentation was 47% higher, due to less variable discharge conditions.
A trap was incorporated in the C18 canal, 120 m upstream of the confluence of
the canal and the central embayment of the estuary. Trap efficiency was calculated as a
sediment removal ratio, or the percentage by which influent sediment load to the trap is
reduced in the effluent load from the trap. Trap efficiency modeling was performed for
varying C18 canal discharge, and a specific discharge (1.7 m3/s) was found to maximize
removal ratio. Trap efficiency simulations were also performed for varying input
sediment concentration to the C18 canal, under a constant discharge. Removal ratio
increased with increasing concentration at concentrations below 0.25 kg/m3, but remained
constant above this value. The trapped load (difference between influent and effluent
loads), however, increased with concentration below values of 7 kg/m3, but decreased
above this value, due to the decreased settling velocity in this highconcentration regime.
Simulations were performed for varying organic content, and removal ratio
decreased with increasing organic content, due to decreasing settling velocity.
Sedimentation in the trap increased with organic content, due to decreasing density.
Influent load to the trap increased with organic content, since less sediment was able to
deposit upstream of the trap.
Future work should include characterizing the nature of organicrich sediment
more comprehensively, as well as developing a monitoring scheme to determine actual
sedimentation rates in a test trap at this location.
CHAPTER 1
INTRODUCTION
1.1 Problem Statement
Sedimentation due to the influx of finegrained particles is an issue affecting
numerous waterways and coastal areas. Frequently, these particles originate far inland,
and are transported into the coastal zone by runoff and streamflow (Rusnak, 1967). In the
humic estuarine regime, inorganic sediment is often complemented by a measurable
organic fraction. This fraction can be produced by either autochthonous sources (e.g.,
native phytoplankton, submerged vegetation) or allochthonous sources (e.g., riverborne
phytoplankton, swamp vegetation, windblown material) (Darnell, 1967). The organic
fraction is transported in the same fashion as inorganic sediment, but the lower density
and therefore lower settling velocity renders the organic fraction highly susceptible to
resuspension even under relatively mild flow conditions (Mehta et al., 1997). While the
transport of inorganic fine fractions has been investigated, the effect of organic content in
fines on sedimentation is largely unstudied. In estuarial muds, the organic fraction tends
towards mixing with the other sediment types, thereby altering the transport behavior of
the inorganic fraction.
Given the presence of organic matter, any solution to the sedimentation problem
in a productive estuary must take into account the organic content of the sediment and the
associated transport behavior. One commonly employed solution to reduce sedimentation
is the implementation of a trap scheme by trenching the submerged bottom. To create a
trenchtrap, the depth at the chosen location is increased by dredging. The increased
depth results in a decreased flow velocity, thereby allowing incoming sediment to settle
in the trap itself, instead of being carried further downstream. The sediment can then be
removed from the trap, rather than dredging the otherwise distributed deposit from a
broader area. Given this background, the objective of this study was to determine the role
organic content plays in the trap efficiency of a selected trap design. By holding the trap
depth and location constant and varying only the organic content of the sediment, the
efficiency of a trap can be assessed for different organic contents and flow discharges.
For the present purpose, efficiency will be determined by the sediment removal ratio,
which is the percentage by which the effluent sediment load (leaving the trap) is reduced
with respect to influent load (entering the trap).
The candidate estuary for this study is the Loxahatchee River, located in
Southeast Florida. Shoaling has occurred in the estuary in several spots, mainly at the
confluence of major tributaries in the central embayment of the estuary, where velocities
are typically low (Sonntag and McPherson, 1984). Several tributaries feed the estuary,
carrying freshwater, nutrients, and sediment. Due to the biologically productive terrain
upstream of the estuary, organicrich sediment is present in the loads carried by the
tributaries. To prevent this sediment from accumulating in the central embayment, a "pre
emptive" trap scheme is a viable option. For the present purpose, the trap scheme will be
implemented in one area of the estuary where a flowregulated tributary feeds the central
embayment. The performance of this scheme will be evaluated with regard to the organic
content of the sediment and the flow hydrograph, in order to determine what role organic
content and discharge have on trap efficiency.
1.2 Problem Approach
Several tasks must be undertaken to determine the efficiency of a trap scheme.
These include:
1) Data collection from the field and from the existing literature to characterize the
nature of the flow. This includes measuring tidal elevations and current velocities
in the estuary, and obtaining streamflow data for major tributaries from the
literature.
2) Data collection to characterize the nature of the sediment. This includes collecting
sediment samples and performing laboratory tests on the samples. Historical
suspended sediment concentration data will also be examined.
3) Modeling the flow field via a hydrodynamic model, in order to determine the
velocities as well as the water surface elevations.
4) Development of a sediment transport model to determine suspended sediment
concentrations. This model will incorporate the sediment characteristics
determined from the laboratory tests.
5) A onetrap scheme will be introduced to the calibrated flow model, and the output
from that model will be applied to the calibrated sediment transport model. The
removal ratio of the trap will be indicative of the efficiency of the trap scheme.
6) The influent and effluent sediment loads through the trap will be recorded in order
to quantify trap efficiency.
1.3 Outline of Chapters
Chapter 2 describes the modeling scheme used to evaluate trap efficiency.
Chapter 3 contains the field and laboratory data collected for this study. Chapter 4
describes the calibration of the model, and Chapter 5 discusses the application of the
model to predict sedimentation and trap efficiency. Conclusions are made in Chapter 6,
followed by the bibliography and biographical sketch.
CHAPTER 2
TRAP EFFICIENCY MODELING
2.1 Introduction
Modeling sediment trap efficiency requires the use of flow and sediment transport
models. A flow model provides water velocities and surface elevations, and this output is
applied to the sediment transport model. These models and the method to calculate the
trap efficiency are described next.
2.2 Flow Modeling
2.2.1 Governing Equations
The NavierStokes equations govern the free surface flows of constant density
and incompressible fluids (Pnueli and Gutfinger, 1992). Applying the hydrostatic
pressure distribution assumption yields threedimensional longwave equations, and these
can be vertically integrated to produce the following twodimensional equations (Casulli,
1990), where x and y are horizontal direction coordinates:
xmomentum:
a(HU) a(HUU) +(HUV) gi 9! ( + 9v}U a8 ( 'UU ( yU
_/+ + = gH + vH + vH yU
Bt ax ay x 8x x)x y Oy (2.1)
ymomentum:
a(HV) +(HUV) +(HVV) S 8 V ( yV
+t + = gH + v + vH y
at 8x y y ax 8x) Oy ) (2.2)
5
continuity:
oc a(HU) (HV)
at Nx ay (2.3)
where H is the water depth, U is the verticallyaveraged horizontal xdirection velocity, V
is the verticallyaveraged horizontal ydirection velocity, t is time, g is the acceleration
due to gravity, Tl is the water surface elevation measured from the undisturbed water
surface, v is the eddy viscosity, and y is the bottom friction dependent coefficient defined
as
g u +v2
C (2.4)
where ub and vb are the horizontal x and y bottom velocity components respectively, and
Cz is the Ch6zy discharge coefficient, which is related to Manning's n by
(H + 1)1/3
z n (2.5)
Solving this system of three partial differential equations (Equations 2.1, 2.2, and
2.3) for the three unknowns (U, V, iq) can be accomplished via a numerical method. The
numerical algorithm used is based on the method developed by Casulli (1990). First, a
characteristic analysis is performed on Equations 2.12.3, in order to determine which
terms must be discretized implicitly, such as the water surface elevation (Eqs. 2.1, 2.2),
and the velocity divergence (Equation 2.3). The advective terms are discretized explicitly
using an upwind scheme, which is unconditionally stable when a EulerianLagrangian
method is used to discretize the terms. This method requires the solution of a 5diagonal
matrix at every time step. It is used in conjunction with an alternatingdirection implicit
(ADI) routine, which results in two simpler, linear tridiagonal matrices (Casulli, 1990).
2.2.2 Model Operation
The 2D vertically averaged hydrodynamic model reported by Marvan (2001) and
used in this study is operated through the MATLAB computational application. The use
of MATLAB allows for the generation of the necessary graphics and data output in a
simple fashion, though computational effort is intensive, due to the necessity of large
matrices. A rectangular grid with square elements is used, with numeric "ones" indicating
the body of water, and "zeros" representing land boundaries. A similar grid is needed for
the input bathymetry, with the depth at mean high water entered into each element.
2.2.3 Flow Boundary Conditions
Flow boundaries are indicated by extending water cells to the grid edge. If
freshwater inflow is desired, a permanent velocity can be imposed at the edge,
corresponding to the desired flow condition. If a nonsteady state inflow is desired,
velocity as a function of time can be implemented. For a tidal flow boundary, a function
specifying the water surface elevation at the boundary can be applied. If no velocity or
elevation is specified at cells which terminate at the grid edge, they become noflow
boundaries in the algorithm.
2.2.4 Flow Model Input/Output Parameters
The area and bathymetry grids described in Section 2.2.2 are required to specify
the domain to be modeled. Other required inputs are the tidal forcing function at the
seaward boundary, the calculation time step, the total simulation time, a file containing
Manning's n coefficient values for each cell, and velocities at the tributary flow
boundaries. The output is three matrices consisting of the water surface elevations, x
direction velocities, and ydirection velocities, for every time step in the simulation.
2.3 Sediment Transport Modeling
2.3.1 Governing AdvectionDiffusion Equation
Advectiondiffusion is formulated using a finitevolume explicit method based on
the quadratic upstream interpolation method (QUIKEST) method by Leonard (1977):
aHC a(HUC) a(HVC) ( HC 5HCl 9( HC 9HC
+ + D + Dxxy + Dyx +yy = S
at ax ay ax ax dy ay ax ay (2.6)
where C is the depthaveraged suspended sediment concentration, and Dij are the
dispersion coefficients calculated via Preston (1985) as follows:
KLU2 + KTV2
Dxx = H
CU2 + V2 (2.7)
SKLV2+ KTU2
Dyy = Hf
CzU2 + V2 (2.8)
(KI KT)UV
Dxy = Dyx =g
z U2 + V2 (2.9)
where KL and KT are the longitudinal and transverse dispersion constants, respectively.
The quantity S is a source/sink term, which accounts for erosion and deposition as the
algebraic sum of the upward and downward fluxes as follows:
S= Qe +Qd (2.10)
The erosion and deposition fluxes Qe and Qd are described in the following sections.
2.3.2 Erosion Flux
The upward erosionall) flux at the bed is computed from:
Qe= EN(Cb s) (2.11)
where EN is the erosion rate constant, zb is the bed shear stress, and zs is the bed shear
strength with respect to erosion. The bed shear stress is computed as
pwn2U2
b (H + r)A (2.12)
where pw is the density of water. The shear strength of the bed is calculated via Mehta
and Parchure (2001):
T, = a,(O e)0 (2.13)
where as is an empirical coefficient dependent on sediment type, D is the solids volume
fraction, Oe is the limiting solids volume fraction (value of D when is=0), and Ps is an
empirical coefficient.
2.3.3 Deposition Flux
The depositional flux Qd is a function of probability, settling velocity, and
concentration, defined as (Krone, 1962):
Qd = pWC (2.14)
where Ws is the settling velocity, and p is the probability of deposition, expressed as
I = d (2.15)
where rd is the critical shear stress for deposition. When the bottom shear stress is greater
than or equal to Td, sediment is unable to deposit. This condition is prevented in the
present analysis by setting d above the highest observed shear stress in the flow model
output.
In general, the settling velocity of fine sediment is dependent on concentration. As
a result the settling velocity differs depending on three identifiable regimes: free settling,
flocculation settling, and hindered settling. In the free settling range, relatively low
concentrations permit the individual flocs to settle without interference from other flocs.
