vxEPA
United States
Environmental Protection
Agency
Evaluation of the State-of-the
Art Contaminated Sediment
Transport and Fate Modeling
System
RESEARCH AND DEVELOPMENT
-------
EPA/600/R-06/108
September 2006
Evaluation of the State-of-the-Art Contaminated Sediment
Transport and Fate Modeling System
by
Earl J. Hayter
Ecosystems Research Division
National Exposure Research Laboratory
Athens, GA 30605
U.S. Environmental Protection Agency
Office of Research and Development
Washington, DC 20460
-------
NOTICE
The information in this document has been funded by the United States Environmental
Protection Agency. It has been subjected to the Agency's peer and administrative review, and it
has been approved for publication as an EPA document. Mention of trade names of commercial
products does not constitute endorsement or recommendation for use.
ABSTRACT
Modeling approaches for evaluating the transport and fate of sediment and associated
contaminants are briefly reviewed. The main emphasis is on: 1) the application of EFDC
(Environmental Fluid Dynamics Code), the state-of-the-art contaminated sediment transport and
fate public domain modeling system, to a 19-mile reach of the Housatonic River, MA; and 2) the
evaluation of a 15-year simulation of sediment and PCB transport and fate in this 19-mile reach.
The development of EFDC has been supported by Regions 1 and 4, the Office of Water, the
Office of Superfund Remediation Technology Innovation (OSRTI), and the Office of Research
and Development (ORD) - NERL/ERD. EFDC is currently being used at the following
Superfund sites: Housatonic River, MA; Kalamazoo River, MI; Lower Duwamish Waterway,
WA; and Portland Harbor, OR.
The evaluation of the modeling results showed that EFDC is capable of simulating the
transport and resultant concentrations of TSS and PCBs in this reach of the Housatonic River
within specified model performance measures. Specifically, a statistical summary of the
performance of the EFDC model for TSS and PCB concentrations found that the relative bias at
the downstream boundary of the model domain (i.e., Rising Pond dam) is well within the model
performance measure of ± 30% for TSS (-11.93%) and just outside the measure for PCB
concentrations (-31.97%). For median relative error, the model performance measure is also ±
30%, and the EFDC model is within the performance measure for both TSS (-27.12%) and PCB
(-3.32%) concentrations. Considering the fact that the model was only minimally calibrated, and
that the system modeled had widely varying hydraulic and morphologic regimes, the EFDC
model's performance, as quantified by relative bias and median relative errors, is considered
good. This demonstrates that EFDC is a robust modeling system that can be successfully
implemented at other contaminated sediment sites.
11
-------
CONTENTS
Page No.
Abstract ii
Figures v
Tables viii
Acronyms ix
1 Introduction 1
1.1 Background 1
1.2 Components of contaminated sediment transport modeling study 2
1.3 Purpose and scope of model evaluation 5
1.4 Organization of report 6
2 Description of EFDC 11
2.1 Description of Model 11
2.2 Hydrodynamics and Transport Model 12
2.3 Sediment Transport Model 16
2.3.1 Non-cohesive sediment transport processes 17
2.3.2 Cohesive sediment transport processes 18
2.4 Contaminant Transport and Fate Model 19
3 Model Application 22
3.1 Overview 22
3.2 Spatial Domain 22
3.3 Model Grid 23
3.4 Model Inputs 23
3.4.1 Geometry, floodplain topography, and river bathymetry 24
3.4.2 Bottom friction and vegetative resistance in the riverbed and
floodplain 24
3.4.3 Sediment bed and floodplain soil composition and PCB
Concentrations 25
3.4.4 Hydraulic characteristics of the four dams 27
3.4.5 Initial conditions 28
3.4.6 Boundary conditions 28
3.5 Model Parameters 29
3.5.1 Partitioning of PCBs in pore water and the water column 29
3.5.2 Sediment - water column PCB exchange 30
3.5.3 Sediment Particle Mixing 30
3.5.4 Volatilization 30
4 Model Evaluation 59
iii
-------
4.1 Model Application 59
4.2 Model-Data Comparisons for Water Column TSS and PCBs 59
4.3 Model Results for Sediment PCB Concentrations 60
4.4 Evaluation of Model Performance 61
4.5 Process-Based Flux Summaries 62
5 Conclusion 82
References 83
Glossary 93
Appendices
A Sediment Properties and Transport 98
B Sediment Gradation Scale 127
IV
-------
FIGURES
Fig. No. Page No.
1.1 Sample CSM for Sediment Site (after USEPA, 2005) 8
1.2 Housatonic River Watershed (after WESTON, 2006) 9
1.3 Housatonic River Reaches 5 through 8 (after WESTON, 2006) 10
3.1 Housatonic River Reaches 5 through 8 (after WESTON, 2006) 33
3.2 Housatonic River between Woods Pond and Great Barrington
(after WESTON, 2006) 34
3.3 Spatial Domain of the EFDC Model Showing Variation in
Bottom Elevation (in meters - NAD 83 (86)) (after WESTON, 2006) 35
3.4a Foodchain Reaches 7a - 7e from Woods Pond Dam to Willow Mill Dam
(after WESTON, 2006) 36
3.4b Foodchain Reaches 7f- 8 from Willow Mill Dam to Rising Pond Dam
(after WESTON, 2006) 37
3.5a Longitudinal Bottom Gradient Profile Showing Foodchain Reaches 7A - 8
(after WESTON, 2006) 38
3.5b Longitudinal Bottom Gradient Profile in Reaches 7 and 8 Showing Location
of the Four Dams (after WESTON, 2006) 38
3.6a Computation grid for EFDC model - upstream boundary is at the top of
this figure 39
3.6b Computation grid for EFDC model - lateral black line in the middle of this
stretch is Columbia Mill dam (arrow points to dam) 40
3.6c Computation grid for EFDC model 41
3.6d Computation grid for EFDC model 42
3.6e Computation grid for EFDC model 43
3.6.f Computation grid for EFDC model - lateral black line in the left third of this
-------
stretch is Willow Mill dam (arrow points to dam) 44
3.6.g Computation grid for EFDC model - lateral black line on the right hand side
(i.e., upstream end) of this stretch is Willow Mill dam 45
3.6.h Computation grid for EFDC model - Stockbridge golf course is in the middle
of this stretch 46
3.6.i Computation grid for EFDC model - lateral black line in the middle of this
stretch (immediately to the left - or downstream - of the red colored cells) is
Glendale dam (arrow points to dam) 47
3.6.j Computation grid for EFDC model - lateral black line in the upper right corner
of this figure is Glendale dam 48
3.6.k Computation grid for EFDC model - the downstream end of this figure is the
upper part of Rising Pond 49
3.6.1 Computation grid for EFDC model - the downstream end of this figure is
Rising Pond (arrow points to Rising Pond dam) 50
3.7a Photo of Stockbridge golf course 51
3.7b Grid in the area of the Stockbridge golf course 52
3.8 Initial Longitudinal Distributions of Grain Sizes for the River Bed Sediment 53
3.9 Initial Longitudinal Distributions of Bulk Densities for the River Bed Sediment 54
3.10 Initial Longitudinal Distributions of Porosities for the River Bed Sediment 55
3.11 Initial Longitudinal Distributions of Sediment Bed PCB Concentrations
for the River Bed Sediment 56
3.12 Initial Longitudinal Distributions of Fractions of Organic Carbon
for the River Bed Sediment 57
3.13 Initial OC-Normalized PCB Concentrations (Segregated into 1-Mile Bins)
versus River Mile (after WESTON, 2006) 58
4. la Comparison of Simulated (Solid Line) and Measured (Red Symbols) TSS
and tPCB Concentrations in the Water Column for 1994-1995. DL = detection limit 67
4. Ib Comparison of Simulated (Solid Line) and Measured (Red Symbols) TSS
and tPCB Concentrations in the Water Column for 1996-1997. DL = detection limit.68
4. Ic Comparison of Simulated (Solid Line) and Measured (Red Symbols) TSS
VI
-------
and tPCB Concentrations in the Water Column for 1998-1999. DL = detection limit.69
4. Id Comparison of Simulated (Solid Line) and Measured (Red Symbols) TSS
and tPCB Concentrations in the Water Column for 2000-2001. DL = detection limit.70
4. le Comparison of Simulated (Solid Line) and Measured (Red Symbols) TSS
and tPCB Concentrations in the Water Column for 2002-2003. DL = detection limit.71
4. If Comparison of Simulated (Solid Line) and Measured (Red Symbols) TSS
and tPCB Concentrations in the Water Column for 2004. DL = detection limit 72
4.2 Comparison of Simulated (Solid Line) and Measured (Red Symbols) TSS
and tPCB Concentrations in the Water Column for April - June 1997.
DL = detection limit 73
4.3 Comparison of Simulated (Solid Line) and Measured (Red Symbols) TSS
and tPCB Concentrations in the Water Column for July - September 2004.
DL = detection limit 74
4.4 Temporal Trend of tPCB Concentrations in Surface Sediments for
Foodchain Reaches 75
4.5 Cross-Plot of Simulated and Measured TSS 76
4.6 Cross-Plot of Simulated and Measured tPCB Concentrations 77
4.7 Probability Distributions of Simulated and Measured TSS Concentrations 78
4.8 Probability Distributions of Simulated and Measured tPCB Concentrations 79
4.9 Process-Based Annual Average Mass Flux Summary for Solids 80
4.10 Process-Based Annual Average Mass Flux Summary for PCBs 81
vn
-------
TABLES
Table No. Page No.
1.1 Typical Elements of a CSM for a Contaminated Sediment Site
(after USEPA, 2005) 7
3.1 Effective Diameters for Non-Cohesive Sediment Classes 32
4.1 Statistical Evaluation of EFDC Model Performance,
January 1994-December 2004 64
4.2 Process-Based Annual Average Mass Flux Summary Tabulation for Solids 65
4.3 Process-Based Annual Average Mass Flux Summary Tabulation for PCBs 66
Vlll
-------
ACRONYMS
CEC Cation Exchange Capacity
COC Chemical of Concern
CSM Conceptual Site Model
DOC Dissolved Organic Carbon
EFDC Environmental Fluid Dynamic Code
FCM Foodchain Model
HSPF Hydrologic Simulation Program FORTRAN
NERL National Exposure Research Laboratory
PCB Polychlorinated Biphenyls
PSA Primary Study Area (Reaches 5 and 6)
SAR Sodium Adsorption Ratio
SSC Suspended Sediment Concentration
TOC Total Organic Carbon
tPCB total PCBs
TSS Total Suspended Solids
USCOE U.S. Corps of Engineers
USEPA U.S. Environmental Protection Agency
USGS U.S. Geological Survey
IX
-------
1 INTRODUCTION
1.1 Background
Remediation of bodies of water such as rivers, reservoirs, lakes, harbors and estuaries
contaminated with PCBs, metals, metalloids, and other toxic chemicals is usually extremely
expensive. The assessment and prediction of the transport and fate of contaminated sediments,
and the associated chemical bioaccumulation are often key issues for both human and ecological
risk assessments and remedial decision-making at Superfund sites, and the need for transparent
and consistent approaches to this issue across sites and across Regions is self evident. Modeling
the transport and fate of sediments and their adsorbed contaminants is often one of the tools used
to assess remediation alternatives. Advanced numerical models that simulate the transport and
fate of contaminants in surface waters are important tasks with which one can understand the
complex physical, chemical and biological processes that govern contaminant transport and fate.