The settling velocity in this range is a function of the drag coefficient and the submerged
weight of the floc. As concentration increases, the collision frequency of flocs increases,
resulting in the formation of larger flocs. These flocs are able to settle quicker due to their
increased mass, and characterize the flocculation settling range. Eventually, the
concentration in the water column reaches a point where a floc is unable to settle quickly
due to significant interference from other flocs, and the limited pore space for the fluid.
This interference reaches a maximum when the water column resembles a bed of mud
with negligible settling (Mehta, 1994). Hwang (1989) formulated a fit of the flocculation
and hindered settling ranges, relating settling velocity to concentration as follows:
aCn"
W=
s (C2 +b2)m (2.16)
where a, b, m, and nw are empirical constants. At free settling concentrations (C<0.25
kg/m3) a constant settling velocity (Wsf) is prescribed. Laboratory tests performed in a
settling column are required to determine the sitespecific constants.
2.3.4 Suspended Sediment Boundary Conditions
The boundary conditions at the tributary connections can be expressed as steady
state concentrations, or sediment rating functions can be applied if unsteady tributary
flows are desired. This also holds true at the tidal entrance, where incoming
concentrations can be specified, varying with tidal stage and/or current velocity.
2.3.5 Sediment Model Input/Output Parameters
The flow output is the primary input for the sediment transport model. The user
selected inputs are time step, total simulation time, dispersion constants, bed dry density
and granular density, fluid density, suspended sediment point sources, the coefficients for
Equation 2.16 (for settling behavior), and the coefficients for Equation 2.13 (for erosional
behavior). The area and bathymetry grids from the flow model are used as well.
2.4 Sedimentation, Sediment Trap, and Trap Efficiency
2.4.1 Sedimentation
Sedimentation at any point in the estuary can be calculated from the deposit
thickness Ah given by
ts
A At QdAt
Ah= = :
i=1 Pd (2.17)
where ts is the total simulation time, At is the time step, i is the time step index, Qdi is the
deposition flux, and Pd is the deposit dry density. The sedimentation rate is then
Ah
SR
ts (2.18)
where SR is the sedimentation rate.
2.4.2 Definition of Trap
In this study, a sediment trap is defined as an area of the submerged bottom
deepened to a depth greater than the surrounding bottom, in order to reduce flow velocity.
The lower velocity should allow sediment to deposit in the trap rather than move past and
deposit elsewhere. This in turn allows for maintenance dredging to be performed at a
specific location (the trap) rather than over a broad submerged area.
2.4.3 Definition of Trap Efficiency
Trap efficiency is defined as the percent by which effluent suspended sediment
load is reduced with respect to the influent suspended sediment load (removal ratio). In a
tidal situation, the seaward edge of the trap will be the influent side during flood tide, and
the effluent side during ebb tide, and vice versa for the landward edge.
2.4.4 Calculation of Trap Efficiency
At each time step in the sediment transport simulation the concentration, the
velocity, and the water surface elevation will be calculated in each cell. The cells that
border the trap and are flownormal are also of interest. Sediment loads can be calculated
for these border cells as follows:
q = UCHAx (2.19)
where q is the sediment load, U is the flow velocity, and Ax is the cell width. The
sediment load on each side of the trap will be used to compute the sediment removal ratio
as follows:
qi q
R=
qi (2.20)
where R is the removal ratio, qi is the influent sediment load, and qe is the effluent
sediment load. The removal ratio will be averaged over a tidal cycle, using the removal
ratio values from each time step.
2.4.5 Calculation of Trap Efficiency as a Function of Organic Content
Simulations will be run for differing organic content of the sediment. The average
removal ratio for the each organic content will be compared for a given flow discharge.
The native organic content will be used as the benchmark case by which to assess the
efficiency of the trap under the other organic content cases. The removal ratio as a
function of organic content will be plotted to determine what effect organic content has
on trap efficiency.
CHAPTER 3
FIELD AND LABORATORY INFORMATION
3.1 Site Description
The Loxahatchee River estuary is contained within Palm Beach and Martin
counties in southeast Florida (Figure 3.1).
0 1 embaymenttl 0
a ,n acce b_ Jupie
c jh aner la 12345 I I l
I Ponn
c 42 Turner Quay
ts 3Pmpano U
4 Dolphin
S46 control structure 5 Marlin
Figure 3.1 Loxahatchee River estuary and tributaries.
The river empties into the Atlantic Ocean via Jupiter Inlet. Three main tributaries
feed the estuary: the Northwest Fork, the North Fork, and the Southwest Fork. Jones
Creek and Sims Creek, two lesser tributaries, also feed the estuary. The Southwest Fork's
upstream reach is referred to as the C18 canal, which was created in 1957/58 to lengthen
the Southwest Fork and facilitate drainage of the westward swampland. Flow through the
C18 canal and the Southwest Fork is regulated by the S46 automated sluice gate
structure. The three forks converge on the estuary approximately 3 km west of the inlet,
at the central embayment. The Intracoastal Waterway (ICWW) intersects the estuary in a
dogleg fashion, east of the Florida East Coast Railroad (FECRR) bridge. Five
navigation/access canals exist on the south shore of the central embayment.
Depths in the estuary range from under 1 m to over 5 m in some channel portions,
with an average depth of just over 1 m. The navigation channel (maintained by the Jupiter
Inlet District) runs westward from Jupiter Inlet, under the FECRR bridge, and through the
central embayment, approximately 14 km upstream from the inlet. Flood shoals exist in
the central embayment mainly due to the influx of sand from the ocean, and smaller
shoals exist at the termini of the three main tributaries. Small islands are located west of
the FECRR bridge, on both sides of the channel.
While the Northwest Fork and North Fork are natural tributaries, as mentioned the
Southwest Fork was lengthened westward in 1957/58 by the construction of the C18
canal and S46 control structure (Figure 3.1), in order to divert flow from the Northwest
Fork to the Southwest Fork. A channel was then constructed to allow for the diversion of
flow from the C18 canal to the Northwest Fork. From this point on, the C18 canal will
indicate the narrow channel section of the Southwest Fork, and the broader section will
be referred to as the Southwest Fork (Figure 3.1). Dredging plans for the C18 canal from
1956 are shown in Figure 3.2 (U.S. Army Corps of Engineers, 1956). The existing
bottom was deepened up to 3 m at some locations to facilitate drainage. The datum
(MSL) refers to the National Geodetic Vertical Datum of 1929 (NGVD).
The Loxahatchee River estuary drains over 1,000 km2 of land through the three
main tributaries, the ICWW, and several minor tributaries. The individual basins that
comprise the watershed are shown in Figure 3.3 and listed in Table 3.1. The watershed
occupies residential areas, agricultural areas, and uninhabited marsh and slough areas.
Y+ Natural bottom (pre1957)
E
.4
o I Dredged canal bottom
/ (post1957)
0 500 1000 1500 2000
Distance from S46 structure (m)
Figure 3.2 Dredging plans for C18 canal, 1956.
The creation of the C18 canal has increased the drainage area of the Southwest
Fork. This canal drains the Loxahatchee Slough, a shallow swamplike feature containing
diverse flora and fauna. Upstream of the Northwest Fork is the Loxahatchee River
proper, which meanders through typical South Florida swampland, all contained within
the Jonathan Dickinson State Park (JDSP). The North Fork drains the east section of
JDSP, which contains extensive swampland and scrubland. As a result of the biologically
productive nature of the watershed, organicrich sediment is present in the runoff which
eventually reaches the estuary.
The freshwater nature of these tributaries ends where seawater mainly from
Jupiter Inlet intrudes. Estuarine conditions (Figure 3.3) persist for 8 km up the Southwest
Fork/C18 canal, 9 km up the North Fork, and 16 km up the Northwest Fork (measured
from the inlet) (Russell and McPherson, 1984). Salinities above 1 ppt have been observed
during lowflow conditions in the Northwest Fork, more than 16 km from the inlet.
J" ,)TrA 1 ~.:) n P Bn l
)IC n ':' 4 ..
onafhdl Dickinson n ,,
 j n Ij^ I. *
T ,TTE PAR
i 1 IN
0 '1
"  4^ ; l h *lle '

S/Intracoastal 545IR 4 inn
Basin___ Area (k ,)
C18 278
Jonathan Dickinson 155
South Indian River 65
Loxahatchee River 6
Loaace Rivr
During high freshwater flow the estuarine boundary naturally moves seaward;
during lowflow conditions the opposite is true. Figure 3.4 shows the salinity in the
Northwest Fork normalized by the salinity at Jupiter Inlet (SaNwF/SaJI), versus freshwater
discharge to the Northwest Fork normalized by the maximum flood tide flow rate
(QNWF/QJI). The flood tide flow rate is calculated as
71P
S T (3.1)
where P is the tidal prism, and T is the (semidiurnal) tidal period.
The values used for salinity, tidal period (T), and tidal prism (P) at Jupiter Inlet
are 35 ppt, 12.42 hr, and 2.93 x 108 m3, respectively (Russell and Goodwin, 1987).
Salinity in the Northwest Fork was measured 11 km from Jupiter Inlet (Russell and
McPherson, 1984). The trend in Figure 3.4 implies that salinity in the tributaries can vary
greatly depending on the freshwater flow condition. This is especially relevant when the
tributary flow is regulated (e.g., Southwest Fork), and salinity is influenced by
management practices.
At the turn of the century the Loxahatchee River estuary was a pristine ecosystem
consisting of mangroves, salt marshes, and scrubland. Prior to World War II agricultural
interests transformed the area into a rural landscape with citrus groves and vegetable
farms. Significant residential development has occurred in the area since World War II.
In response to this development the estuary was designated an aquatic preserve in 1984.
Nonetheless, the construction of homes along the shoreline continues. Currently, the
entire estuarial shoreline in the central embayment as well as a significant portion of the
tributary shorelines is residentially occupied. A chronology of noteworthy events
0.8
QjI = 20,587 m3/s
0.7
0.6
0.3
0.2
0.1
0.0000 0.0002 0.0004 0.0006 0.0008 0.0010 0.0012 0.0014
QNWF/QJI
Figure 3.4 Normalized salinity in the Northwest Fork vs. normalized freshwater
discharge to the Northwest Fork.
influencing the hydrology of the Loxahatchee River during the 19281980 period is given
in Table 3.2.
Recreational boating is widely practiced in the estuary by local residents. Access
is necessary to the upstream areas for recreational activities, and also to the open sea and
the ICWW. Many of the natural and artificial access routes have shoaled in recent years
(Antonini et al., 1998), leading to hazardous boating practices such as highspeed
entry/exit to prevent grounding of the vessel. The canals adjacent to the south shore of the
central embayment (Figure 3.1) are especially susceptible to shoaling (Sonntag and
McPherson, 1984), directly affecting boaters who rely on these canals for access.
Aside from boating, the ecological health of the estuary is also sensitive to
shoaling. Seagrass plays an integral part in this environment (Mehta et al., 1990).
U 0.4  1
shoaling. Seagrass plays an integral part in this environment (Mehta et al., 1990).
Table 3.2 Chronology of hydrologyrelated events in the Loxahatchee River watershed,
19281980
Year Event
1928 Small ditch dredged to divert water from Loxahatchee Marsh to
Southwest Fork
1947 Jupiter Inlet permanently stabilized for navigation
19571958 C18 canal created to divert water from Northwest Fork to
Southwest Fork
19701971 Severe drought throughout watershed
1974 C 14 canal created to allow flow to be diverted from C 18 to
Northwest Fork
19761977 U.S. Army Corps of Engineers dredged lower estuary
19771978 Oyster bars near FECRR Bridge dredged to improve navigation
and flushing in the central embayment
1978 Sewage treatment plant began discharging up to 2.5 m3/s to
Northwest Fork
1980 Operation of S46 structure modified to allow for more water
storage in C 18 canal
1980 Three channels dredged in central embayment for improved
navigation
Seagrass coverage maps (Jupiter Inlet District, 1999) indicate shifting and shrinking
seagrass communities in the central embayment.