However, no single assessment approach is appropriate for all sites, so there must also be
flexibility in the rigor and scope of assessments while maintaining the consistency of principles.
The National Exposure Research Laboratory' s Ecosystem Research Division in Athens,
GA has a research program entitled "Contaminated Sediment Transport and Fate Modeling", the
goal of which is to develop a consensus framework for transport/fate/bioaccumulation modeling
at Superfund sites. This framework is to include modeling protocols for applying the component
contaminated sediment transport and bioaccumulation models to evaluate proposed remediation
measures at contaminated sediment Superfund sites. To accomplish this task, the following five
research objectives are being performed:
1. Evaluation of existing contaminated sediment mass fate and transport models and
adsorbed contaminant bioaccumulation models. Existing, public-domain
contaminated sediment transport models were evaluated in 2003 (see Imhoff et al.
2003), and existing chemical bioaccumulation models were evaluated in 2004 (see
Imhoff etal 2004).
2. Testing of highest ranked contaminated sediment transport model. Based on the
review performed by Imhoff et al. (2003), the highest ranked contaminated sediment
transport model, EFDC (Environmental Fluid Dynamic Code), was subsequently
tested in the following types of surface water bodies: river (Housatonic River, MA;
reservoir (Lake Hartwell, GA/SC); salt-wedge estuary (Lower Duwamish Waterway,
WA); and partially stratified estuary (St John-Ortega-Cedar Rivers, FL). The purpose
of this testing was to evaluate the ability of EFDC to simulate the hydrodynamics,
sediment transport and contaminant transport and fate in these different types of
surface waters. In the Lower Duwamish Waterway, only the ability of EFDC to
simulate the barotropic and baroclinic circulation in a salt-wedge estuary was
evaluated (Arega and Hayter, 2006). In the St John-Ortega-Cedar Rivers, the ability
of EFDC to simulate the hydrodynamics and sediment transport in a micro-tidal,
partially stratified estuary was evaluated (Hayter et al., 2003). In both the Housatonic
-------
River and Lake Hartwell, the ability of EFDC to simulate the hydrodynamics,
sediment and contaminant transport and fate were evaluated.
3. Evaluation of EFDC by modeling the transport and fate of sediments and
contaminants over a minimum of 10-years at a demonstration site. This evaluation is
presented in this report and is discussed in Section 1.3.
4. Develop new modules for EFDC to address the identified sediment-related needs of
OSRTI and the Regions. In 2003, OSRTI identified contaminated sediment-related
research priorities that included the development of models to simulate processes such
as the vertical transport of contaminants dissolved in the pore water of sediment out of
the sediment and up through an overlying sediment cap. In response to these
identified needs, algorithms to simulate the following processes have been (or are
currently being) developed and incorporated into EFDC:
a) Simulation of consolidation due to sediment self-weight and cap-induced
overburden, and the resulting upward flux of dissolved contaminants. This
module has been completed and tested. A paper that will be submitted to a
peer-reviewed journal is currently being prepared.
b) Simulation of wave-induced resuspension of highly organic sediments and the
associated contaminants. This module is currently being developed under
contract to the University of Florida.
c) Linking sediment transport, eutrophication and diagenesis modules in EFDC
to account for resuspension and settling of inorganic sediment and organic
matter. This work is to be performed by Tetra Tech under a Work
Assignment starting in FY2007.
5. Develop a consensus framework for modeling remedial alternatives in surface waters.
Building from the upgraded version of EFDC, a consensus framework for
transport/fate and bioaccumulation modeling at contaminated sediment Superfund
sites will be developed. This framework will include protocols for applying the
component watershed loading model, the transport and fate modeling system, and the
chemical bioaccumulation model.
1.2 Components of contaminated sediment transport modeling study
For the sake of completeness, the components of a complete and technically defensible
contaminated sediment transport modeling study are briefly reviewed in this section. The reader
should refer to EPA's Contaminated Sediment Remediation Guidance for Hazardous Waste Sites
(USEPA, 2005) for a more in-depth discussion of this topic.
Develop Conceptual Site Model: A conceptual site model (CSM) of a contaminated
sediment site is a representation of an environmental system (e.g., watershed) and the
physical, chemical, and biological processes that govern the transport of sediments and
the transport, fate and transformation of contaminants from sources to receptors.
Important elements of a CSM include information about both point and nonpoint
-------
sediment and contaminant sources, transport pathways (both over land and in the water
body), and exposure pathways (USEPA, 2005). Summarizing this information in one
place helps in identifying data gaps and areas of uncertainty that might impact the
subsequent remedial investigation (RI) and feasibility study (FS). The initial version of a
CSM is usually a set of hypotheses derived from existing site data and possibility
knowledge gained from other sites. The subsequent site investigation is a collection of
field and laboratory studies conducted to test these hypotheses and quantify the
qualitative descriptions in the initial CSM. The initial CSM is modified as additional
source, pathway, and contaminant information is collected and analyzed during the site
investigation. A thorough CSM along with a site tour are invaluable in determining
whether or not a modeling study needs to be performed, and if so, what level of
analysis/model is required. Typical elements of a CSM for a contaminated sediment site
are listed in Table 1.1. An example schematization of a contaminated sediment CSM that
focuses on sediment and contaminant transport and fate processes is shown in Figure 1.1.
Determine Whether a Modeling Study is Needed/Appropriate: The following questions
(modified from USEPA, 2005) are useful, but not inclusive, for determining the appropriate use
(if at all) of site-specific mathematical models:
Are historical data and/or simple quantitative techniques available to determine the
validity of the hypotheses in the CSM with the desired accuracy?
Have the spatial extent, degree of heterogeneity, and levels of contamination at the
site been defined?
Have all significant ongoing sources of contamination been defined and their fluxes
measured?
Do sufficient data exist to support the use of a mathematical model, and if not, are
time and resources available to collect the required data to achieve the desired level
of confidence in model results?
Are time and resources available to perform the modeling study itself?
In theory, the answers to the first three bullets should be given in the CSM. They are
included here since the answers to these questions should be considered in addressing
this issue. If the decision is made that some type of modeling is needed, the following
material should be useful in deciding what type of model (or level of analysis) should
be used.
Determine the Appropriate Level of Analysis: As in the previous step, the CSM
should be consulted during this step. This step concerns determining if the most
significant (i.e., first-order) processes and interactions that control the transport and/or
fate of sediment and contaminants, as identified in the CSM, can be simulated with
one or more existing contaminated sediment transport and fate models. If it is
determined that there are existing models capable of simulating these governing
processes, then the types of models (e.g., analytical, empirical, numerical) that have
this capability should be identified. The model types that do not have this capability
should not be used. If it is determined that there are no existing models capable of
simulating, at a minimum, the most significant processes and interactions, then other
tools or methods for evaluating proposed approaches should be identified and used. If
it is determined that one or more models or types of mathematical models capable of
-------
simulating the controlling transport and fate processes and interactions exist, then the
process described previously should be used to choose the appropriate type of model.
As shown in Imhoff etal. (2003), there are existing public domain numerical models
that can simulate most of the physical, chemical and biological processes and
interactions (e.g., those shown in Figure 1.1) that control the transport and fate of
sediment and contaminants in water bodies, e.g., EFDC.
Choose the most appropriate model: If the decision is made to apply a numerical
model at a contaminated sediment site, selection of the most appropriate contaminated
sediment transport and fate model to use at a specific site is one of the critical steps in
the modeling program. Familiarity with existing sediment and contaminant transport
models is essential to perform this step. Comprehensive technical reviews of available
sediment and contaminant transport and fate models and chemical bioaccumulation
models have been conducted by the EPA's ORD National Exposure Research
Laboratory - please refer to Imhoff et al. (2003) and Imhoff et al. (2004).
Conduct a complete modeling study: Whenever numerical models are used, the
following steps should be performed to yield a scientifically defensible modeling
study: verification, calibration, validation, sensitivity analysis, and uncertainty
analysis (the latter is not practical to perform with transport and fate models). These
steps are discussed in the following:
Model verification: This step involves evaluation of 1) model theory, 2) consistency of
the computer code with model theory, and 3) the computer code for integrity in the
calculations. Model verification should be documented, or if the model is new, it
should be peer-reviewed by an independent party. Whenever possible, public domain
verified models, calibrated and validated to site-specific conditions should be used.
Model calibration: Uses site-specific information from a time period of record to
adjust model parameters in the governing equations (e.g., bottom friction coefficient
in hydrodynamic models) to obtain an optimal agreement between a measured data set
and model calculations for the simulated state variables.
Model validation: Also referred to as model confirmation. This step consists of a
demonstration that the calibrated model accurately reproduces known conditions over
a different time period with the physical parameters and forcing functions changed to
reflect the conditions during the new simulation period. The parameters adjusted
during calibration should not be adjusted during validation. Model results from the
validation simulation should be compared to the data set. If an acceptable level of
agreement is achieved between the data and model simulations, then the model can be
considered validated, at least for the range of conditions defined by the calibration and
validation data sets. If an acceptable level of agreement is not achieved, then analysis
should be performed to determine possible reasons for the differences between the
model simulations and data. The latter sometimes leads to refinement of the model
(e.g., using a finer model grid) or to the addition of one or more physical/chemical
processes represented in the model.
-------
Sensitivity analysis: This process consists of varying each of the input parameters by a
fixed percent (while holding the other parameters constant) to determine how the
model predictions vary. The resulting variations in simulated state variables are a
measure of the sensitivity of model predictions to the parameter whose value was
varied.
Uncertainty analysis: This process consists of propagating the relative error in each
parameter (that was varied during the sensitivity analysis) to determine the resulting
error in the model predictions. A probabilistic model, e.g., Monte Carlo Analysis, is
one method of performing an uncertainty analysis. While quantitative uncertainty
analyses are possible and practical to perform on watershed loading and food chain
models, they are not so at present on transport and fate models. As a result, a
thorough sensitivity analysis should be performed for the transport and fate models.