3.2 Tributary Flows
Tributary flow data were obtained from USGS streamflow gage data, for all years
available (19712000 N.W. Fork, 19801982 N. Fork, 19592000 S.W. Fork). Cumulative
frequency distribution curves have been constructed (Figs. 3.5ac) to designate the
median and extreme flow events (Table 3.3).
The Northwest Fork carries substantially more flow than the North Fork, while
the regulated C18 canal/Southwest Fork carries more flow on average than both. Under
storm conditions the Northwest Fork discharges greater amounts than the other
tributaries.
1
0.9
1 J  
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60
Flow rate (m3Is)
re 3.5 Cumulative frequency distribution of tributary flow a) Northwest Fork.
0.9
0.8
0.7 
0.6
0.5
0.4
0.3 
0.2
0.1
0
0.2 0.4
3 1 1.2
Flow rate (m3/s)
1.4 1.6 1.8 2
Figu
0
C,
.3
t
0)
>
om
I
E
C)
0U
S
C.
0
b) North Fork.
The criterion for opening the gates of the S46 structure is based on water level
behind the structure. When the level exceeds a predetermined mark, the sluice gates are
5 10 15 20 25 30
Flow rate (m3/s)
c) Southwest Fork.
Table 3.3 Median, high, and maximum flows in three tributaries
Tributary Median Flow High Flow Maximum Flow
(50%) (m3/s) (90%) (m3/s) (100%) (m3/s)
Northwest Fork 7.0 x 101 4.1 x 100 6.1 x 101
North Fork 1.0 x 101 2.1 x 10' 1.9 x 100
Southwest Fork 1.3 x 10 7.8 x 100 3.2 x 101
opened until the level recedes by 30 cm (Russell and McPherson, 1984), at which point
the gates are closed. This flow regulation has resulted in a discontinuous flow record;
there have been weeks when no flow has passed the structure, and days when storm flows
have been released. The flows from the North and Northwest Forks are by contrast more
continuous and more immediately responsive to the rainfall/runoff condition. During the
normal wet season, the level behind the S46 structure may not be sufficiently high
enough to release flow, while the other tributaries are freely discharging to the estuary.
Figure 3.6 illustrates this point, showing flow data for a onemonth period from the three
tributaries.
4.5
4
3.5
 NWF
2.5 NF
Sa SWF
M 2
1.5
0 5 10 15 20 25 30
Time (d)
Figure 3.6 Tributary flow over a one month period, 10/10/8111/10/81.
3.3 Tides
Ultrasonic water level gauges (Model #220, Infinities USA, Daytona Beach, FL)
with stilling wells were installed to measure tidal elevations over more than a month's
time. The locations were chosen so as to span a large reach of the estuary, and also to
facilitate gauge placement and retrieval. Thus the eastern gauge was placed on a bridge
pier, as was the gauge in the Southwest Fork. The Northwest Fork gauge was attached to
a navigation post. The gauge locations are shown in Figure 3.7. Characteristic tidal
ranges are given in Table 3.4, and sample records from the three gauges are shown in
Figure 3.8. All tidal elevations use North American Vertical Datum 1988 (NAVD 88) as
the datum. NAVD 88 is the most recent datum, and replaces NGVD. The tidal ranges
indicate the total change in water surface elevation between high and low tide, while the
phase lag indicates the difference in time between high/low tide at gauge UFG1 and the
other gauges. The effect of bottom friction is manifested in reduced tidal range and lag
between UFG1 and the other two gauges.
Figure 3.7 Locations of tide gauges (UFG#), current transects (T#), grab samples (GS#),
suspended sediment data locations (SSD). A= tide gauge, * = current transect, = grab
sample, m = suspended sediment data location.
Table 3.4 Spring/neap tidal ranges and phase lags for three gauges
Phase lag from
Gauge ID Spring tidal Neap tidal UFG1 (min)
range (m) range (m) high tide low tide
UFG1 0.90 0.66 0 0
UFG2 0.85 0.65 21 60
UFG3 0.86 0.64 28 60
3.4 Currents
An Acoustic Doppler Current Profiler (Workhorse 1200kHz ADCP, RD
Instruments, San Diego, CA) was used to profile current velocities at the transects shown
in Figure 3.7. The water velocity was profiled over most of the depth, while data from the
upper 72 cm were lost due to the need to submerge the device at all times.
Simpson and Oltmann (1993) provide a method to interpolate velocity
measurements throughout the water column using a powerlaw velocity profile, in which
the velocity (u) is a function of elevation above bottom (z) to the 1/6th power (i.e., uz/6).
 UFG1
 UFG2
I aUFG3
Time (h)
Figure 3.8 Sample records of tidal measurements at three locations (09/14/0009/15/00).
MSL datum is taken as NAVD 88.
The upper 72 cm were interpolated using this profile equation across an incremental
crosssectional distance, and the resultant velocity profile was integrated vertically from
bottom to surface to yield a mean velocity. With a known crosssectional area, the
discharge through the transect was then estimated. Table 3.5 contains the discharge
through two transects (Figure 3.7), at times after high tide at gauge UFG1. These values
are for ebb flow.
1 1 35 43 45
_ I I _ _ _ _ _
Table 3.5 Selected transect discharges and times of discharge, 08/30/00
Transect Transect Transect maximum Time after high tide Discharge
area (m2) depth (m) (gauge UFG1) (min) (m3/s)
3 472 4.2 233 416
4 398 3.8 260 326
3.5 Sedimentation
With reference to sedimentation, the central embayment is influenced mainly by
the daily flushing due to tide, while the upstream tributary locations are influenced
primarily by riverine discharge and rainfall events. Estimates of the net accumulation of
sediment have been made for three areas of the estuary.
Sonntag and McPherson (1984) estimated shoaling rates in the south shore access
canals (Figure 3.1) to be between 1.53.0 x 102 m/yr. They also estimated shoaling rates
in the central embayment to be as low as 2.5 x 104 m/yr, suggesting that circulation and
flushing in the embayment inhibit substantial accumulation. In the present study poling
depths (obtained by pushing a graduated pole into the bottom until a hard substrate is
reached) in the C18 canal were determined to estimate sedimentation rates along the
length of the canal. Since the bottom was dredged at the time of the construction of the
canal in 1957/58, the bed thickness can be considered to represent the subsequent
accumulation in the interim 42 year period. Figure 3.9 shows these thicknesses along the
canal length. Table 3.6 compares the shoaling rates in the three areas. Sediment thickness
increases with distance from the S46 structure, possibly due to the large erosional forces
near the structure (when flow is released), and the reduction of theses forces as the flow
moves along the canal, allowing more sediment to deposit.
Table 3.6 Shoaling rates in three locations of the estuary
Location Shoaling rate
(m/yr)
C18 canal 2.1 x 102
Access canals 1.53.0 x 102
Central embayment 2.5 x 104
2c
E
UA
.
"0 Mean value=0.92 m
0.5 
S480 m from S46 structure (Section 5.1)
0 200 400 600 800 1000 1200 1400 1600 1800 2000
Distance from S46 (m)
Figure 3.9 Poling depth bed thicknesses in the C18 canal. Straight line indicates mean
trend.
3.6 Sediment Rating Curves
Sediment rating curves relating suspended sediment concentration to discharge
were developed using data from three locations in the estuary (Figure 3.7). Discharge
data from the three tributaries were compared to USGS suspended sediment data for the
same tributaries.
Sonntag and McPherson (1984) reported two values of suspended sediment
concentration (0.059 kg/m3, 0.017 kg/m3) with corresponding flow data for the C18
canal (31 m3/s, 28 m3/s) and a mean concentration value for the duration (198082) of
their study (0.014 kg/m3). The median flow for the C18 canal (1.3 m3/s) from Figure
3.5c was correlated to this mean value of concentration in the present study. For the
Northwest Fork mean and maximum concentration values were reported for the study
period without corresponding flows. The maximum flow during this period (11.2 m3/s)
was recorded by a USGS streamflow gauge and that value was correlated with the
maximum concentration (0.023 kg/m3), while the median flow for the entire flow record
(0.7 m3/s) was correlated with the mean concentration (0.016 kg/m3). The same
procedure was followed for the North Fork. The resulting points were connected to create
sediment rating curves. A fit in the form of
C = asQbs (3.2)
was used (Miiller and F6rstner, 1968), where as and bs are sitespecific coefficients. The
values used and the rating curve coefficients are shown in Table 3.7.
Table 3.7 Median and high flow concentration data and coefficients for Equation 3.2
Median flow High flow (=99%) as bs
Tributary concentration concentration coefficient coefficient
(kg/m3) (kg/m3)
Northwest Fork 0.011 0.023 0.012 0.27
North Fork 0.010 0.018 0.018 0.02
Southwest Fork 0.014 0.059 0.012 0.49
The flows that correlate with the high flow concentrations were of differing
cumulative frequencies, but all were over 99%. Data for the upstream boundary of the
North Fork (JDSP) were not available; values only existed for a location at the opening to
the central embayment (Figure 3.7).
The coefficient as is, in a sense, indicative of the erodibility of the upstream
banks/bed. The North Fork data were obtained at a downstream location, so it is difficult
to make comparisons with the other two tributaries. However, between the Northwest
Fork and Southwest Fork one can infer that the degree of erodibility is equal, as
manifested by the as values.
The exponent bs is indicative of the intensity of the erosional forces present in the
river. These rating curves suggest that the erosional forces of the C18 canal/Southwest
Fork are relatively higher than the Northwest Fork. The course of the Northwest Fork is
far more meandering than the C18 canal/Southwest Fork, and the crosssection is wider
as well. This would indicate a less hydraulically efficient Northwest Fork that is less
capable of eroding and/or transporting sediment than the C18 canal/Southwest Fork.
It is important to note that the transport of fine sediment is largely dependent on
the supply of sediment, which indicates that the rating curves are sensitive to the
availability of fine sediment. If the source of fine sediment for the C18 canal was to be
depleted or enhanced, then the rating curve would be modified to represent, respectively,
less or more suspended sediment concentration for the same discharge.
3.7 Sediment Composition
3.7.1 Organic Fraction Composition
Possible sources of organic material in the Loxahatchee River range from simple
organisms such as protozoa to complex organisms such as vertebrates. As these organic
materials are degraded, they become mixed with the native inorganic sediment, especially
silts and clays. While the specific sources of organic material and their relative
contribution to the organic fraction are unknown, some of the likely sources are listed in
Table 3.8 (Twenhofel, 1950).
Table 3.8 Likely organisms in the Loxahatchee River and corresponding organic
materials
Organism Organic materials produced
Protozoans calcite, chitin, acanthin
Annelids calcite, phosphate, chitin
Mollusks calcite, chitin
Arthropods calcium carbonate, chitin
Vertebrates keratin, cellophane
Plants calcium carbonate, cellulose,
resins, fats, gums, waxes
3.7.2 Inorganic Fraction Composition
Grab samples collected within the estuary were visually classified as sand,
sand/mud mixture, and mud. Samples from the central embayment and the eastern reach
of the estuary were largely sand and shell, while samples from the western reaches
(Figure 3.7) were mainly mixtures of sand and mud. Samples from within the tributaries
and along the banks contained mostly fine sediment.
The inorganic sediment fraction was subjected to xray diffraction to determine
the dominant minerals. Other than the quartz sand fraction, kaolinite, smectite, illite,
pyrite, and rutile were present (in order of decreasing amounts).
Wet sieve analyses (Ingram, 1971) of the finer samples from the western reach of
the estuary were conducted to determine the percentage of fines (<74 microns). In order
to determine the organic content (loss on ignition) of the sediment, the samples were
dried at 550 C, ground in a mortar with a pestle, and heated at 5500 C. The difference in
mass between before and after the hightemperature heating is the mass of the organic
fraction of the sample (Gross, 1971).