1.3 Purpose and scope of model evaluation
The purpose of this research is to evaluate the ability of EFDC, the state-of-the-art
contaminated sediment modeling system, to simulate the transport and fate of a contaminant over
a time period of at least 10 years. This time period was chosen since models would normally
have to be run over a multi-decade time period to evaluate the effectiveness of various remedial
measures, e.g., dredging, capping, dredging and capping, monitored natural recovery (MNR), in
reducing the contaminant concentrations in both the sediment and water column. The
demonstration site chosen was the Housatonic River in western Massachusetts. Specifically, a
19-mile reach of the Housatonic River immediately downstream of Woods Pond (see Figure 1.2)
was chosen as the modeling domain. Reaches 5 through 8 of the Housatonic River are shown in
Figure 1.3, and the 19-mile reach encompasses Reaches 7 and 8. Also shown in Figure 1.3 are
Reaches 5 and 6; these reaches constitute the Primary Study Area (PSA), and were previously
modeled by EPA Region 1 as described in the Modeling Framework Document (WESTON,
2004a), the Model Calibration Report (WESTON, 2004b), and the Model Validation Report
(WESTON, 2006). For the purposes of this model evaluation, the same version of EFDC used by
Region 1 was applied to Reaches 7 and 8. The application of EFDC to Reaches 7 and 8 is
described in detail in Section 3.2. The strategy in applying EFDC to Reaches 7 and 8 was to test
its performance in simulating the transport of sediment and PCBs over a multi-year period
without the benefit of calibration or validation, thereby testing its robustness to yield satisfactory
comparisons with data collected at a sampling station a short distance downstream of Reach 8,
i.e., downstream of the downstream boundary of the modeling domain of Reaches 7 and 8. While
the model was not calibrated or validated, hydrodynamic, sediment and PCB related
parameterizations were changed in order to represent the vastly different hydraulic and
morphologic regimes in Reaches 7 and 8 compared to those in the PSA. Especially considering
the material presented in the previous section, it is important for the reader to understand that the
model of Reaches 7 and 8 is in the traditional sense uncalibrated and unvalidated, and thus model
results will naturally have a higher level of uncertainty associated with them than results obtained
by a calibrated and validated model.
-------
1.4 Organization of report
A description of EFDC is presented in Section 2, while a description of the application of
EFDC to Reaches 7 and 8 is presented in Section 3. The evaluation of EFDC's ability to simulate
the transport of sediment and PCBs in these reaches is discussed in Section 4, and conclusions
from this model evaluation study are presented in Section 5. In Appendix A, properties and
transport processes of both cohesive and noncohesive sediments are described, while Appendix B
contains a sediment gradation scale.
-------
Table 1.1 Typical Elements of a CSM for a Contaminated Sediment Site (after USEPA,
2005)
Sources of contaminants of concern
Upland soils
Floodplain soils
Surface water
Groundwater
Non-aqueous phase liquids (NAPL) and other
source materials
Sediment "hot spots"
Outfalls, including combined sewer outfalls
and storm water runoff outfalls
Atmospheric contaminants
Exposure pathways for humans
Fish/shellfish ingestion
Dermal uptake from wading, swimming
Water ingestion
Inhalation of volatiles
Exposure pathways for biota
Fish/shellfish/benthic invertebrate ingestion
Incidental ingestion of sediment
Direct uptake from water
Contaminant transport pathways
Sediment resuspension and deposition
Surface water transport
Runoff
Bank erosion
Groundwater advection
Bioturbation
Molecular diffusion
Food chain
Human receptors
Recreational fishers
Subsistence fishers
Waders/swimmers/birdwatchers
Workers and transients
Ecological receptors
Benthic/epibenthic invertebrates
Bottom-dwelling/pelagic fish
Mammals and birds (e.g., mink, otter,
heron, bald eagle)
-------
Figure 1.1 Sample CSM for Sediment Site (after USEPA, 2005)
Highlight 2-13: Sample Conceptual Site Model Focusing on Sediment-Water Interaction
4:
Peclpcaro'
~\^
F f
Murdpsl j
a&rag** ^^B
t, C
Source: Modified from Sediment Management Workgroup (SMWG)
-------
XV"
Lang Sound
I I
Figure 1.2 Housatonic River Watershed (after WESTON, 2006)
-------
Nk >*T
.-> ?
ifr "lib li» A I* ar
ll,i« w, U » yfisB^ral
Figure 1.3 Housatonic River Reaches 5 through 8 (after WESTON, 2006)
10
-------
2 DESCRIPTION of EFDC
2.1 Description of Model
As discussed in the previous section, the numerical model evaluated in this study was the
Environmental Fluid Dynamics Code (EFDC) (Hamrick, 1992). Imhoff etal. (2003) evaluated
public domain, contaminated sediment fate and transport models based on their numerical
scheme, physical and chemical processes simulated, model support availability, and application
history. EFDC was the top ranked model. EFDC is currently maintained by Tetra Tech, Inc. and
supported by the U.S. EPA. EFDC is a three-dimensional (3D) public domain modeling system
that has been widely used in water quality and contaminant transport studies. The application
history of EFDC includes: simulating wetting and drying processes of the hydrodynamics and
sediment transport in Morro Bay (Ji et a/., 2000); thermal discharge study in Conowingo Pond
(Hamrick and Mills, 2000); simulating Lake Okeechobee hydrodynamics, thermal, and sediment
transport processes (Jin etal., 2002); studying tidal intrusion and its impact on larval dispersion
in the James River estuary (Shen et a/., 1999); modeling hydrodynamics and sediment transport
in the middle Atlantic Bight (Kim et a/., 1997); and modeling the hydrodynamics and water
quality in Peconic Bay (Tetra Tech, 1999). EFDC has also been used to develop TMDLs in the
following water bodies: Charles River, MA; Mashapaug Pond RI; Christiana River, DE and PA;
Wissahickon Creek, PA; Cape Fear River, NC; Neuse River, NC; Jordon Lake, NC; Boone
Reservoir, NC; Charleston Harbor, SC; Savannah River, GA; Brunswick Harbor, GA; Lake
Allatoona, GA; Southern Four Basins, GA; St. Johns River, FL; Fenholloway River, FL; Myakka
River Estuary, FL; Mobile Bay, AL; Ward Cover, AL; Alabama River, AL; Flint Creek, AL;
Lake Jordon, AL; Lake Mitchell, AL; Logan Martin Lake, AL; Lay Lake, AL; Lake Neeley
Henry, AL; Yazoo River, MS; Escatawpa River, MS; St. Louis Bay, MS; East Fork Little Miami
River, OH; Ten Killer Ferry Lake, OK; Lake Wister, OK; Armanda Bayou, TX; Arroyo
Colorado, TX; San Diego Bay, CA; Los Angeles River, CA; Los Angeles Harbor, CA; Big Bear
Lake, CA; Canyon Creek, CA; Clear Lake, CA; a section of the Sacramento River, CA; and
South Puget Sound, WA. As stated previously, EFDC is also currently being used at the
following Superfund sites: Housatonic River, MA; Kalamazoo River, MI; Lower Duwamish
Waterway, WA; and Portland Harbor, OR.
The EFDC model is a public domain, surface water modeling system incorporating fully
integrated hydrodynamics. It solves the 3D, vertically hydrostatic, free surface, turbulence
averaged equations of motion. EFDC is extremely versatile, and can be used for ID, 2D-laterally
averaged (2DV), 2D-vertically averaged (2DH), or 3D simulations of rivers, lakes, reservoirs,
estuaries, coastal seas, and wetlands.
For realistic representation of horizontal boundaries, the governing equations are
formulated such that the horizontal coordinates, x and y, are curvilinear. To provide uniform
resolution in the vertical direction, the sigma (stretching) transformation is used. The equations
of motion and transport solved in EFDC are turbulence-averaged, because prior to averaging,
although they represent a closed set of instantaneous velocities and concentrations, they cannot
be solved for turbulent flows. A statistical approach is applied, where the instantaneous values
are decomposed into mean and fluctuating values to enable the solution. Additional terms that
11
-------
represent the turbulence terms are introduced to the equations for the mean flow. Turbulent
equations of motion are formulated to utilize the Boussinesq approximation for variable density.
The Boussinesq approximation accounts for variations in density only in the gravity term. This
assumption simplifies the governing equations significantly, but may introduce large errors when
density gradients are large. The resulting governing equations, presented in the next section,
include parameterized, Reynolds-averaged stress and flux terms that account for the turbulent
diffusion of momentum, heat and salt. The turbulence parameterization in EFDC is based on the
Mellor and Yamada (1982) level 2.5 closure scheme, as modified by Galerpin etal. (1988), that
relates turbulent correlation terms to the mean state variables. The EFDC model also solves
several transport and transformation equations for different dissolved and suspended constituents,
including suspended sediments, toxic contaminants, and water quality state variables. An
overview of the governing equations is given in the following; detailed descriptions of the model
formulation and numerical solution technique used in EFDC are provided by Hamrick (1992).
Additional capabilities of EFDC include: 1) simulation of wetting and drying of flood plains, mud
flats, and tidal marshes; 2) integrated, near-field mixing zone model; 3) simulation of hydraulic
control structures such as dams and culverts; and 4) simulation of wave boundary layers and
wave-induced mean currents.
2.2 Hydrodynamics and Transport Model
The 3D, Reynolds-averaged equations of continuity (Eq. 2.1), linear momentum (Eqs. 2.2
and 2.3), hydrostatic pressure (Eq. 2.4), equation of state (Eq. 2.5) and transport equations for
salinity and temperature (Eqs. 2.6 and 2.7) written for curvilinear-orthogonal horizontal
coordinates and a sigma vertical coordinate are given by the following:
d(me) | d(myHu) | d(mxHv) | d(mw) = Q
dt dx dy dz
d(mHu) d(m Huu) d(mHvu) d(mwu) . , d(m) dmx^TT
- - + - - - + - - + - - - - (mf + v - - u -)Hv =
dt dx dy dz dx dy
u <2-2>
dx dx dx dz dz
d(mHv) d(mHuv) d(mHw) d(mwv) . , d(m) dmx.TT
- - + - - - + - - + - - + (mf + v - + u -)Hu =
dt dx dy dz dx dy
dv (2.3)
dH dHdp
u ,
mH - ~m
x
dy dy dy dz dz
12
-------
Po
(24)
= p(p,S,T) (2.5)
d(mHS) | dK/faS1) | d(mxHvS) | d(mM^) = H dz |
ft 5r <9y 3z 2z
| _ |
ft dx dy dz S
where u and v are the mean horizontal velocity components in (x,y) coordinates; mx and my are
the square roots of the diagonal components of the metric tensor, and m = mxmy is the Jacobian or
square root of the metric tensor determinant;/? is the pressure in excess of the reference pressure,
, where p0 is the reference density;/is the Coriolis parameter for latitudinal
Po
variation; Av is the vertical turbulent viscosity; and Ab is the vertical turbulent diffusivity. The
buoyancy b in Equation 2.4 is the normalized deviation of density from the reference value.
Equation 2.5 is the equation of state that calculates water density (p) as functions ofp, salinity
(S) and temperature (7).