Those samples which contained over 59% fines and 10% organic content were
combined as a composite sample representative of the western reach of the estuary. All of
the samples combined happened to be collected upstream of the central embayment,
suggesting that the finer sediments are present in the lowerenergy locations of the
estuary. Table 3.9 contains the individual sample information for the composite sample.
A positive correlation between percentage of fines and organic content is suggested by
these data. The composite sample was tested for bulk (in situ) and dry (0% water content)
densities (Lewis and McConchie, 1994) (Table 3.10). Granular (or particle) density was
then calculated via the mass balance
PdPw
Pw +Pd Pb (3.3)
Additional relations are given to compute the porosity, ns, and the water content, we.
n. =1 Pd
P, (3.4)
nsPw
S(1n)ps (3.5)
Before the individual samples were combined, their mass was recorded (Table
3.8). The mass of each sample was then multiplied by the organic content. This value was
summed for all four samples, and divided by the total composite sample mass. This yields
the organic content for the composite sample (15%).
Table 3.9 Percent fines and percent organic content for combined samples
Sample Percent fines (<74 ulm) Percent organic content Mass contributed to
ID (by weight) (by weight) composite sample (kg)
1/2 59 13 0.98
3/4 63 13 0.92
5 69 18 0.76
6 87 19 0.72
Table 3.10 Density measurements of composite sample
Bulk density Dry density Granular density
(kg/m3) (kg/m3) (kg/m3)
1218 336 2559
1230 428 2038
1336 562 2367
3.8 Sediment Erodibility
The erosion behavior of finegrained sediment in the upper reaches of the estuary
was determined via the Particle Entrainment Simulator (PES). The device, originally
designed by Tsai and Lick (1986), consists of a cylindrical column that is filled with a
slurry of the sediment. The slurry is allowed to settle and consolidate for a specified
period of time (in the reported tests, three consolidation times were used: 24, 72 and 96
hours). Once the requisite consolidation time is achieved, the cylinder is loaded onto the
apparatus, which consists of a perforated cylindrical disk that oscillates within the
cylinder, above the sedimentwater interface. The period of oscillation is measured with a
digital tachometer, and converted to a shear stress value. Samples are taken from the
overlying water at specified time intervals, and analyzed for suspended sediment
concentration via gravimetric analysis. Tsai and Lick (1986) provide a description of the
PES testing procedure. Once the timeconcentrationshear stress data are obtained, plots
of erosion rate or flux (kg/m2s) versus shear stress (Pa) can be developed (Figure 3.10).
For each test, by extending the bestfit line to zero erosion rate, the shear strength
of the bed (with respect to erosion), rs, can be estimated. The data in Figure 3.10 indicate
that longer consolidation times result in a higher shear strength. The erosion rate
constant, EN, is the slope of the erosion rate vs. shear stress curve. An empirical
relationship between erosion rate constant and shear strength in the form
SN = ENOexp(X;S)
(3.6)
0.0007
0.0004
024 hr
0 ,072hr
/A96hr
0.0003
2
0.0002 z 0
0 A
0.0001
0 0A
0
0 0.1 0.2 0.3 0.4 0.5 0.6
Shear stress (Pa)
Figure 3.10 Erosion rate vs. shear stress for beds with three pretest consolidation times.
Intersection of each mean trend line with the shear stress axis gives the shear strength.
The slope of the line gives the erosion rate constant.
can be inferred (Mehta and Parchure, 2001), where ENO is the limiting (Ts=0) erosion rate
constant of 0.2 kg/Ns, and Xs and ks are sedimentspecific coefficients. Equation 3.5 is
plotted in Figure 3.11 along with the results from the PES. The coefficients s and ,s were
found to be 14.0 and 0.44, respectively. Table 3.11 gives the parameters and results of the
tests. The trend in Figure 3.11 implies that sediment beds with higher shear strengths
have lower erosion fluxes than beds with lower shear strength for a given applied shear
stress, as would be expected (Mehta and Parchure, 2001).
I
a' 0.1
z
S 0 Eq. 3.6
o PES
S0.001
0.0001
0 0.05 0.1 0.15 0.2
Shear strength (Pa)
Figure 3.11 Erosion rate constant vs. shear strength.
Table 3.11 PES test consolidation times, erosion rate constants, shear strengths
Consolidation time Erosion rate constant EN Shear strength s
(hr) (kgN/s) (Pa)
24 0.0012 0.110
72 0.00062 0.125
96 0.00060 0.140
3.9 Settling Velocity
The settling velocity can be modeled according to Equation 2.16, for which a
settling column is used to obtain data for the settling velocity at different initial
suspension concentrations. Accordingly, a slurry of native water and the composite fine
sample was vigorously mixed and poured into a 2 m tall and 0.1 m diameter cylindrical
settling column. Withdrawal tubes located at eight elevations allow for sampling of the
slurry through time, and the samples were gravimetrically analyzed for concentration.
The timeconcentrationelevation data were entered into a MATLABbased program
which uses a routine developed by Ross (1988) to calculate the settling velocity at each
elevation and time. The program presents a plot of settling velocity versus concentration,
and a fit of the above equation is obtained by varying the a, b, m, and nw coefficients in
Equation 2.16. Hwang (1989) provides a detailed description of the procedure. The
parameters of the testing performed here are given in Table 3.12. The data from all tests
were combined to yield Figure 3.12.
Table 3.12 Settling column testing parameters
Initial Sampling time Sampling tube
Concentration intervals elevations
(kg/m3) (min) (m from bottom)
18.33 0, 5, 15, 30, 60, 120, 180 0.05, 0.15, 0.3, 0.55, 0.8, 1.05, 1.3, 1.55
9.88 same as above same as above
3.40 same as above same as above
The coefficients used to fit Equation 2.16 to Figure 3.12 were a=0.189, b=6.4,
m=1.8, and nw=1.8. At concentrations below 0.25 kg/m3, a constant (free settling)
velocity (Wsf) is specified. Above that concentration, and below concentrations of 7
kg/m3, the flocculation settling zone is indicated, where particles collide, aggregate and
form larger, faster settling flocs, with increasing concentration. Above 7 kg/m3 the
hindered settling zone occurs, where interference from other flocs reduces settling
velocity with increasing concentration.
0.01
0.001
.2 0.0001 
> /
0
SWs=1.8 x 105 ri/s
 
0.00001
Free settling zone Flocculation settling zone Hindered settling zone
0.000001 .
0.01 0.1 1 10 1(
Concentration (kglm3)
Figure 3.12 Settling velocity vs. concentration for composite fine sample.
CHAPTER 4
MODEL CALIBRATION
4.1 Introduction
Calibrations of the flow and sediment transport models are described in this
chapter. The development of the computational grid is addressed first, followed by the
flow model, including the initial and boundary conditions, and model calibration. The
final section describes the sediment transport model and its operation.
The objective of this development, as noted in Chapter 1, was to evaluate
sediment trap efficiency as a function of organic content. The C18 canal was chosen for
the site of the trap, since this canal seemingly supplies much of the finegrained deposit
found in the central embayment. The function of the trap would be to intercept the
incoming sediment before it reaches the wide region of the central embayment. Thus the
focus in the simulations will be sediment transport from the C18 canal to the central
embayment.
Certain aspects of the estuary have been idealized in model formulation, in order
to minimize computational time and to avoid potential errors. These idealizations are as
follows:
1) Tidal variation in the ICWW is not considered, due to the complexity of
implementing multiple tidal forcing in the model, in the absence of synchronous
tidal data from the ICWW.
2) The north arm of the ICWW is extended to a distance of 420 m only, in order to
minimize computational time.
3) The two lesser tributaries, Sims Creek and Jones Creek, are retained, but not
considered as sources of freshwater input, which is likely to be minor.
4) The three islands west of the FECRR bridge (Figure 3.1) are excluded, since each
is smaller in area than a single cell. It is deemed more accurate to exclude the
islands rather than represent them in an really disproportionate manner.
5) The C18 canal is "bent" to reduce total grid size, in order to reduce
computational time.
4.2 Grid Development
Due to the narrowness of the C18 canal, which is approximately 75 m wide at the
junction with the Southwest Fork and less than 40 m wide at the S46 structure, it was
decided to represent it as a single cell channel, with a cell size of 60 x 60 m. As a single
cell channel, the bending of the C18 canal has no effect on computation. The same cell
size was then conveniently extended to the rest of the model domain. A bathymetry file
was created using available data (Mehta et al., 1990). Data from a 2000 survey (Cutcher,
2000) were compared to the bathymetry, and corrections were made for areas that had
shoaled or eroded during the period between the two surveys. The input bathymetry is
shown in Figure 4.1.
In the computational grid (Figure 4.2), each land cell was assigned the number
zero, while each water cell was assigned a one. Flow boundaries (black cells) were
extended to the edges of the grid, such that "ones" were at the edges of the grid. The
estuary entrance at Jupiter Inlet is a flow boundary, as are the three main tributaries, and
4.5
S4.0
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0 meters
Figure 4.1 Bathymetry of Loxahatchee River estuary, as used in the flow/sediment
transport model.
the two lesser tributaries. The south arm of the ICWW is also a flow boundary. The north
arm of the ICWW was not extended to the grid edge because of the aforementioned
reasons. Jupiter Inlet is at the bottom edge of the grid. The relative locations of the tide
gauges and ADCP current measurement transects are also shown in Figure 4.2.
Sims
Creek
Jones
Creek
E
Iwest FoIII I I
hwestForklC18 canal
Northwest Fork
* Gauge UPG3
Gaug
PUF
G1
Inn
 z
Abv~tranaect3
x Jupiter Inlet
Figure 4.2 Computational grid, with flow boundaries in black cells, simulated tide gauge
and current transect locations.
4.3 Flow Model Calibration
4.3.1 Initial and Boundary Conditions
At the beginning of a simulation, the water surface elevation was chosen to be
uniform, with all cells at high water. In reality the water surface is nonuniform, due to
GaugeUFG2
ADCP ansetFork
ADCP transact 4
ii i
M I i I I
amplitude change and phase lag. Velocities throughout the domain were correspondingly
set to zero at the start of each simulation. It was observed that a full tidal cycle was
required to be simulated before the water surface elevations reached a quasisteady state.
This was verified by recording velocities and water surface elevations at the locations of
the three gauges over multiple tidal periods.
Since tidal forcing at the south arm of the ICWW was not considered, neither
velocity nor water surface elevation was specified at the grid edge. If no specification is
made at grid edge cells, the landward cell edge is treated as a wall in the model (i.e., a no
flow boundary). The computational scheme uses velocity values from adjacent cells, but
also imposes a zero velocity node at the cell edge. The closed boundary of the north arm
of the ICWW (Figure 4.2) is a noflow boundary as well, since land cells are adjacent.
Tidal forcing at Jupiter Inlet (Figure 4.2) is perhaps the most important boundary
condition in this system, because it is the mechanism by which the majority of the water
flows through the estuary. The data obtained from the three gauges (Figure 3.7) were
used to estimate this forcing. The raw data contained a trend in water surface elevation
that was also observed at a station on the northeast Florida coast. The trends followed
similar increases and decreases in midtide elevation, and it was hypothesized that
onshore winds may have created increased elevations in the estuary. The wind record
from two offshore sites (20 and 120 nm east of Cape Canaveral, Florida) was averaged
and correlated with the midtide elevation (Figure 4.3). The data show a positive
correlation between wind speed and midtide elevation.
0.3
0.2
S3 4 5 7 9 10 11 12
0*
0.2
0.3
*.
Wind speed (m/s)
Figure 4.3 Midtide elevation at UFG1 vs. average wind speed from two offshore buoys.