The sigma (stretching) transformation and mapping of the vertical coordinate is given as
where z* is the physical vertical coordinate, and h and % are the depth below and the
displacement about the undisturbed physical vertical coordinate origin, z* = 0, respectively, and
H = h + £ is the total depth. The vertical velocity in z coordinates, w, is related to the physical
vertical velocity w * by
* ,diE u d£ v d£. ,, . , u dh v dh. ,_ _
w = w -z( + - + -) + (l-z)( + ) (2.9)
dt mx dx my dy mx dx my dy
13
-------
The solutions of Eqs. 2.2, 2.3, 2.6 and 2.7 require the values for the vertical turbulent
viscosity and diffusivity and the source and sink terms. The vertical eddy viscosity and
diffusivity, Av and A},, are parameterized according to the level 2.5 (second-order) turbulence
closure model of Mellor and Yamada (1982), as modified by Galperin etal. (1988), in which the
vertical eddy viscosities are calculated based on the turbulent kinetic energy and the turbulent
macroscale equations. The Mellor and Yamada level 2.5 (MY2.5) turbulence closure model is
derived by starting from the Reynolds stress and turbulent heat flux equations under the
assumption of a nearly isotropic environment, where the Reynolds stress is generated due to the
exchange of momentum in the turbulent mixing process. To make the turbulence equations
closed, all empirical constants are obtained by assuming that turbulent heat production is
primarily balanced by turbulent dissipation. The vertical eddy viscosities are determined
according to the local Richardson number given as
A critical Richardson number, Rq = 0.20, was found at which turbulence and mixing cease to
exist (Mellor and Yamada, 1982). Galperin et al. (1988) introduced a length scale limitation in
the MY scheme by imposing an upper limit for the mixing length to account for the limitation of
the vertical turbulent excursions in stably stratified flows. They also modified and introduced
stability functions that account for reduced or enhanced vertical mixing for different stratification
regimes.
The vertical turbulent viscosity and diffusivity are related to the turbulent intensity, q ,
turbulent length scale, / and a Richardson number^ as follows:
Av = vql = 0.4(1 + 36Rq )-' (1 + 6Rq)-' (1 + *Rq )ql (2.11)
(2.12)
where Av and Ab are stability functions that account for reduced and enhanced vertical mixing or
transport in stable and unstable vertical, density-stratified environments, respectively. The
turbulence intensity (q2) and the turbulence length scale (/) are computed using the following two
transport equations:
14
-------
d(mHq2) d(myHuq2) d(mxHvq2} d(mwq2) ^Jj ^ ^
1 1 1 = h (A
dt dx dy dz dz q
.
mAn
d(mHq2l) d(myHuq2l) d(mxHvq2l} d(mwq2l)
d2u. ,52
m
dz
The above two equations include a wall proximity function, W = 1 + E2l(KL) 2, that
assures a positive value of diffusion coefficient^1 = (H)~l (z~l + (1 - z)~:)). B},.,Ei, E2, and £5
are empirical constants with values 0.4, 16.6, 1.8, 1.33, and 0.25, respectively. All terms with
Q's (Qu, Qv, Qq, Qi, Qs, QT) are sub-grid scale sink-source terms that are modeled as sub-grid
scale horizontal diffusion. The vertical diffusivity, Aq, is in general taken to be equal to the
vertical turbulent viscosity, Av.
The vertical boundary conditions for the solutions of the momentum equations are based
on the specification of the kinematic shear stresses. At the bottom, the bed shear stresses are
computed using the near bed velocity components (wl3 Vj) as:
v,) (2.15)
Is'
where the bottom drag coefficient ch = ( - )2 , where K is the von Karman constant, A
b
is the dimensionless thickness of the bottom layer, z0 = z0*/H is the dimensionless roughness
height, and z0* is roughness height in m. At the surface layer, the shear stresses are computed
using the u, v components of the wind velocity (MW,VW) above the water surface (usually
measured at 10 m above the surface) and are given as:
(2.16)
15
-------
Wherec = 0.001-(0.8 + 0.065J(M 2+v2)) and p and p are the air and water densities,
A,
respectively. No flux vertical boundary conditions are used for the transport equations.
Numerically, EFDC is second-order accurate both in space and time. A staggered grid or
C grid provides the framework for the second-order accurate spatial finite differencing used to
solve the equations of motion. Integration over time involves an internal-external mode splitting
procedure separating the internal shear, or baroclinic mode, from the external free surface gravity
wave, or barotropic mode. In the external mode, the model uses a semi-implicit scheme that
allows the use of relatively large time steps. The internal equations are solved at the same time
step as the external equations, and are implicit with respect to vertical diffusion. Details of the
finite difference numerical schemes used in the EFDC model are given in Hamrick (1992), and
will not be presented in this report.
2.3 Sediment Transport Model
This section describes the sediment transport module in EFDC. To provide the requisite
background for the discussion of sediment transport in this report, a brief overview of sediment
properties, with an emphasis on the properties of cohesive sediment, is given in Appendix A.
The sediment transport module in EFDC solves the transport equation for suspended
cohesive and noncohesive sediment for multiple size classes. Its capabilities include the
following:
Simulates bedload transport of multiple size classes of noncohesive sediment
Simulates noncohesive and cohesive sediment settling, deposition and
resuspension/entrainment
Uses a bed model that divides the bed into layers of varying thickness in order to
represent vertical profiles in grain size distribution, porosity, bulk density, and fraction of
sediment in each layer that is composed of specified size classes of cohesive and
noncohesive sediment
Simulates formation of an armored surficial layer
Has a consolidation model to simulate consolidation of a bed composed of fine-grained
sediment.
The generic transport equation solved in EFDC for a dissolved (e.g., chemical
contaminant) or suspended (e.g., sediment) constituent having a mass per unit volume
concentration C, is
dm m HC dm HuC dm HvC dm m wC dm m w C
y1 y- 1 £ 1**
where Ky and KH are the vertical and horizontal turbulent diffusion coefficients, respectively; ws
16
-------
is a positive settling velocity when C represents the mass concentration of suspended sediment;
and Qc represents external sources or sinks and reactive internal sources or sinks. For sediment,
C = Sj, where Sj represents the concentration of thejth sediment class. The solution procedure is
the same as that for the salinity and heat transport equations, which use a high-order upwind
difference solution scheme for the advection terms (Hamrick, 1992). Although the advection
scheme is designed to minimize numerical diffusion, a small amount of horizontal diffusion
remains inherent in the numerical scheme. As such, the horizontal diffusion terms in (2.17) are
omitted by setting KH equal to zero.
2.3.1 Noncohesive sediment transport processes
The process formulations used in EFDC for modeling noncohesive sediment transport in
Reaches 7 and 8 are given in this section. Where applicable, reference is made to equations in
Appendix A.
Incipient motion of a given class size of noncohesive sediment is determined using Eqs.
A. 10 and A. 13. Once the applied bed shear stress exceeds the critical shear stress for incipient
motion, the mode of transport, i.e., bedload or in suspension, is determined using the logic
expressed in Eq. A.21.
Bedload transport is determined using the modified Engelund-Hansen formulation (Wu et
a/., 2000; Engelund and Hansen, 1967) given by:
(2.18)
where q^ is the bedload transport rate (mass per unit width per unit time) in the direction of the
near-bottom flow velocity;/?^ is the fraction of grain size classy in the surface bed layer; and/' is
the friction factor defined as:
(2.19)
where U= current speed. The term ey represents the relative magnitude of exposure and hiding
due to non-uniformity of the grain size class fractions within the surficial bed layer. The
modified Engelund-Hansen method uses the following exposure and hiding formulation, given by
Wuetal. (2000):
(2.20)
where the probability of exposure, pej, and the probability of hiding,/?^, are given by:
17
-------
and =
where Dt = diameters of hidden particles, D}- = diameters of exposed particles, andfnj is the
fraction of they'* noncohesive size class in the surficial bed layer.
Suspended sediment transport, which occurs when the grain-related shear velocity
exceeds the settling velocity for a specific grain size class, is a function of the excess shear stress
(i.e., the difference between the grain-related bed shear stress and grain-related critical shear
stress), the near-bed equilibrium suspended sediment concentration and its corresponding
reference distance above the bed surface. The near-bed equilibrium concentration is the
suspended sediment concentration at a reference height, zeq, above the bed surface. It represents
the maximum suspension concentration. Some researchers take zeq to be equal to the thickness of
the bedload transport zone. The method of calculating the near-bed equilibrium concentration,
Ceq, and the reference distance above the bed surface for bed material that consists of multiple
noncohesive sediment size classes and that accounts for the effect of bed armoring [developed by
Garcia and Parker (1991) (see Equation A.25)] is used in the EFDC model. For this model, in
which multiple noncohesive sediment size classes are simulated, the equilibrium concentrations
for each size class are adjusted by multiplying by their respective sediment volume fractions in
the surface layer of the bed.
The settling velocity for noncohesive sediment particles, wsc, is given by van Rijn (1984b)
(see Equation A.20). The deposition rate of a particular size class is equal to the product of the
settling velocity and the suspended sediment concentration for that size class, i.e., Q wscj.
2.3.2 Cohesive sediment transport processes
The formulations used to represent the resuspension, settling, and deposition of cohesive
sediment in the EFDC model are briefly described in this section. The deposition rate for
suspended cohesive sediment is given by Equation A.38. The following settling velocity
equation (in units of meters per day) is used:
-Cwl) (222)
coh
where Cwi is the washload concentration (determined to be 5 mg/L through calibration), and Ccoh
is the concentration of suspended cohesive sediment (WESTON, 2004b).
The resuspension rate of cohesive sediment, Ecoh, is modeled using the following
excess shear stress power law formulation (Lick et al., 1994):
Ecoh = fcohMsedce (2.23)
18
-------
where M= 6.98 g m"2 s"1, n = l.59,fcoh= fraction of cohesive sediment in the surficial bed
layer, TCS = critical shear stress for erosion, and T& = bed shear stress.
2.4 Contaminant Transport and Fate Model
This section describes the three-phase contaminant transport and fate module incorporated
in EFDC. Use of a three-phase partitioning model explicitly accounts for the freely dissolved
contaminant, the phase (or fraction) that is bioavailable via waterborne exposures, and is a better
representation of the bioavailable fraction than a two-phase partitioning model as is used in other
contaminant fate models such as HSCTM-2D (Hayter etal. 1999). There are several important
processes controlling PCB fate, e.g., partitioning, that must be represented in the model. These
are briefly described next.
Nonionic organic chemicals, such as PCBs, can be distributed in various phases in aquatic
ecosystems. One representation of this distribution is that the chemicals are partitioned among
the particulate organic matter (POM), the dissolved organic matter (DOM), and also the freely
dissolved form (US EPA, 1998). The degree of partitioning, as characterized by the dissolved
(free plus DOC-complexed) and particulate fractions,^ and^/p, respectively, is an important
parameter that controls the fate of chemicals. This is because the transport of both the dissolved
and particulate chemical phases is related to this phase distribution (USEPA, 1998).
In the EFDC model, it is assumed that the total PCB (tPCB) load is distributed among the
three phases mentioned previously, i.e., freely dissolved PCBs, DOC-complexed PCBs, and
sorbed or POC-bound PCBs, and that the PCBs are in equilibrium with across all these phases.
While the actual time it takes to reach complete equilibrium can be very long, it is often assumed
that equilibrium between the dissolved and particulate phases occurs over a time scale of only a
few hours to a day (Jepsen et a/., 1995). This is the basis of the equilibrium partitioning
assumption that is commonly used in the field of contaminant transport and fate modeling.
Transport processes that effect the fate of PCBs, and that are represented in the EFDC model, are
discussed next.
Both dissolved and particulate-bound PCBs are advected by the flow in the river.
Adsorbed PCBs are transported with sediment particles as the latter are moved as a result of bed
load, suspended load, deposition, and resuspension as simulated by the sediment transport model.