To obtain a tidal record without the effect of this variation, filtering was
performed. For each tidal cycle, the following calculation was made:
7=HT + TLT
0 = 2 (4.1)
where rio is the midtide elevation, *RHT is the water surface elevation at high tide, and iTLT
is the elevation at low tide. The midtide elevation was subtracted from each measured
elevation within the same tidal cycle. The rio variation is also shown in Figure 4.4a.
Figure 4.4b shows the tidal record obtained from the raw data in Figure 4.4a, after
subtracting the midtide trend.
Tidal elevations and discharge measurements were used to calibrate and validate
the flow model. For the dischargeedata to be of use however, the tidal data must be
synchronous. The ADCP current data were obtained while only one gauge (UFG1) was
operating (08/30/00). Once all three gauges were functioning two weeks later (09/13/00),
the UFG1 data during the ADCP measurements were compared with the UFG1 data from
the fully functional period. An appropriately matching (i.e., equal tidal range) tidal record
from the later period was assumed to be representative of the earlier period.
A Fourier analysis (Ochi, 1990) of tidal representation of the form
Stcos2nnnt .si(2nnt I
n(t) = a0 + > a.cos bsn 
n=, ( Ndt) Ndt)) (4.2)
was performed on the entire tidal record (09/13/0010/13/00). In Equation 4.2, rj is the
water surface elevation, ao is the offset elevation, subscript n is the harmonic constituent
index, an and bn are the harmonic coefficients, and dt is the time step. The resulting
equation, carried to N=4, was used as the forcing at Jupiter Inlet. The derived coefficients
are given in Table 4.1.
Table 4.1 Fourier coefficients for tidal forcing at Jupiter Inlet
n an (m) bn (m)
0 0.015 
1 0.416 0.024
2 0.012 0.022
3 0.010 0.0036
4 (=N) 0.0053 0.0068
E Raw data UFG1
o a
So o  Midtide trend
0.2
0.4
0.6
0.8
Time (h)
Figure 4.4 Tidal data from UFG1, 09/14/0010/13/00. a) Raw data. Time origin is 12:00
am.
0.8
0.6
0.4
: 0.2
1 1 2 2 3 3 M41 4 0
I
C 0.2
0.4
0.6
0.8
Time (h)
b) After midtide trend is removed. Time origin is 12:00 am.
The tidal forcing was modified by an amplitude multiplier (1.17) and a phase
correction (0.62 h) until the simulated tide at the UFG1 location matched the UFG1
measured data. This procedure was carried out so that the simulated tidal record closely
matched the measured tidal record from the period of the ADCP current measurements.
For tidal simulation, Manning's n was assumed to be 0.041 throughout the
modeled domain as a first approximation (based on preliminary simulations). This
coefficient is manifested in Equations 2.4 and 2.5, representing the effect of bottom
friction in the verticallyintegrated momentum equations (Eqs. 2.12.2). Chow (1959)
provides qualitative descriptions of waterways and their corresponding n values. For
major streams with irregular and rough sections, a minimum value of 0.035 and a
maximum value of 0.10 are prescribed for n.
In order to correctly simulate the measured tides at the gauge locations UFG2 and
UFG3, bottom friction had to be increased via Manning's n. First, the central embayment
was assigned a higher value (0.052), in order to account for the increased presence of
shoals and depressions in the central embayment. This increase in n enhanced agreement
of tidal elevations at gauge UFG2 (Northwest Fork), though the phase lag of the
simulated tide at UFG3 was not large enough. To increase the phase lag, bottom friction
was further increased by specifying an n value of 0.057 at the throat of the Southwest
Fork, and throughout the C18 canal (Figure 4.5). The increased flow resistance caused
by the constriction of the Southwest Fork into the C18 canal, as well as the presence of
shoals in the area, were represented in the higher n value.
The predicted tides are compared with those measured at the three gauge locations
in Figures 4.6ac. Agreement is best at UFG1, while the simulations at the other two
. n.. O. 07
n=0.052
,All other cells n=0.041
Figure 4.5 Selection of Manning's n throughout model domain.
locations do not entirely agree in amplitude or phase. Also, the filtering process described
earlier introduced some irregularities, which can be seen in the measured data for gauge
UFG3 (Figure 4.6c). Overall, however, the model seems to perform reasonably well in
regard to water level simulation.
4.3.2 Comparison of Measured and Simulated Discharges
Once simulated tidal elevations were adjusted for at all three locations, simulated
discharges through the transects indicated in Table 3.5 and Figure 3.7 were compared
with the corresponding measurements. The transects used in the model domain are shown
in Figure 4.2. At each transect, the discharge through each of the cells were summed to
obtain the total discharge through the transect. The model results thus obtained are
0.5
0.4
0.3
0.2
0.1
E
0 0
S5 10 15 2
W 0.1
0.2
0.3
0.4
0.5
Time (h)
Figure 4.6 Measured tide vs. model result, 09/20/00. a) UFG1
0.5
0.4 m
0.3
0.2
0.1
cU
0 0
B 5 1 15 21
W 0.1
0.2
0.3
0.4
0.5
Time (h)
b) UFG2.
1 0.1 A Gauge UFG3
c Model results
S5 1 15 20 25
W 0.1
0.2
0.3
0.4 VA AV
A A
0.5
Time (h)
c) UFG3.
compared with the ADCPderived data in Table 4.2. Differences between the two sets of
discharges can be attributed to the idealized crosssection in the model, as well as error
resulting from the interpolation process (for the upper 72 cm of water column) described
in Section 3.4.
Table 4.2 Comparison of measured (08/30/00) and simulated discharges at two transects
Transect Time after high tide Measured Simulated % difference*
at UFG1 (min) discharge (m3/s) discharge (m3/s)
3 233 416 455 +9
4 260 326 370 +14
*100% x (simulatedmeasured)/measured
4.3.3 Neap Tide Simulation
In order to validate the model, a similar filtering was carried out for a neap tidal
cycle, using the same amplitude multiplier and phase correction from the previous (spring
tide) analysis. The Fourier coefficients are given in Table 4.3, and the computed tidal
elevations are compared to the measurements (Figs. 4.7ac). Though the simulated tides
are not wholly in phase with the observed tides, the amplitudes are predicted reasonably
well by the model.
Table 4.3 Fourier coefficients for neap tidal forcing at Jupiter Inlet flow boundary
n an (m) bn (m)
0 0.0155 
1 0.314 0.058
2 0.0088 0.022
3 0.0195 0.000085
4 0.0026 0.012
Time (h)
Figure 4.7 Measured neap tide vs. model result, 10/02/00. a) UFG1.
48
0.5
0.4
0.3
0.2
0.1
5M \ 5 /10
0.2
oj 0.1 ,2 i
0.3
0.4
0.5
Time (h)
b) UFG2.
0.5
0.4
0.3
0.2
E 0
0.1
o\
0
I A 5 / 10
\A
0.1
A
0.2
A A
0.3
0.4
Time (h)
c) UFG3.
4.3.4 Comparison with Analytical Solution for a Cooscillating Tide
In order to validate the predictive ability of the model, a scenario with a known
analytic solution was examined. When a tidal wave propagates into a onedimensional
channel bounded by a wall at it's terminus (Figure 4.8), the wave is reflected back,
producing a cooscillating tide. The horizontal velocity at the wall must remain zero,
while maximum velocity occurs at the open end of the channel. At a given location,
maximum velocities occur at onequarter tidal period after high tide, and onequarter tidal
period after low tide.
Reflecting wall
Tidal amplitude at_ ,
channel entrance L M I1
Total channel length  L h Water deplh
Figure 4.8 Definition sketch for tide entering a channel with a reflecting wall.
The analytical solution for velocity in a onedimensional channel with frictional
damping and a reflecting wall is (Ippen and Harleman, 1966):
Lu L ko [e' cos(at kx) e" cost + kx)]
h 0 2 + k2 2(cos2kL + cosh 2L) (4.3)
where u is the velocity, 1TL is the mouth tidal amplitude, h is the mean water depth, Co is
the frictionless tidal wave celerity, ko is the frictionless wave number, k is the damped
wave number, ji is the wave damping coefficient [(k2ko02)12], a is the wave frequency, x
is the channel axis coordinate, and L is the channel length (Figure 4.8). The values used
to solve Equation 4.3 are as follows: rlL=0.45 m, h=1.62 m, Co=4 m/s, ko=3.53 x 105
rad/m, V=2.19 x 105 rad/m, k=4.15 x 10"5 rad/m, L=2200 m, and a=1.405 x 104 rad/s.
The tidal amplitude at the mouth (T0L) was selected based on the observed tidal amplitude
at the mouth of the C18 canal. The water depth was the actual mean depth in the C18
canal, which was used to compute Co [=(gh)1/2] and ko (=o/Co). The length of the channel
was selected to be the actual length of the canal. The wave frequency corresponded to the
semidiurnal (M2) tidal period of 12.42 h.
The damped wave number, k (and therefore t), was varied until the predicted
velocities matched the simulation performed with the flow model. A measure of the
damping is the ratio, ko/k. The selected value of this ratio was 0.85. For comparison,
models of the Panama Canal have yielded damping ratios between 0.79 and 0.91 (Ippen
and Harleman, 1966). Velocities through the channel were calculated at t=T/4.
Model simulation was performed by specifying a zeroflow condition at the S46
control structure (Figure 4.2). The tidal forcing developed in Section 4.3.1 was applied.
The simulated velocities in the C18 canal were recorded onequarter tidal period after
high tide. The comparison between the analytical solution and simulated solution is
shown in Figure 4.9.
At the wall (Figure 4.8) there is no velocity, while the velocities in the canal
between the wall and the entrance to the canal are slightly underestimated by the model.
A lower damping ratio (i.e., more damping) would reduce the analytically solved
velocities at the intermediate canal locations and improve agreement, though the velocity
at the entrance would then be overestimated (in relation to the analytical solution).
0.1
0.09
0.08
0.07
0.06
0.05 _________________ Eq. 4.3
/ Model
S 4 ts
S0.04
0.03
0.02
0.01
0
0 500 1000 1500 2000 2500
Distance from S46 structure (m)
Figure 4.9 Comparison of Equation 4.3 and model results for C1 8 canal.
Overall, this comparison shows that the flow behavior in the C18 canal, under zero net
discharge, closely follows the analytical solution.
4.3.5 Tributary Boundary Conditions
The flow boundary conditions for the three main tributaries were formulated in an
identical manner. For example, the historical Northwest Fork flow data were compiled
and subjected to a frequency analysis, in order to develop a cumulative frequency
distribution (Section 3.2). The discharge (0.7 m3/s) corresponding to the 50% cumulative
frequency (median value) was converted to a velocity by dividing the discharge by the
flow crosssectional area (753 m2). That velocity (0.00093 m/s) was then specified at the
flow boundary, at each boundary cell. The same process was repeated for the North Fork
and the Southwest Fork, with their respective 50% discharges. To obtain boundary
conditions for highflow scenarios, the same process was done using the 100% discharge
(the highest observed value).
Sims Creek and Jones Creek were treated as noflow boundaries, since no
velocity or elevation was imposed at the grid edge. This results in essentially a wall
condition. The short lengths of these tributaries (300 m for Sims Creek, 900 m for Jones
Creek) suggests that a wall condition may not be wholly inaccurate.
4.4 Sediment Transport Model Calibration
4.4.1 Sediment Bed Properties
In order to meet the objective of this study, the density of the bed as a function of
organic content was determined. Such a relation was developed for fine sediments from
three sites in Florida (Rodriguez et al., 1997, Mehta et al., 1994), as well as from the
Loxahatchee River. The three sites are the Lower Kissimmee River, the Taylor Creek
Nubbin Slough basins, and the Rodman Reservoir. All locations harbor organicrich fine
sediments mixed with sand, as does the Loxahatchee River.