There is also a vertical diffusive flux of PCBs that occurs in proportion to the gradient between
the total dissolved concentration in the water column and that in the pore water of bed sediment.
This diffusive flux is due to molecular diffusion and bioturbation. In addition, advective
transport due to groundwater flow may also result in a significant mass flux of other, less
hydrophobic contaminants.
Another PCB fate and transport process, volatilization, is also simulated in the EFDC
model. Volatilization is the loss of freely dissolved chemicals via transfer from the water column
to the atmosphere. Given the relatively short residence times in Reaches 7 and 8 and the high
degree of chlorination of the PCB mixture found in the Housatonic River, the air data that have
been collected during the project, and the low rate of volatilization of PCBs from the water
column, volatilization was described as a secondary loss mechanism in the CSM for the PSA, but
was nonetheless included in the model.
19
-------
The transport equation for the freely dissolved chemical is:
dt (mxmyHCw) + dx (myHuCw) + dy (mxHvCw) + dz (mxmywCw
(2.24)
where Cw is the mass of freely dissolved contaminant per unit total volume, ^is the mass of
contaminant sorbed to sediment class /' per mass of sediment, %D is the mass of contaminant
sorbed to dissolved material j per unit mass of dissolved material, $ is the porosity, i//w is the
fraction of the freely dissolved contaminant available for sorption, Ka is the adsorption rate, Kd is
the desorption rate, and y is a net linearized decay rate coefficient. The sorption kinetics are
based on the Langmuir isotherm (Chapra, 1997) with % denoting the saturation adsorbed mass
per carrier mass. The solids and dissolved material (i.e., DOC) concentrations, S and D,
respectively, are defined as mass per unit total volume. The index j is the number of
contaminants, and the index /' is the number of classes of solids, i.e., organic paniculate matter
and inorganic sediment. The transport equation for the contaminant adsorbed to DOC is:
(mxmyWD'XJD
= o.
(2.25)
The transport equation for the contaminant adsorbed to suspended solids is:
: (myHuS'x's ) + dy (mxHvS'x's ) + dz(n
myWs Xs' z\m/ny H z\ Xs>)
, . .J r Y .
+mm,H I
(2.26)
The concentrations (in units of sorbed mass per unit total volume) of chemicals adsorbed to DOC
and solids, CD and Cs, respectively, are defined as:
(2.27)
C1 =D
^- ^
20
-------
(2.28)
Introducing Equations 2.27 and 2.28 into Equations 2.24 - 2.26 gives
dt (mxmyHCw ) + dx (myHuCw ) + dy (mxHvCw ) + dt (mxmywCw
0£w + mxmyH
-" )
m m H
x y
!+ d
mm
dt (mxmyHCs ) + ^ (myHuCs
(mxmywCs
' ) f ^
(K'ds + r) Q
(2.29)
(2.30)
(2.31)
The EFDC sorbed contaminant transport formulation currently assumes equilibrium partitioning
with the adsorption and desorption terms in Equations 2.30 and 2.31 being equal, such that:
C,.
(2.32)
(2.33)
21
-------
3 MODEL APPLICATION
3.1 Overview
As previously described, the EFDC modeling domain included Reaches 7 (Woods Pond
Dam to the headwaters of Rising Pond in Great Barrington) and 8 (Rising Pond), a distance of 19
miles. The levels of PCB contamination in Reaches 7 and 8 relative to those in Reaches 5 and 6
are, in general, lower, but PCB concentrations in the sediment and particulate organic matter
(POM) exhibit greater variation among their subreaches than in the PSA. Figure 3.1 shows the
locations of Reaches 5 through 8. The remainder of this section describes the application of
EFDC to Reaches 7 and 8. Results from an 11-year simulation of this model are described in
Section 4.
3.2 Spatial Domain
The spatial domain of the EFDC model of Reaches 7 and 8, hereafter referred to as the
EFDC model, includes approximately 30.6 km (19 miles) of the Housatonic River, with the
upstream boundary of the domain located at the outlet of Woods Pond and the downstream
boundary at Rising Pond Dam (Figure 3.2). The domain also includes the floodplain inside the 1 -
ppm PCB soil concentration isopleth, which in most cases is coincident with the 10-year
floodplain. Figure 3.3 shows the entire spatial domain of the EFDC model. The Housatonic
River watershed area that drains into the upstream boundary of the EFDC model is 421.3 km2
(162.6 mi2), and the area of the watershed between the upstream and downstream boundaries is
302.0 km2 (116.6 mi2). Figures 3.4a and 3.4b show nine foodchain reaches (labeled as Reaches
7a, 7b, 7c, 7d, 7e, 7f, 7g, 7h and 8) into which Reaches 7 and 8 were subdivided for the purposes
of foodchain modeling performed by EPA. There will be references to these nine foodchain
reaches in the subsequent parts of this report.
Figure 3.2 shows the local drainage areas (in light green) and the seven tributaries that
were explicitly represented in the HSPF model. HSPF was used to simulate non-point source
water flows and the transported sediment loads into the Housatonic River during runoff events.
These simulated nonpoint source flows and the sediment loads conveyed by that runoff from the
local drainage areas were added directly to the river channel within each local drainage area. The
modeling domain of the HSPF watershed model used by EPA in the PSA modeling study
extended to the USGS gage in Great Barrington, located approximately 1 mile south of Rising
Pond Dam. The HSPF-simulated flow and associated solids concentration time series in the
seven tributaries (i.e., Washington Mountain Brook, Laurel Brook, Greenwater Brook, Hop
Brook, West Brook, Konkapot Brook, and Larrywaug Brook) are represented as point sources in
the EFDC model. Additional discussion of boundary conditions is presented below.
As shown in Figures 3.5a and 3.5b, the longitudinal bottom profile of the Housatonic
River along the EFDC model domain divides Reaches 7 and 8 into the following six hydraulic
sections with markedly different bathymetric and morphological features:
22
-------
Reaches 7A and 7B - From the upstream boundary (Woods Pond Dam) to Columbia
Mill Dam. The average gradient in these reaches is 0.0032 meter per meter (m/m).
Reach 7C and a portion of 7D - From Columbia Mill Dam to the town of Lee. As
shown in Figure 3.5b, the town of Lee represents a marked change in the river bottom
gradient, with the gradient decreasing from 0.0020 m/m upstream of Lee to 0.00075
m/m downstream of Lee.
Reach 7D remainder and 7E - From Lee to Willow Mill Dam. In this subreach, the
bottom gradient, 0.00075 m/m, is less than the two upstream subreaches, and it
contains several meanders.
Reaches 7F and 7G - From Willow Mill Dam to Glendale Dam. These meandering
reaches have the smallest average bottom gradient, 0.00017 m/m, and consequently,
the lowest flow velocities in the EFDC model domain. The Stockbridge Golf Course
is located in this section.
Reach 7H - From Glendale Dam to the upstream limit of the backwaters from Rising
Pond Dam. As shown in Figure 3.5b, this section has the highest bottom gradient,
0.0042 m/m, and consequently, the highest flow velocities in the EFDC model.
Reach 8 - Rising Pond, the impoundment created by Rising Pond Dam. Rising Pond,
unlike the much wider Woods Pond at the downstream end of the PSA, resembles a
run-of-the-river reservoir. The downstream end of each of these six hydraulic sections
was located at a hydraulic control point (e.g., dam, break-in-grade). These figures
also show the predicted river centerline water surface profile at a snapshot in time
(low flow).
3.3 Model Grid
Starting at the upstream boundary of the EFDC model domain, Figures 3.6a-3.61 present
the vertically integrated, orthogonal-curvilinear grid used to model Reaches 7 and 8. This model
grid, composed of 4,938 cells, is shown in that series of 12 images (with overlap on both ends of
each image) to give the reader a useful view of the spatial variability. The gray cells represent
the floodplain, whereas the blue to red colored cells represent the Housatonic River. In most
locations, the river channel is one cell wide, except for Rising Pond and the other impoundments.
In general, the grid resolves the features of the EFDC model domain very well (e.g., meanders in
Reaches 7D through 7G). Figures 3.7a and 3.7b, respectively, show a comparison between an
aerial photograph of the Stockbridge Golf Course and the computational grid for the same area.
3.4 Model Inputs
To simulate the transport and fate of PCBs in the EFDC model domain, the EFDC model
required the following hydrodynamic, sediment, and PCB inputs:
Geometry of the model domain, floodplain topography, and river bathymetry.
23
-------
Bottom friction and vegetative resistance in the riverbed and floodplain.
Sediment bed and floodplain soil composition and associated PCB concentrations.
Hydraulic characteristics of four dams.
Initial conditions for hydrodynamic, sediment, and contaminant transport modules.
Boundary conditions for hydrodynamic, sediment, and contaminant transport modules.
These input parameters are discussed in more detail in the following:
3.4.1 Geometry, floodplain topography, and river bathymetry
The river bathymetry and floodplain topography were developed from a number of
different sources for the main channel, Rising Pond, and the floodplain. Detailed surveys of 77
cross-sections in Reaches 7 and 8 were conducted during summer 2005; these were used to
describe the bathymetry of grid cells within the river channel and Rising Pond. The surveyed
cross-sections provide bottom elevations across the channel at approximately 1-m spacing
between surveyed points. Water surface elevation was also recorded at the time of the survey.
The bathymetry within the main channel was defined by assigning the average bed elevation for a
given cross-section to the EFDC grid cell at that location. Linear interpolation was used to assign
the bottom elevation in channel cells between surveyed cross-sections. Bathymetry in Rising
Pond was developed using cross-sections surveyed in 1998 and 2005 from an analysis performed
in ArcGIS 8.3 with the Spatial Analyst extension. A Triangulated Irregular Network (TIN)
surface was selected as the approach for incorporating these data into the model grid. The
resulting average-value field was joined to the attributes of the model grid. EFDC floodplain grid
cells were assigned bottom elevations from a 5-ft interval digital elevation model (DEM)
developed from USGS contour data.
3.4.2 Bottom friction and vegetative resistance in the river bed and floodplain
For the effective bottom roughness, z0, an effective roughness height of 0.04 m was
assigned to the channel cells in unarmored free-flowing reaches (i.e, lower half of 7D and 7F),
and a roughness of 0.06 m was assigned to the channel cells in the armored free-flowing reaches
(Reaches 7 A, 7C, upper half of 7D, and 7H). These values reflect the hydraulically rougher river
bottom in most of Reaches 7 and 8 as compared to the PSA. In the impoundments formed by the
four dams (i.e., Reaches 7B, 7E, 7G, and 8), a roughness height of 0.02 m was used since the
sediment beds in the impoundments will, in general, be smoother than in the steeper, free-flowing
reaches. A z0 value of 0.04 m was also uniformly applied to floodplain cells. Submerged aquatic
vegetation is not prevalent in Reaches 7 and 8. As a result, the effects of aquatic vegetation were
not represented in the EFDC model. In addition, the effect of friction from floodplain vegetation,
which is a second order effect at most, was also not represented in the EFDC model.
3.4.3 Sediment bed and floodplain soil composition and PCB concentrations.
24
-------
The sediment bed within the river channel and soil on the floodplain were analyzed
separately to establish the initial conditions for bed properties. Core samples taken at numerous
locations throughout the river channel and floodplain were used to determine the spatially
varying sediment grain size distribution across the EFDC model domain. Because of limited data
on sediment properties with depth in the bed, it was assumed that the bed properties of all five
bed layers used in both river channel and floodplain cells were the same. The following
thicknesses used for the five bed layers were the same as used in the PSA model: 7 cm for the
surface layer, 8.24 cm for the first subsurface layer, and 15.24 cm for the three remaining
subsurface layers.
Five grain size classes of solids, one cohesive and four non-cohesive, were used to
represent the range of sediment, floodplain soil, and suspended solids grain sizes in Reaches 7
and 8. The cohesive grain size class was specified as < 63 um, and ranges were defined for four
non-cohesive grain size ranges: 63 to 250 um (very fine to fine sand), 250 to 2,000 urn (medium
to very coarse sand), 2,000 to 8,000 um (very coarse sand to medium gravel), and > 8,000 um
(medium gravel and coarser). Because the EFDC model tracks each size class as a single particle
size, it was necessary to establish a nominal grain size for each class. The effective diameters
used in the model to characterize the four non-cohesive size classes were set equal to the average
of the representative sediment diameter values determined using the following three methods:
Based on the median diameter (D50) of particles within each size class
Based on settling velocities
Based on critical shear velocities
A brief description of each of these three methods is given next. The first step was to separate
the grain size data into four sets, one for each of the noncohesive sediment class ranges.
Median Diameter Method: For each non-cohesive size class, a D50 was calculated for each
sample. The DSO'S from all samples were then combined, and a mean was calculated as the
effective diameter for each size class. The effective diameters calculated with this method are
given in Table 3.1.
Settling Velocity Method: In this method, the settling velocity was calculated for each diameter
represented by the geometric mean diameter between two sieve sizes. As described in Section
2.3, the equations given by van Rijn (1984a) (see Equation A.20) were used to compute the
settling velocity. Once the settling velocities for each grain size were determined, a normalized
settling velocity was calculated. The equivalent particle diameter was then back-calculated using
the settling velocity equation. The effective diameters for the four size classes calculated with
this method are also given in Table 3.1.
Critical Shear Velocity Method: The third method was based on the weighted critical shear
velocities. First, the critical shear stress was calculated by the van Rijn (1984b) formulation (see
Equations A. 10 and A. 13) and using the geometric mean diameter between two sieve sizes. The
critical shear velocities were then calculated from the critical shear stresses, and then they were
weighted using the normalized data set to find the effective critical shear velocity. Lastly, the
25
-------
equivalent grain size diameter for each sample was calculated. The effective diameters calculated
with this method are again given in Table 3.1.
As seen in Table 3.1, the mean effective diameters for the three methods were very
similar, with values ranging from 149 to 179 um for particles in the 63 to 250 um grain size class,
585 to 646 um for the 250 um to 2 mm class, 3,913 to 4,146 um for the 2,000 um to 8,000 um
class, and 13,442 to 13,723 um for particles in the > 8,000 um class. For the EFDC model
implementation, the effective diameters for the four non-cohesive grain sizes were calculated as
the arithmetic mean of the values obtained from the three methods; these effective diameters are
159, 625, 3,993 and 13,560 um, respectively.
Specification of initial conditions for the percent composition of the five grain size classes
for each model grid cell is required. To determine these initial conditions, the river channel was
divided into longitudinal spatial bins, each of which contained approximately the same number of
sediment samples, and the model grid cells within each bin were assigned the measured average
value of the mass fraction within each size class range. Figure 3.8 shows the initial longitudinal
distributions of grain sizes for the upper 60.96 cm (divided into four 15.24 cm layers) of river bed
sediment. River Mile 124.3 is at the outlet of Woods Pond and River Mile 105.2 is at Rising Pond
Dam. This figure shows that, because of the limited data (in comparison to the data available for
the PSA), the grain size distributions between the impoundments formed by Columbia Mill Dam,
Willow Mill Dam, Glendale Dam, and Rising Pond Dam were assumed to be constant. In
addition, the longitudinal grain size distributions were assumed to be constant within the
impoundments. Note that the general pattern within the river channel is for the percentages of
cohesive sediment to increase in the impoundments and decrease in the reaches between
impoundments, as expected. The same general pattern is observed for non-cohesive class 1,
whereas the percentages of non-cohesive classes 2-4 increase with increasing bottom gradient.
The largest non-cohesive size class was made immobile to represent the armoring that exists in
Reaches 7A, 7C, and 7H.
The specification of initial conditions for grain size for floodplain cells was based on a
spatial weighting analysis of the floodplain soil data. Grain sizes were derived from the data
using an inverse distance approach with a 10-m2 interpolation grid. As described previously,
properties developed for the surface layer were also applied to the deeper bed layers due to the
limited data at depth.
The sediment bulk density and porosity were determined from core samples that were
analyzed for solids content. The data were averaged over the same spatial bins as the grain size
distribution data described previously, and the average bulk density and porosity within each bin
were calculated from the sediment specific gravity and the average sediment density. The model
grid cells within each bin were assigned the average bulk density and porosity calculated for that
bin. Spatial plots of the initial conditions of bed bulk density and porosity are given in Figures
3.9 and 3.10, respectively. Bulk density and porosity for floodplain soil were determined from
soil cores with solids content data, assuming a specific gravity of 2.65.
Spatial distributions of sediment PCBs and fraction organic carbon (foc) were determined
from field measurements and assigned to the channel cells as the initial conditions at the
beginning of the simulation. Initial PCB concentrations and foc of the sediment bed are plotted
26
-------
versus river mile in Figures 3.11 and 3.12. Initial OC-normalized PCB concentrations segregated
into 1-mile bins are shown in Figure 3.13 (top figure uses a logarithmic scale and bottom figure
uses an arithmetic scale). The green dotted vertical lines in this figure show the locations of
(starting at the left) Woods Pond Dam, Columbia Mill Dam, Willow Mill Dam, Glendale Dam,
and Rising Pond Dam. PCB concentrations are greater in the impoundments upstream of these
dams than in the free-flowing rivers upstream of the impoundments. This is particularly
noticeable in the impoundment formed by Columbia Mill Dam.
To compute dissolved and particulate PCB concentrations with three-phase equilibrium
partitioning used in EFDC, it is necessary to properly represent carbon-normalized PCB
concentrations in the sediment. Because PCB and TOC data generally follow lognormal
distributions, specification of average PCB and TOC concentrations in a bin representing a river
reach would result in an inaccurate carbon-normalized PCB concentration. This issue was
resolved by specifying the average PCB concentration and an additional input that is a nominal
value for the organic carbon content of the sediment, TOC*, such that the ratio of the two yields
the appropriate carbon-normalized PCB concentration. The average PCB and TOC*
concentrations estimated for a given bin were then assigned to the grid cells within that bin.
Modified inverse distance weighting was used to create a fine-scale (3-m2 grid)
distribution of total PCB (tPCB) concentrations in the floodplain soil based on the available data.
A similar approach was used to develop initial conditions for foc in the floodplain soil by creating
a distribution of foc on a 10-m2 grid using the inverse distance weighting approach. The foc
concentrations within a model grid cell were averaged to develop the initial conditions for the
model.
The EFDC model also requires inputs for dissolved organic carbon (DOC) in the water
column, DOC in the sediment pore water, and fraction organic carbon (foc) on the TSS in the
water column. These parameters are necessary for partitioning the PCBs among particulate,
DOC-complexed, and truly dissolved phases. As in the PSA model, porewater DOC and total
organic carbon (TOC) in the sediment and the DOC and foc in the water column were assumed to
be constant over time. This assumption eliminated the computational complexity of modeling
organic carbon production and fate within the sediment and water column. The same values used
in the PSA model for DOC in the water column and DOC and TOC in the sediment were also
used in the EFDC model of Reaches 7 and 8. However, the foc values on the suspended sediment
in the water column were specified, as determined from data, as 10% on the cohesive size class
and 2% on the four non-cohesive size classes.
3.4.4 Hydraulic characteristics of the four dams
As shown in Figure 3.5, there are four dams within the EFDC model domain. Physical
characteristics (e.g., spillway elevation and length) and hydraulic properties (e.g., flow-stage
rating curve) of these four dams were obtained from Harza (2001) and BBL (1994). In EFDC,
these hydraulic structures are specified as control structures connecting specific upstream and
downstream cells. This specification in EFDC allows water, solids, and PCBs to be properly
transferred from upstream to downstream of the control structure. To account for the fact that the
upstream cell widths were different from the spillway lengths of the dams, the flow-stage rating
curves had to be adjusted so that for a given stage height (in this case, stage height represents the
27
-------
difference between the upstream water surface elevation and the spillway crest elevation), the
correct discharge flowed over the spillway. Specifically, the coefficients in the original rating
curves were adjusted to accomplish this.
3.4.5 Initial Conditions
To begin a simulation, the initial water depth must be defined for each EFDC grid cell.
The hydrodynamic module in EFDC is then run in a spin-up mode for a relatively short period of
time (typically, on the order of a few days) so that the subsequent model simulation is
independent of the hydrodynamic initial conditions. For the EFDC model, the initial water depth
in each channel cell in free-flowing reaches of the river was set to 1.0 m. In the impoundment
areas of the four dams, the initial cell water depths were set equal to the difference between the
dam spillway elevation plus 2 cm and the cell bottom elevations. This procedure ensured that the
impoundments were full at the start of the spin-up, and thus water flowed over the spillways from
the start of the simulation. The initial conditions for the grid cells within the floodplain were
assigned the minimum water depth of 10 cm, indicating a dry cell, signaling EFDC to skip
calculations in these cells. A hydrodynamic spin-up time of four days was used for the EFDC
model.
The sediment transport model was spun-up for one year of simulated time (starting the
simulation using the spun-up hydrodynamic module) to establish more spatially representative
initial sediment bed conditions than those determined using the procedure described in Section
3.4.3. The bed composition at the end of this one year simulation was used as the initial bed
conditions for the subsequent model simulation.
For the start of both the one year sediment spin-up and the subsequent 11-year model
simulation, the initial suspended sediment concentrations for cohesive sediment were set to
spatially uniform values of 5 mg/L. Zero initial concentrations were used for the four non-
cohesive sediment size classes and for tPCBs.
3.4.6 Boundary conditions
Time series of flow, suspended solids concentrations, and particulate and dissolved tPCB
concentrations calculated at the downstream boundary of the PSA model were used as the
upstream boundary conditions for the EFDC model.
As described in the previous sections, the HSPF-simulated runoff and the cohesive and
non-cohesive solids loads conveyed by the runoff from the local drainage areas shown in Figure
3.2 were added directly to the river channel within each local drainage area. These nonpoint
loads were uniformly distributed to the channel cells within each local drainage area. For
example, if there were 50 channel cells within a particular local drainage area, then the HSPF-
calculated nonpoint source loads from the local drainage area on both sides of the river channel
were added, the total load was divided by 50, and the result added to each of the 50 channel cells.