The granular density of sediment was found to vary with organic content as
follows:
p, = 16.5(Oc) + 2650 (4.4)
yielding an appropriate density of 2,650 kg/m3 for a sediment with no organic content
(Oc=0%). The bulk density was found to vary according to:
Pb = 1568e0.l8(c) 0.9(Oc)+1114 (4.5)
The dry density was then computed via the mass balance
(Pb Pw)Ps
d (P Pw) (4.6)
Equations 4.4, 4.5, and 4.6, and the data used to create them are shown in Figure 4.10,
including the samples analyzed in the present study (Oc=15%). The granular, bulk, and
dry densities decrease with increasing organic content, though the dry and bulk densities
level off to near constant values (112 kg/m3 and 1,051 kg/m3, respectively) at an organic
content greater than 70%.
3000
2500 + *
/ ** *"t $
S2000
S* Granular density
1500 _
S 4. Bulk density
a 1000 oo Dry density
A + Loxahatchee River
500 4
^A A AA AA A A
0
0 10 20 30 40 50 60 70
Organic content (%)
Figure 4.10 Variation of granular, bulk, and dry densities with organic content using data
from three Florida locations and the Loxahatchee River.
4.4.2 Suspended Sediment Boundary Conditions
To simulate the input of sediment into the estuary, suspended sediment
concentrations must be specified at the appropriate boundary cells. At the Southwest
Fork/C18 canal flow boundary (Figure 4.2), Equation 3.2 with the appropriate
coefficients (Table 3.7) was applied. At the Northwest Fork the same equation was
similarly used. Sediment input from the North Fork was not considered, since the data
used to obtain the rating curve were from a central embayment location (Figure 3.7). Due
to the relatively low flow of the North Fork, it is reasonable to assume that the main
source of suspended sediment at the data location was the inlet, and not an upstream
source.
Sediment inputs at the Jupiter Inlet/ICWW boundaries were not considered, since
fine sand is the major component of the suspended sediment load at those locations
(Sonntag and McPherson, 1984). Sims Creek and Jones Creek (Figures 3.1, 4.2) were
assumed to contribute negligible fine sediment loads. Table 4.4 shows the locations
where sediment inputs were implemented, and the corresponding suspended sediment
concentrations for median and maximum flows.
Table 4.4 Input sediment concentrations at the tributaries
Tributary Median flow (50%) Maximum flow (100%)
concentration (kg/m3) concentration (kg/m3)
Northwest Fork 0.011 0.030
Southwest Fork 0.014 0.066
North Fork 0.000 0.000
Sims Creek 0.000 0.000
Jones Creek 0.000 0.000
4.4.3 Governing Transport Equation
The governing advectiondiffusion transport equation (Equation 2.6) requires
values of the longitudinal and transverse dispersion constants, KL and KT. Marvin (2001)
selected values of 13 and 1.2, respectively, when modeling the Cedar/Ortega River
system in northern Florida. These values were adopted in the present study, and a
sensitivity analysis was performed by varying KL and simulating deposition in the C18
canal under identical flows and input suspended sediment concentrations. This analysis is
described later in Section 4.4.5.
4.4.4 Erosion Function Calibration
The bed shear strength (with respect to erosion) is calculated via Equation 2.13
(Mehta and Parchure, 2001), with the coefficients related to organic content. Note
however that in what follows the erosion function has been disabled for all simulations,
since the organicrich fine sediment was only present in the upstream reaches of the
estuary.
4.4.5 Deposition Function Calibration
The probability of deposition, p (Equation 2.15), requires a value for the critical
shear stress for deposition. This is set to the highest observed shear stress so that
deposition is possible at all times. In these simulations, Td was set to 20 Pa.
The settling velocity is characterized by Equation 2.16. The velocity scaling
coefficient, a, is used to account for the variation of organic content, with higher organic
content resulting in lower settling velocity. Using avalues from Lake Okeechobee
(Hwang, 1989), the Ortega River (Marvin, 2001), and the Loxahatchee River, the
variation of "a" with organic content (Oc in %) was formulated as follows:
a = 1.3 x 107(Oc)4 + 7.1 x 106(Oc) 1.7 x 104(Oc)2 + 6.6 x 104(Oc) + 0.2 (4.7)
Equation 4.7 is applicable in the range of Oc between 0 and 44% and is plotted in Figure
4.1 la. Values for the other coefficients of Equation 2.16 were found to be b=6.4, m=1.8,
and nw=1.8 (Section 3.9). The use of these coefficients in Equation 2.16 yields the
settling velocity in m/s. Figure 4.11b illustrates the variation of this velocity (at a
representative concentration C=0.5 kg/m3) with organic content. As expected, the settling
0
14
3 0.1
o
0
0.01
'' *.
10 20 30 40 5
Organic content (%)
Figure 4.11 Variation of settling velocity parameters with organic content. a) coefficient a
8.00E05
7.00E05 "
6.00E05
E 5.00E05 
S4.00E05
3.00E05
2 00E05 
0 5 10 15 20 25
Organic content (%)
30 35 40 45
 Eq. 4.7
B Hwang (1989)
* Marvan (2001)
+ Present study
1.00E05
0.00E+00
_~WHi
b) Settling velocity.
velocity decreases with increasing organic content. For example, a twofold increase in
organic content, from 20% to 40%, yields a 56% decrease in settling velocity.
4.4.6 Sensitivity Analysis for Longitudinal Dispersion Constant
The selection of the dispersion constants (Equations 2.72.9) was based on a
sensitivity analysis to determine if the values used by Marvin (2001) were reasonable for
this study. In the C18 canal (Figure 4.2), only longitudinal dispersion has an effect on
the transport of sediment, because of the onecell width of this canal. Sedimentation rate
was chosen as the result of interest, since it is the primary focus in the canal.
Sedimentation rate in the C18 canal was calculated via Equations 2.172.18 for
three different values of KL, and the calculated rate is shown in Figure 4.12. The plot
shows that even with a 27fold increase in KL, the sedimentation rate decreases only 7%
at the location (60 m) of the greatest difference between the results. This implies that the
results of a given simulation are not greatly affected by the value of KL, especially in this
range of values (2>KL>55). Therefore the original values of KL and KT were retained (13
and 1.2, respectively).
In order to explain the relatively minor differences in sedimentation rate with
varying KL, the Peclet number can be calculated for these simulations as
LU
Pe = LU
Dx (4.8)
where L is the channel length, U is the xdirection velocity, and Dxx is the dispersion
coefficient in the xdirection (Equation 2.7). This dimensionless number represents the
ratio of advection to diffusion. Therefore, for Pe>l, advection is the dominant transport
mechanism, and for Pe
m, U=0.014 m/s (median C18 flow velocity), and the varying KL values to compute Dxx,
0.00012
0.0001
E 0.00008
KL=2
O 0.00006
U K,=13
d$K1=55
M 0.00004
0.00002
0
0 200 400 600 800 1000 1200
Distance from S46 structure (m)
Figure 4.12 Sedimentation rate in C18 canal as a function of longitudinal dispersion
constant, KL.
Pe varies from 24 to 672, indicating that advection is dominant, under these values of KL.
Therefore the transport behavior is not greatly affected by KL.
4.4.7 Comparison with Analytical Solution for AdvectionDiffusion
In order to verify the performance of the advectiondiffusion mechanism of the
model, a comparison was made between an analytical solution and model prediction. The
transport of a conservative substance in a 1D channel is governed by the equation
(Harleman, 1966):
ac ac a D
+U x Da a )
at ax ax ( ax (4.9)
where C is the concentration of the substance, U is the horizontal velocity, and Dx is the
xdirection turbulent diffusion coefficient. A relevant case is a channel separated by a
barrier, with the upstream end containing a steady substance concentration Co, and the
downstream end having no concentration (Figure 4.13). At time t=O the barrier is lifted
and the motion of the substance in the downstream direction at a velocity U is tracked.
The initial and boundary conditions are:
C(O,t)=Co t> 0
C(x,) = 0 x > 0
C(o, t) = 0 t 0
C=Co Barrier (at t=0)
I
 _so r_"; = C t
x=0 +x
Figure 4.13 Definition sketch for substance concentration in uniform 1D channel, with
barrier separating zones of constant concentration and zero concentration at t=0.
which indicate a constant concentration at x=0, at all times. When the barrier is lifted (at
t=0), the solution for concentration at any distance and subsequent time is given by
(Harleman, 1966):
Ux
1 Dx X + Ut 1 x Ut
C(x, t) = Co IeD erfc +t + )erf (4.10)
2 2D^J 2 2VDt (4.10)
Using Co=0.066 kg/m3, U=0.32 m/s, and Dx=30 m2/s, the concentrations at t=750 s and
t=1,500 s were calculated for x=0 to 1,200 m. The concentration and velocity values
correspond to the maximum flow condition in the C18 canal. The resulting concentration
profiles are plotted in Figure 4.14.
An analogous situation was simulated using the numerical model by eliminating
tidal forcing, and specifying a constant discharge (and hence velocity) and input
suspended sediment concentration at the S46 structure. The deposition and erosion
functions were disabled so that strictly transport was modeled. The discharge was
maintained at 31 m3/s (100% flow), corresponding to a velocity of 0.32 m/s, and a
constant sediment concentration input of 0.066 kg/m3 was imposed on the C18 canal
flow boundary, as computed via the sediment rating curve for the C18 (Equation 3.2).
The model was run for 1,500 s, and the concentrations in the C18 canal were recorded.
Figure 4.14 shows the comparison between the analytical solution and model prediction.
The t=750 s comparison shows that the numerical model allows the concentration front to
move quicker (0.08 m/s faster at C=0.007 kg/m3) than the analytical solution. The
comparison at t=1,500 s shows better agreement for the concentration predictions (no
difference at C=0.007 kg/m3).
The discrepancy between the model versus analytical results might be explained
by the numerical scheme, which causes oscillations in velocity, which deviate
approximately 4% from mean (at 100% C18 discharge), at a frequency of 0.3077
cycles/h. While the dispersion coefficients (Equations 2.72.9) in the model are functions
of the Chdzy coefficient, velocity, depth, and the dispersion constants, the analytical
solution requires only a constant dispersion coefficient. Therefore an oscillating velocity
would also cause the dispersion coefficients to oscillate. Nevertheless, the dispersion
constant KL from the numerical model was modified for this comparison so that Dxx
(Equation 2.7) was equal to Dx from the analytical solution (30 m2/s).
4.4.8 Comparison with Flume Data for Deposition Under Turbulent Flow
The deposition algorithm of the model was used to compare the output with
laboratory data for deposition in a flume under turbulent flow, in order to validate the
algorithm.
Deposition of fine cohesive sediment under turbulent flow was investigated by
Krone (1962) in a recirculating laboratory flume, using sediment from the San Francisco
Bay. Under high flow conditions (no deposition) and a water depth of 0.3 m, sediment
was suspended throughout the flume at the beginning of the test, and then the velocity
was reduced to a constant value so as to allow for deposition of the suspended material.
The suspension was recirculated so that the sediment concentration at the beginning of
the flume was equal to the concentration at the end of the flume. Tests were conducted
for times up to 200 h. The suspended sediment concentration was measured in the return
flow through time, at five different flow velocities. At concentrations above 0.3 kg/m3,
two distinct deposition phases were observed, corresponding to differing suspension
concentrations. Each phase was characterized by a different deposition rate. Below 0.3
kg/m3 a linear relationship was found between deposition rate and bed shear stress (Tb).