The HSPF-simulated time series of flow and solids loads from the seven tributaries (i.e.,
Washington Mountain Brook, Laurel Brook, Greenwater Brook, Hop Brook, West Brook,
Konkapot Brook, and Larrywaug Brook) shown in Figure 3.2 were represented as direct inputs to
the seven river channel cells located at the confluence of these seven tributaries with the river.
28
-------
Outflows at the EFDC model boundary at the Rising Pond Dam were controlled by the
stage-discharge rating curve and the spillway crest elevation for this dam specified in BBL
(1994).
3.5 Model Parameters
The parameter values and functions used in the EFDC model in the formulas for cohesive
sediment settling and deposition, sediment-water column exchange of dissolved PCBs via
diffusion and of sorbed PCBs via sediment particle mixing, PCB volatilization, as well as for
these parameters needed for three-phase partitioning calculations, are the same as those used in
the PSA model (WESTON, 2004b) . Parameters related to the simulation of contaminant
transport and fate are briefly described next.
3.5.1 Partitioning of PCB s in pore water and the water column
An assumption in equilibrium partitioning theory is that the carbon-normalized PCB
concentrations are proportional to the freely dissolved pore water concentrations. A Pore Water
Partitioning Study and a Supplemental Surface Water Partitioning Study found that a three-phase
partitioning model is a reasonable representation of PCB partitioning in the Housatonic River
(Mathew et al., 2002; BBL and QEA, 2003). In these studies, the data were analyzed for
individual congeners and as tPCBs, and it was assumed that partitioning to organic carbon is a
function of the contaminant-specific octanol-water partition coefficient, Kow (Karickhoff et al.,
1979; Karickhoff, 1981; Di Toro etal, 1985). As reported by Mathew et al. (2002), analysis of
PCB congener data found the following results: for sorbed carbon Koc ~ Kow, and for dissolved
carbon KDoc = C(DOcKow -0.1 Kow. These results are in general agreement with those found by Di
Toro (1985), EPA (1998), and Burkhard (2000).
The studies mentioned in the previous paragraph also found that it was acceptable to
model tPCBs instead of individual PCB congeners. As a result, a tPCB partition coefficient that
reflects the congener distribution in the sediment and pore water needed to be derived. To
accomplish this task, it was assumed that Koc = Kow, with Kow values determined using the results
presented by Hawker and Connell (1988). A weighted average value of log(Kow) = 6.5 was found
using the pore water fractions of dissolved PCB homologues. A value of QDOC =0.1 was
determined by fitting the partitioning data with the three-phase partitioning model (WESTON,
2004b).
An analysis of water column partitioning data also indicated that a three-phase
partitioning model is a reasonable representation for partitioning in the water column (WESTON,
2004b). However, that analysis also showed that DOC complexation is less important in the
water column than in the sediment. BBL and QEA (2003) also reported that DOC complexation
in the water column of the Housatonic River was of less importance relative to that in the
sediment bed. This is due to, among other factors, lower DOC concentrations in the water
column than in the sediment, and differences in the nature of the DOM present in the water
column compared to that in the sediment pore water. The latter factor was the reason the water
column data were analyzed separately to determine a value of KDoc for the water column. It was
29
-------
found that the water column data were best reproduced using QDOC = 0.01 (WESTON, 2004b).
This value is at the lower end of values of QDOC found by Butcher et al. (1998) for tetra through
hexachlorobiphenyl homologues in the water column of the upper Hudson River; those values
ranged from 0.011 to 0.049. Given that DOC in the water column usually forms weaker
complexes with non-polar, organic contaminants (NPOCs), such as PCBs, than those formed with
DOC in the pore water of sediment beds, the results reported by Butcher et al. are in general
agreement with the results found from analysis of Housatonic River pore water - sediment data.
3.5.2 Sediment - water column PCBs exchange
As mentioned previously, exchange of contaminants between the water column and the
bed sediment occur by the diffusive flux of dissolved contaminants between the water column
and pore water in the sediment, and by transport of contaminants that are adsorbed on sediment
that undergoes deposition and resuspension. The contaminant diffusive flux, equal to the product
of a mass transfer coefficient (i.e., effective diffusion coefficient) times the concentration gradient
of the freely dissolved contaminant between the water column and sediment pore water, is known
be an important transport mechanism (Thomann and Mueller, 1987; Thibodeaux et al., 2002).
The diffusive flux is typically of increased importance relative to the resuspension and deposition
induced particulate flux during baseflow periods when the particulate flux is usually reduced. It
also has a greater impact on water column concentrations during baseflow conditions because
there is less dilution of the contaminant mass that diffuses from the sediment due to the smaller
volume of water above the sediment. The mass transfer coefficient is a function of the dissolved
pore water concentration and the mass of dissolved contaminant transferred to the water column
during baseflow conditions. Analyses of sediment-water column mass flux data sets for the PSA
portion of the Housatonic River was used to establish that the order of magnitude of the average
PCB mass transfer coefficient was from 1 to 10 cm/day. The value of 1.5 cm/day used in this
study was determined during calibration of simulated results to water column data (WESTON,
2004b).
3.5.3 Sediment Particle Mixing
Mixing of sediment particles that compose the bed is caused by both physical and
biological processes. Physical mixing processes include resuspension, bedload transport, and
deposition. Bioturbation is an important biological mixing process that results in the vertical
transport of contaminants adsorbed to sediment within the bed sediment (Di Toro, 2001). In the
EFDC model, this process is simulated as a mixing of sediment between the sediment layers in
the bed. This particle mixing is set proportional to a particle-mixing coefficient. A mixing
coefficient of 10"9 m2/s that is in the upper range of values reported by Boudreau (1994) was used
for the upper 7 cm of the bed sediment in the EFDC model.
3.5.4 Volatilization
Volatilization of organic contaminants such as PCBs, is proportional to the concentration
gradient across the air-water interface, and can be modeled using two film transfer theory
(Whitman, 1923). It also depends on the water depth, flow velocity, and wind speed. The
driving force for volatilization is the partial pressure gradient in the air and the concentration
30
-------
gradient in the water, and the volatilization flux in each phase can be described using Pick's Law
of Diffusion. The transfer coefficient due to volatilization is a function of the transfer
coefficients for both liquid and gas phases. Analysis based on congener-specific Henry's Law
constants and the relative site-specific congener composition of samples collected in the PSA of
the Housatonic River found that the loss of PCBs via volatilization is a relatively minor
transport/fate process. Nevertheless, volatilization was simulated in the EFDC model of Reaches
7 and 8 using the PSA derived parameter value.
31
-------
Table 3.1
Effective Diameters for Non-Cohesive Sediment Classes
Parameter
Mean deff (|Jm)
Method (1)
Grain Size Distribution
63 -
250
|Jm
150
250
|Jm
- 2
mm
585
2-8
mm
4146
> 8
mm
13,723
Method (2)
Weighted Settling
Velocities
63 -
250
|Jm
179
250
|Jm
- 2
mm
646
2-8
mm
3913
> 8
mm
13,442
Method (3)
Weighted
Critical Shear Velocities
63-
250
|Jm
149
250
|Jm
- 2
mm
643
2-8
mm
3919
> 8 mm
13,515
32
-------
Reach 6 (Woods Pond)
LEGEND:
o TownCity
Roads
Reach Division Line
I Housatonic River
| | State Parfc
Municipal Boundary
1Q-\fearFlQQdplain
ns D i Kilometers
Model Validation Report
GE)Housatonic River Site
Rest of River
FIGURE 6.2-1
HOUSATONIC RIVER,
PRIMARY STUDY AREA
(REA CHES 5 AND 6) AND
REACHES 7 AND 8
Figure 3.1 Housatonic River Reaches 5 through 8 (after WESTON, 2006)
33
-------
Great Barrington
Roads
^drography
I | Local Drainage
1 1 I HSPF Watershed Model Segments
\
'
5 Miles
Figure 3.2 Housatonic River between Woods Pond and Great Barrington (after
WESTON, 2006)
34
-------
Housatonic River Downstream (Reaches 7 & 8) Model
Bottom Elev (m)
214 6209.00 291
Figure 3.3 Spatial Domain of the EFDC Model Showing Variation in Bottom Elevation
(in meters - NAD 83 (86)) (after WESTON, 2006). North is up in this figure.
35
-------
Segment Boundary (Dam)
Segment Boundary (No Dam)
EFDC Grid
Sediment Sample
(Synoptic OC only)
Figure 3.4a Foodchain Reaches 7a - 7e from Woods Pond Dam to Willow Mill Dam (after
WESTON, 2006). North is up in this figure.
36
-------
Segment Boundary (Dam)
Segment Boundary (No Dam)
EFDC Grid
Sediment Sample
(Synoptic OC only)
Figure 3.4b Foodchain Reaches 7f- 8 from Willow Mill Dam to Rising Pond Dam (after
WESTON, 2006). North is up in this figure.
37
-------
233
233
273
Housatonic River Downstream Model
Wataf Surface- Profile
223
213
V.
Bottom
Water -SLrtsze
TC
aha pan £l
^»«h 7D arid
JE
7F tt«
BtathTQ
v
%s
a
2S33 i33C> T5CC "CC23 12i33 15XO 17iCC 2ECC3 22i33 2i333 275CC 32S33
Dlciancwi |m|
Figure 3.5a Longitudinal Bottom Gradient Profile Showing Foodchain Reaches 7A - 8
(after WESTON, 2006)
290
280
270
g-260
I 250
E 240
230
220
210
Housatonic River Downstream Model
Water Surface Profile
Legend
Time: 4072.50
Intial Bottom
^ Bottom
Water Surface
Glendale Dam
Columbia
Mill Dam
Rising Pond
Lee
Willow Mill
, Dam ,
0 2500 5000 7500 10000 12500 15000 17500 20000 22500 25000 27500 30000 32500
Distance (m)
Figure 3.5b Longitudinal Bottom Gradient Profile in Reaches 1 and 8 Showing Location of
the Four Dams (after WESTON, 2006)
38
-------
Housatonic River Downstream Model
Figure 3.6a Computation grid for EFDC model - upstream boundary is at the top of this
figure. North is up in this figure.
39
-------
Housatonic River Downstream Model
Figure 3.6b Computation grid for EFDC model - lateral black line in the middle of this
stretch is Columbia Mill dam (arrow points to the dam). North is up in this
figure.
40
-------
Housatonic River Downstream Model
Water Depth
.1037 4050.00 8.573
Figure 3.6c Computation grid for EFDC model. North is up in this figure.
41
-------
Housatonic River Downstream Model
Water Depth
.1037 4050.00 8.573
Figure 3.6d Computation grid for EFDC model. North is up in this figure.
42
-------
Housatonic River Downstream Model
Water Depth
.1037 4050.00 8.573
Figure 3.6e Computation grid for EFDC model. North is up in this figure.
43
-------
Housatonic River Downstream Model
Figure 3.6.f Computation grid for EFDC model - lateral black line in the left third of this
stretch is Willow Mill dam (arrow points to the dam). North is up in this
figure.