This trend was extended to zero shear stress, to yield a deposition rate that corresponded
0.07
0.06
0.05
S0.04 Eq. 4.10, t=750 s
 Eq. 4.10, t=1,500
 2D model, t=750 s
0.03 I2D model, t=1,500 s
0
0.02
0.01
0 200 400 600 800 1000 1200
Distance (m)
Figure 4.14 Comparison of analytical solution and numerical model prediction for
concentration in a 1D channel, at t=750 s and 1500 s.
to a constant settling velocity, i.e., dC/dt Ws/h. The sediment concentration through
time in this regime (<0.3 kg/m3) was related to a Ws as follows:
= C exp Ph (4.11)
where C is the instantaneous concentration, Co is the initial concentration, p is the
probability of deposition (Equation 2.15), t is time, Ws is the (constant) settling velocity,
and h is the water depth. From data taken after concentrations had reached 0.3 kg/m3, the
settling velocity was found to be 6.6 x 106 m/s, and the probability of deposition was
found to be 1(rb/0.06), where 0.06 Pa is the critical shear stress for deposition. One test
was used for comparison in the present study. In that test the flow velocity was 0.134
m/s, and a suspended sediment concentration of 0.3 kg/m3 was observed after 43 h of the
117 h test. Therefore only the last 74 h were used in this comparison.
To recreate the above laboratory conditions in the sediment transport model, the
depth in the C18 canal was reduced to 0.3 m, the discharge was modified to produce a
flow velocity of 0.134 m/s, and initial concentration in the canal was selected to be 0.3
kg/m3. The value of rd was changed from 20 Pa (Section 4.4.5) to 0.06 Pa (Equation 2.15,
Section 2.3.3), and the erosion function was disabled. The value of tb (0.05 Pa) was
determined via Equation 2.12. After the first time step, the recirculating aspect of the
flume was mimicked by using the concentration at the end of the canal as the input
concentration at the beginning of the canal for every successive time step. As noted, only
the constant settling velocity regime was modeled, by using the flumederived value of
6.6 x 106 m/s. Concentrations were recorded for 74 h, which corresponds to the duration
of the flume test. The model result is plotted against flume data in Figure 4.15.
In Figure 4.15, the model result follows the exponential decay observed in the
flume tests, which should occur when a constant settling velocity is prescribed according
to Equation 4.12. It should be noted that if a concentration or timedependent settling
velocity were introduced, the decay in sediment concentration would not be exponential.
The comparison does show that the sediment transport model accurately predicts
depositional behavior under turbulent flow, which is essential for the simulations
described next in Chapter 5.
 Model results
* Flume data
0 10 20 30 40 50 60 70 80
Time (h)
Figure 4.15 Comparison of flume data (Krone,
under turbulent flow.
1962) with model result for deposition
CHAPTER 5
SEDIMENTATION AND TRAP EFFICIENCY
5.1 Historic Sedimentation Rates
The calibrated flow and sediment transport models were applied to quantify the
sedimentation rate as a function of discharge in the C18 canal. Flow velocities and water
elevations were obtained by running the flow model for various C18 discharges. These
velocities and elevations were used in the sediment transport model to simulate
sedimentation rates at a selected location in the canal. The sediment rating curve for the
canal (Equation 3.2) was used to determine the input suspended sediment concentration
at the S46 boundary (Figure 4.2). The selected flows and concentrations are given in
Table 5.1. The Northwest Fork and the North Fork were maintained at their respective
50% flow conditions (Table 3.3) along with the corresponding input concentrations
(Table 3.7). This was done because the S46 structure frequently releases flows to the C
18 canal that are not synchronous with the other tributaries (Figure 3.6).
Table 5.1 C18 discharges and input concentrations for determination of sedimentation
rate as a function of discharge
C18 discharge C18 input suspended sediment
(m3/s) concentration (kg/m3)
1.5 0.012
10.5 0.038
19.0 0.051
37.0 0.070
The selected location to calculate the sedimentation rate (Equation 2.18) was 480
m (8 grid cells) downstream of the S46 structure. This site was selected because the
poling depth data (Figure 3.9) contained a measurement (0.63 m) which fell near the
mean poling depth trend (0.67 m) at this approximate location. The simulations for the
eight discharges were run for 5 tidal cycles, and the sedimentation rate was calculated at
the end of each simulation. An equation was fit to the result, with the form
SR = [l exp(pQcs1)] (5.1)
where SR is the sedimentation rate in m/d, a and P are sitespecific coefficients, and Qc18
is the C18 discharge in m3/s. Bestfit values of coefficients a and P were found to be 3.0
x 104 and 0.138, respectively. Figure 5.1 shows the model results and the bestfit
Equation 5.1.
3.50E04
3.00E04
2.50E04
5 2.00E04
C Model results
.9 Eq. 5.1
r 1.50E04
E
"0
C0 1.00E04
5.00E05
0.OOE+00
0 5 10 15 20 25 30 35 40
C18 discharge (m3/s)
Figure 5.1 Sedimentation rate 480 m downstream of the S46 structure in the C18 canal
as a function of canal discharge.
The results in Figure 5.1 imply that as discharge increases the change in
sedimentation rate decreases. For example, a doubling of flow from 2.5 to 5 m3/s results
in a 71% increase in sedimentation rate, while a similar twofold increase from 10 to 20
m3/s results in a 25% increase in sedimentation rate. At lower flows, the sedimentation
rate is much more sensitive to discharge than at higher flows. This is explained in the
following analysis.
In Figure 5.1, the trend of increasing sedimentation is evidently related to
discharge; however, increasing discharge also increases concentration via the sediment
rating curve (Equation 3.2). To separate the two effects, two scenarios were simulated.
Using the same C18 discharge (1.7 m3/s), different input concentrations were
implemented in the C18 canal, and the sedimentation rate curve was plotted against
concentration (Figure 5.2). As observed, the input concentration has a linear correlation
with sedimentation rate, due to the linear increase in depositional flux resulting from
increased concentration (Equation 2.14).
Simulations were then performed with constant input concentration (0.014
kg/m3), under differing flows. The results are shown in Figure 5.3. These simulations
show that at flows below 12 m3/s, increasing the discharge increases the sedimentation
rate, but above this discharge the sedimentation rate decreases. This trend can be
explained by taking an individual suspended particle, and tracking it under different
(constant) flow velocities (Figure 5.4).
S1.20E04
1.00E04
8.00E05
4.00E05
42 .OOE05
00.E+00
0 0.01 0.02 0.03 0.04 0.05 0.06
Concentration (kglm3)
Figure 5.2 Sedimentation rate as a function of input suspended sediment concentration in
the C18 canal, under a discharge of 1.7 m3/s.
9.00E05
8.00E05
7.00E05
S 6.00E05
2 5.00E05
C
4.00E05 
E
3.00E05
2.00E05
1.00E05
0.OOE+00
0 5 10 15 20 25 30 35 40
C18 discharge (m'/s)
Figure 5.3 Sedimentation rate as a function of discharge in the C18 canal, with input
suspended sediment concentration of 0.014 kg/m3.
Particle
W,
x=O =a \ x=b x=C
Track of particle (under U=Ub
Figure 5.4 Schematic diagram of a suspended sediment particle subjected to constant
flow velocity and settling velocity.
As seen in Figure 5.4, as a particle at x=0 and suspended at the water surface
settles at a rate equal to the settling velocity Ws, it is moved longitudinally by the water
velocity U. Thus, for example, under a velocity Ua=0.051 m/s (corresponding to a C18
discharge of 5 m3/s), a particle with settling velocity Ws=l x 105 m/s and at the water
surface will deposit 8,262 m downstream from it's original location, given the canal
water depth h=1.62 m. In contrast, under a velocity Uc=0.1 m/s (discharge=10 m3/s), the
same particle will deposit 16,686 m downstream. Finally, under velocity Ub=0.075 m/s
(discharge=7 m3/s), the particle will deposit 12,474 m downstream. Therefore if the
sedimentation rate is measured at x=b, then one can expect that this rate will be
maximum under a velocity of Ub.
Equation 5.1 was applied to a historical flow record for the C18 canal (03/01/81
01/18/91) using the daily discharge values (QC18) to determine the daily sedimentation
rate (SR). The sum of daily sedimentation over ten years yields the tenyear
sedimentation. The flow record used is shown in Figure 5.5.
The regulation of the C18 canal by the S46 structure is manifested in the high
frequency of zerodischarge periods (54% of the days) and spikes. The deposition rate
was found to be 0.15 m over the tenyear period (0.015 m/yr). The deposition rate
estimate made via poling depths in the C18 canal (Section 3.5) was 0.021 m/yr (Table
3.6). The poling depth estimate accounted for the actual sediment bed thickness, which
contained a mixture of sand and fine sediment. If the value is corrected to account for
only fine sediment (using the average percent fines value from Table 3.8 of 70%), the
40 
30
E 25
0
M 20
u
0 15
6
10
5
0
I j ____________ ___________ _____________________________________________
_ ILl ~i~ij
0 500 1000 1500 2000 2500 3000 3500
Time from 03/01/1981 (d)
Figure 5.5 Historical flow record for C18 canal, from 03/01/8101/18/91.
tenyear fine deposition thickness would be 0.147 m (0.0147 m/yr), which is within 3%
of the model prediction. This procedure is based on the assumption that if only fine
sediment was present in the estuary, it would deposit at a thickness proportional to it's
mass fraction. This assumption does not account for the different bulk and dry densities
of the sand and fine fraction, however.
To compare sedimentation in a regulated tributary (C18) versus one that is
hypothetically unregulated, the flow record for the Northwest Fork from the same period
as Figure 5.5 was prorated so the total discharge over the tenyear period was the same
as for the C18 canal (Figure 5.6). This was done by summing the daily discharge over
the tenyear period in the C18 canal (6,245 m3/s) and the Northwest Fork (7,217 m3/s),
and multiplying each daily discharge value for the Northwest Fork by the ratio (0.865) of
the tenyear cumulative discharges. As a result of this procedure, the tenyear cumulative
discharge in the Northwest Fork becomes equal to the tenyear cumulative discharge in
the C18 canal. This procedure creates a hypothetical unregulated C18 flow record, that
contains a prevalent background discharge as only 4.5% of the days had zero flow, as
compared to 54% for the existing (regulated) C18 flow record.
Equation 5.1 was applied to the unregulated C18 canal flow record, resulting in a
tenyear deposition thickness of 0.22 m (0.022 m/yr). This implies that episodic
discharges from the C18 canal actually result in lower sedimentation rates than if the
flows were more evenly distributed, as in the Northwest Fork (Table 5.2). Figure 5.7
shows the cumulative deposition for the two flow records over the tenyear period.
The deposition under the unregulated flow record is greater than it is under
regulated flow. This is a consequence of the nearconstant discharge present in the
unregulated flow record. As Figure 5.1 shows, sedimentation rate increases greatly with
discharge at lower values, but this increasing trend is attenuated at higher discharges.
Therefore the higher occurrence of low flows in the unregulated C18 flow record allows
for relatively higher deposition as compared to the existing regulated flow record. This
can be illustrated by considering sedimentation over a sample period, such as two days. If
the first day had an average flow rate of 25 m3/s, and the second day had no flow, the
total twoday sedimentation would be 2.9 x 104 m (Equation 5.1). If, however, the flow
was divided evenly over the two days, resulting in 12.5 m3/s of flow each day, the total
sedimentation would be 4.9 x 10"4 m, a 70% increase.
av,
E
J 25
3 0)
C o
0) 20
E
m S 15
6
10
5
0 500 1000 1500 2000 2500 3000 3500
Time from 03/01/1981 (d)
Figure 5.6 Assumed C18 canal flow record obtained by prorating the measured record
for the Northwest Fork for the period 03/01/8101/18/91.
Table 5.2 Sedimentation rate in the C18 canal
Flow record Sedimentation rate (Eq. 5.1) (m/yr)
Existing (regulated) 0.015
Hypothetical (unregulated) 0.022
5.2 Sediment Trap and Trap Efficiency Calculation
Once the flow and sediment transport models were calibrated and tested (Chapter
4), a trap was incorporated in the C18 canal (Figure 5.8). This location was chosen
because it is near the area of greatest postdredging deposit thickness, with a poling depth
of approximately 1.2 m (Figure 3.9). A dredging depth of 3 m (from original bed depth),
width of 60 m (one cell), and length 180 m (3 cells) were chosen as sufficient to reduce
the velocity in the canal, and allow a measurable amount of sediment to settle. When
0.25
0.2 
*. 0.15
SC18
L unregulated
0
aC 18
___________________ regulated
_4 0.1
0.05
0 1 2 3 4 5 6 7 8 9 10
Time (yr)
Figure 5.7 Cumulative deposition over tenyear period (03/01/8101/18/91) for existing
regulated versus assumed unregulated C18 canal discharge.
simulating a cumulative 50% flow magnitude (1.3 m3/s) and regular tidal forcing (Section
4.3.1) in the C18 canal, the maximum velocity over the trap was found to be 0.04 m/s.
Over the same area, with no trap in place, the maximum velocity was 0.12 m/s, which
indicates a 67% decrease in velocity over the trap.
In order to incorporate the trap, the bathymetry file was modified at the selected
cells. The flow model was run with the trap in place, in order to generate the new
velocities and water surface elevations. The output from the flow model was then applied
to the sediment transport model. Sediment removal ratio, as defined in Equation 2.20,
was calculated from the influent and effluent sediment loads in units of kg/s (Equation
2.19) at the cells adjacent to the trap on the landward and oceanside edges of the trap
(Figure 5.8).
Freshwater flow with input concentration
Flow direction
on ebb tide
i Influent cell
 Effluent cell
C18 canal
Central embayment
I i I I IJ II
I. + I I
Figure 5.8 Portion of the computational grid.
adjacent influent/effluent cells also shown.
Three trap cells are shown in black. Trap
Influent and effluent sediment loads were calculated for each time step, and the
removal ratio averaged over one ebb tidal cycle as follows:
M
YRi
R av i=
ave M (5.2)
where Rave is the ebb tideaveraged removal ratio, i is the index for each time step At, Ri
is the removal ratio from a single time step, and M is the total number of time steps over
an ebb tidal period, as follows:
(5.3)
M=
2At
where T is the tidal period.
Flood tidal data were not used to calculate the removal ratio because the effluent
load (landward edge) was affected by the input concentration at the C18 canal. In
addition, the influent load (seaward edge) contained only the sediment which had escaped
the trap on the previous ebb tide. It was observed that C18 canal 100% flow (31 m3/s)
was the only discharge that produced a constant seaward flow while the ocean tide was
flooding. In this case the removal ratio was calculated for the entire tidal period.
5.2.1 Trap Efficiency as a Function of C18 Canal Discharge
If discharge through the S46 control structure has an effect on trap efficiency,
then flow management practice will affect trap efficiency as well. To explore this,
simulations were performed using the sediment rating curves developed in Section 3.6,
and applied to 8 flow cases. Table 5.3 gives the flows used and the corresponding
sediment concentration. The organic content used was 15%, reflecting the mean organic
content of the native sediment. Removal ratios were calculated only during periods of ebb
flow through the trap (Section 5.2), and plotted against C18 canal discharge (Figure 5.9).
Table 5.3 Flows and suspended sediment concentrations for trap efficiency as a function
of C18 canal discharge
C18 flow (ma/s) C18 suspended sediment
concentration (kg/m3)
0.42 0.008
0.50 0.009
1.27 0.013
1.45 0.014
1.69 0.016
2.54 0.019
4.25 0.024
37.00 0.070
0.6
0.5
0.4
> 0.3
o
E
0.2
0.1
0
0.1 1 10 100
C18 discharge (m3/s)
Figure 5.9 Removal ratio in the presence of trap as a function of C18 canal discharge.
These simulations show that the removal ratio is maximum at a C18 discharge of
approximately 1.7 m3/s. At a higher discharge (and thus velocity), particles are
transported past the trap. At a lower discharge, the same particles settle before the trap.
To illustrate this behavior, Figure 5.10 shows the tracks of a particle under three different
velocities U (with UaU
105 m/s, a particle under a flow velocity of Ua=0.012 m/s would deposit before the trap.
Under a flow velocity of Ub=0.018 m/s, the particle would arrive at the trap at a depth of
1.06 m, where it would then be subjected to a lower velocity (Ubt=0.006 m/s) caused by
the increase in depth over the trap. It would then fall 0.6 m between the entrance to the
trap and the end of the trap, and thus would be captured (at a depth of 1.66 m). A particle
under a flow velocity of Uc=0.03 m/s, however, would arrive at the trap at a depth of 0.62
m. At the trap the flow velocity would reduce to Uct=0.011 m/s, but the particle would
only fall another 0.34 m, which would leave it at a depth of 0.96 m when it left the trap,
still suspended in the flow. Therefore for this trapping scheme and sediment type, a
discharge of 1.7 m3/s results in the maximum removal ratio. Above this discharge
particles are moving at a velocity such that more of them are able to pass the trap, and
below this discharge more particles deposit before the trap.
Particle tracks under
different velocities
Particle a U b U U
7I h
Figure 5.10 Schematic diagram of a particle passing over a trap at three different
velocities.
This hypothesis can be tested by moving the trap further upstream. The trap was
moved 480 m upstream (1260 m from S46 structure) from it's previous location, and the
removal ratio was calculated for three C18 canal discharges (0.42, 1.45, and 4.25 m3/s).
Table 5.4 compares the removal ratios for the three discharges, for the two trap locations.
The two discharges below 1.7 m3/s, for which removal ratio is a maximum, had increased
removal ratios when the trap was moved upstream. The discharge above 1.7 m3/s had a
decreased removal ratio at the new trap location. These observations confirm the
mechanisms of Figure 5.10.
Table 5.4 Removal ratio for three C18 discharges, with two different trap locations
C18 discharge (m3/s) Removal ratio with Removal ratio with trap
original trap location 480 m further upstream
0.42 0.17 0.70
1.45 0.48 0.49
4.25 0.35 0.30
5.2.2 Trap Efficiency as a Function of C18 Canal Sediment Concentration
The sediment rating curves developed in Section 3.6 were used in the previous
simulation to determine input concentrations at the C18 canal. Note that all the
concentrations in Tables 5.1 and 5.3 correspond to the free settling range, within which a
constant settling velocity is prescribed (Section 3.9). In reality, highconcentration mud
suspensions are common just above the sediment bed, with concentrations that may rise
above 1 kg/m3 (Winterwerp, 1999). Under these conditions, settling would go beyond the
free range, and thus trap efficiency would be affected by the settling regime, e.g., free
settling versus flocculation settling. In order to quantitatively account for this effect using
depthaveraged modeling (in which a vertical profile in concentration does not occur),
concentration in the C18 canal was increased under a constant flow (1.7 m3/s), and the
removal ratio calculated. Table 5.5 shows the selected input concentrations and the
resulting settling regime. Figure 5.11 shows the simulations.
Table 5.5 Simulation parameters for trap efficiency as a function of input sediment
concentration to the C18 canal (Qc18=1.7 m3/s)
C18 input sediment Settling zone
concentration
(kg/m3)
0.014 free settling
0.10 free settling
0.25 free settling
0.757 flocculation settling
725 hindered settling
UAU 4U
0.01 0.1 1 10
Concentration (kg/m3)
Figure 5.11 Removal ratio as a function of input suspended sediment concentration to the
C18 canal.
The trend in Figure 5.11 implies that the removal ratio is not affected by
concentration at values greater than 0.1 kg/m3. Note however that the formula to calculate
removal ratio (Equation 2.20) normalizes the difference between influent and effluent
load by the influent load. To eliminate the effect of normalizing the difference between
influent and effluent loads, Figure 5.12 shows the nonnormalized difference, qiqe,
p
plotted against input concentration. As a result of not normalizing the load difference,
observe that increasing input concentration does result in higher load being trapped. The
rate of increase begins to decrease at concentrations above 5 kg/m3, which is just below
the hindered settling range (z7 kg/m3 for this sediment type, Section 3.9). In this range a
lower settling velocity decreases the amount of load that can be trapped. This simulation
also serves as a sensitivity analysis for the settling velocity in the model. Since settling
velocity is computed as a function of sediment concentration (Equation 2.16), Figure 5.12
also shows the variation of trapped load with settling velocity.
0.5
0.45
0.4
0.35
S 0.3
S 0.25
0.2
0.15
0.1
0.05
0
0.01 0.1 1 10 100
Concentration (kg/m3)
Figure 5.12 Difference between influent and effluent loads (trapped load) as a function of
input suspended sediment concentration to the C18 canal.
5.2.3 Trap Efficiency as a Function of Organic Content
To determine the effect of organic content on trap efficiency, simulations were
performed at a C18 canal discharge of 1.7 m3/s, and at an input suspended sediment
concentration of 1 kg/m3. This value was selected because the removal ratio is
independent of concentration in this range (Figure 5.11). Table 5.6 shows the avalues
(settling velocity scaling coefficient in Equation 2.16), and the sediment dry and granular
densities as functions of organic content. The results are shown in Figure 5.13.
The trend in Figure 5.13 indicates that sediment with a high organic content (and
therefore low density and settling velocity) is more difficult to remove from suspension
than sediment with relatively low organic content. To examine the trend of the amount of
sediment actually being trapped, sedimentation rate in the trap was calculated via
Equation 2.18. Marvin (2001) observed consolidation rates for an organicrich Florida
sediment, where selfweight consolidation of the sediment resulted in a 64% decrease in
bed height after 200 h. This correction was applied to each sedimentation rate to account
for consolidation of the sediment bed in the trap. The resulting sedimentation (shoaling)
rate versus organic content plot is shown in Figure 5.14.
Table 5.6 Values used for coefficient a (Eq. 2.16), dry density, and granular density, for
simulation of removal ratio as a function of organic content
Organic content avalue Dry density (Eq. 4.6) Granular density (Eq. 4.4)
(%) (Eq. 4.7) (kg/m3) (kg/m3)
0 0.2000 2650 2650
5 0.1998 1210 2568
10 0.1954 590 2485
15 0.1890 331 2403
20 0.1812 220 2320
25 0.1700 172 2238
30 0.1532 149 2155
35 0.1240 138 2073
40 0.0760 131 1990
0.9
0.8
0.7
.2 0.6
II
0 .5
S0.4
0.3
0.2
0.1
0
0 5 10 15 20 25 30 35 40
Organic content (%)
Figure 5.13 Removal ratio as a function of organic content at a constant discharge of 1.7
m3/s.
1.00E03
9.00E04
8.00E04
0 7.00E04
E
S3.00E04
2.00E04
1.00E04
0.00E+00
0% 5% 10% 15% 20% 25% 30% 35% 40%
Organic content (%)
Figure 5.14 Sedimentation (shoaling) rate in trap as a function of organic content at a
constant discharge of 1.7 m I/s, assuming uniform consolidation.
83
The trend in Figure 5.14 implies that sediment with higher organic content shoals
at a higher rate than sediment with lower organic content. Two phenomena contribute to
this trend. Firstly, the decrease in dry density with increasing organic content (Figure 4.9)
leads to a greater deposition thickness (Equation 2.17). Secondly the influent load rises as
organic content is increased (due to less deposition upstream of the trap via lower settling
velocity), which increases the potential amount of sediment that can be trapped. The mass
deposited upstream of the trap (in terms of mass) versus organic content is shown in
Figure 5.15. As the trend shows, sediment with low organic content tends to deposit at a
higher rate upstream of the trap, due to the high settling velocity (and therefore
deposition flux, Equation 2.17).
35000
30000
g 25000
00
I 20000
o 15000
0C
a)
U)
cu 10000
5000
0
0 5 10 15 20 25 30 35 40
Organic content (%)
Figure 5.15 Mass deposited in the C18 canal (upstream of the trap, Figure 5.8) as a
function of organic content.