44
-------
Housatonic River Downstream Model
Figure 3.6.g Computation grid for EFDC model - lateral black line on the right hand side
(i.e., upstream end) of this stretch is Willow Mill dam. North is up in this
figure.
45
-------
Housatonic River Downstream Model
Water Depth
.1037 4050.00 8.573
Figure 3.6.h Computation grid for EFDC model - Stockbridge golf course is in the middle
of this stretch. North is up in this figure.
46
-------
Housatonic River Downstream Model
Figure 3.6.1 Computation grid for EFDC model - lateral black line in the middle of this
stretch is Glendale dam (arrow points to the dam). North is up in this figure.
47
-------
Housatonic River Downstream Model
Figure 3.6.J Computation grid for EFDC model - lateral black line in the upper right corner
of this figure is Glendale dam. North is up in this figure.
48
-------
Housatonic River Downstream Model
Figure 3.6.k Computation grid for EFDC model - the downstream end of this figure is the
upper part of Rising Pond. North is up in this figure.
49
-------
Housatonic River Downstream Model
Figure 3.6.1 Computation grid for EFDC model - the arrow points to Rising Pond dam.
North is up in this figure.
50
-------
*&*?-
'#/.-' ., "/
*"IM
'.*%;*t£
Figure 3.7a Photo of Stockbridge golf course
51
-------
Housatonic River Downstream Model
Bottom Elev
245 4018.00 255
Figure 3.7b Grid in the area of the Stockbridge golf course. North is up in this figure.
52
-------
::.
Ifl
IS
mo
B3
63
44
0
j-
-
~.
_
La
fsr4([
-F
Jrm
f-e-
' '
i
itm<
con.
NC1
NC2
^ £
125 124 123 122 121 120 119 118 117 116 115 114 113 112 111 110 109 108 107 106 105
100
B3
63
3
1
100
BO
63
44
23
3
1
100
BO
63
44
23
'-
":
'linn
15
-
~i i i i i i
!5
Laye
_ _ 1 ITT7
1 1 1 1 1 1 II 1 1 1 1 14-H-
124 123
1 1 1 1 1 1 II 1 1 1 1 ll-H-
124 123
r3(
w
1
r2;
-
J
1
6f'-i;
22
ijr-1
22
1
*
121 120 119 118 117 116
jn
f i il 1 1 1 1 1 1 n 1 1 1 1 1 1 ii 1 1 1 1 1 1 1 ii 1 1 1 1 1 1 n 1 1 1 1 1 1 Ii
121 120 119 113 117
1 1 1 1 1 1
1
; Layer 1 (Iff1 - 24")
:
, ,,, | | | |,
'6
1
115 114 113 1'
1
2 111 110
i f '
115 114 113 112 111 11C
11
K
,,,,, , t
1 1 1 !
i
i-
9 108 107 10-6 10
i-
9 108 107 106 10
h
t-
j-
i i b
=
.2
4-1
u
ra--»
I-^P
u-l^
i/i
i/i
ra
2
c
g
"£
u
ra^,
t-^
U-£l
in
^i
ra
c
o
"5
u
n-^
td
i/i
w
n
s
125 124 123 122 121 120 115 118 117 11B 115 114 113 112 111 110 109 108 107 106 105
River Mile
Figure 3.8 Initial Longitudinal Distributions of Grain Sizes for the River Bed Sediment
53
-------
3000
geT2000
S|
Ji .3 1000
"5 """
0
1;
3000
4-1
'£^2000
Q -
J£ ^1000
"3 "~"
CO
0
1;
3000
>
'W^-2000
Q -
^ 5*1000
"5 """"
CD
0
1;
3000
*-«
Ji .S'lOOO
m
0
: 7A 7B 7C :
; ; ; U; M
| Layer 4 (0" - 6") ;
!5 124 123 122 121
: = ; U; U
Layer 3 (6" -12") :
!5 124 123 122 121
: Layer 2 (12" -18")
!5 124 123 122 121
I ; ; U; U
\ Layer 1 (18" - 24") ;
7D 7E 7F I 7Q 7H 8 ~.
;u ;\_;
120 119 118 117 116 115 114 113 112 111 110 109 108 107 106 10
; : : : ::
: : ; ; \ \
120 119 118 117 116 115 114 113 112 111 110 109 108 107 106 1C
: : : : : :.
:": ;U : ^
120 119 118 117 116 115 114 113 112 111 110 109 108 107 106 1C
I I I I , I ..,,..,,. I I 1 1 1 1 , , 1 1 1 , 1 :,"
125 124 123 122 121 120 119 118 117 116 115 114 113 112 111 110 109 108 107 106 105
River Mile
Figure 3.9 Initial Longitudinal Distributions of Bulk Densities for the River Bed Sediment
54
-------
1QO
g
".
1
g
>
1
g
>.
1
80
20
0
12
100
80
20
0
12
100
80
20
0
12
100
80
60
20
0
: 7A : 7B : 7C : 7D : 7E : 7F 7G : 7H :
; ; ; ; ; ; ; r j ;
' i i ft ri in: ill :
Layer 4 (0" - 6") : ; ; : ; ;
5 124 123 122 121 120 119 118 117 116 115 114 113 112 111 110 109 108 107 1C
i i ni i i i in i
- : : H n ;n; ill ;
Layer3(r-12;') ; ! ; ; ; ;
5 124 123 122 121 120 119 118 117 116 115 114 113 112 111 110 109 108 107 1C
- i H I M U i
' n H ^ \\ -
Layer 2 (12" -18") : ; ; : ; ;
5 124 123 122 121 120 119 118 117 116 115 114 113 112 111 110 109 108 107 1C
:j rtr: M in
Layer 1 (18" - 24") ; ; ; ; ; ;
8 !;
-~
H
6 10
n
:~
-_
6 10
ii
::
-_
6 10
I 1:
-_
125 124 123 122 121 120 119 118 117 116 115 114 113 112 111 110 109 108 107 106 105
River Mile
Figure 3.10 Initial Longitudinal Distributions of Porosities for the River Bed Sediment
55
-------
100
1 10
m 5
OOJ
\x
- 0)
01
1:
100
1 10
03 ^
V
Q. =5
01
1:
100
^
00 "O
£i
*- B) 1
^
01
1:
100
i 10
DO 5
o S>
Q. ₯
*- O) 1
0.1
7f^ 7B 7C 70 7E 7F
\ - - -
i i n
Layer 4 (
-------
10
u
o
u
o
u
o
u
o
8
6
/I
9
0
1;
10
8
6
4
2
0
1:
10
8
6
4
2
0
1;
10
8
6
4
2
0
7A - 7B - 7C - 7D 7E 7F -7
: : j j : :
Layer 4 (tf1 - b") 1 \ \ \
M ,1 | j | ;,,| | | | | i In ii in ill 1 1 1 1 liniii
5 124 123 122 121 120 119 118 117 116 115 114 113 112 111 110
_M u r
Layer 3 (6"^T2'1)
, , , i i ; i ; , , i i i i i ; i , , , , ; , , , , i i i i i i , , , ; , ,
5 124 123 122 121 120 119 118 117 116 115 114 113 112 111 110
: I ' 1 j (
Layer 2(1-2"- 1B")
, , , 1 | j | ; , , | | | | | i 1 1 1 1 1 i 1 1 1 1 1 1 1 1 1 1 1 1 1 i 1 1
5 124 123 122 121 120 119 118 117 116 115 114 113 112 111 110
i i r~ i i l
Layer i (iv - wj
Mil li I Ill I I I I I i I II II II Mil I 1 1 1 Ill II II
G 7H - 8 -
: : :-
: : ;;
: ::
: : ::
, , | | | , , , | :,-
109 108 107 106 1C
--
: : :-
1 M
: : :.
, , i i i , , , i ,
109 108 107 106 1C
: : :J
! ! ^
= : il
j >
"I
, , I | | , , , | ,-
109 108 107 106 1C
: : :J
- : ::
: : !-
j =:
"
1 | i 1 1 1 f
125 124 123 122 121 120 119 118 117 116 115 114 113 112 111 110 109 108 107 106 105
River Mile
Figure 3.12 Initial Longitudinal Distributions of Fractions of Organic Carbon for the River Bed Sediment
57
-------
58
-------
OC Normalized PCBs, Individual Samples and 1 Mile Bins
114
River Mile
100
1000
o
o
-S1
"S
0)
to
o
Q.
104
OC Normalized PCBs, Individual Samples and 1 Mile Bins
124
iL-TO
7^rr*?*
-------
4 MODEL EVALUATION
4.1 Model Application
The EFDC model was run to simulate sediment and tPCB transport in Reaches 7 and 8
over an 11-year period, from January 1, 1994 - December 31, 2004. This period was selected
because it includes the vast majority of the data record available for Reaches 7 and 8. For the
simulation, the EFDC model was spun-up using the boundary conditions for the one year period
prior to January 1, 1994. The simulated sediment bed conditions at the end of this one year spin-
up were used as the initial conditions for the simulation. The initial PCB concentrations at the
beginning of the 11 -year simulation were set equal to those at the beginning of the one-year spin-
up run since these initial concentrations were determined from data.
4.2 Model-Data Comparisons for Water Column TSS and PCBs
The only data set for water column PCB and TSS concentrations used to develop model-
data comparisons was collected at Division Street (the USGS Great Barrington gaging station),
located approximately 1 mile downstream of Rising Pond Dam. Simulated TSS and tPCB
concentrations at the outlet of Rising Pond (shown as black lines) are compared with the data
(shown as red squares) in Figures 4. la through 4. If in which the state variable is plotted on a log
scale. To make it easier to visualize the differences between model results and data, model
results corresponding to the time of data collection are indicated by blue open circles on days
when TSS or tPCB data were collected. Few data were collected in 1994-1995 (Figure 4. la).
However, in subsequent years, the sampling frequency increased to bimonthly to monthly in most
periods, and more frequently in some periods. In particular, there was a period from April to
May 1997 when relatively high frequency TSS sampling was conducted.
TSS concentrations measured at the outlet of Rising Pond were typically in the range of
1.5 to 15 mg/L. The low end of the range of simulated TSS concentrations decreases to
approximately 2.0 mg/L during low-flow conditions, consistent with the data. Simulated TSS
concentrations vary in response to changes in the hydrograph (upper panel), reaching values
greater than 100 mg/L several times per year. Data are typically not available at the time of these
simulated peak TSS concentrations. However, during March - May 1997, two high flow events
with peak flows greater than 1,000 cfs occurred. TSS measurements through this period ranged
from approximately 3 to 30 mg/L and showed considerable scatter. The TSS data also exhibited
a relationship (within the scatter) with flow corresponding to the peak flows of the two events.
Simulated TSS values during this period are well within the range of the measurements, and also
capture some of the relationship exhibited by the measurements. For example, the simulated and
measured peak TSS values are similar. Throughout the remainder of the simulation period, the
model results are in general agreement with the data, but sometimes underestimate the
measurements. The simulated TSS values are also typically less variable than the measurements.
These differences may be attributable, at least in part, to differences between the HSPF simulated
nonpoint runoff and the flows and solids loadings from the seven modeled tributaries that are
used as input to the EFDC model compared with the actual flows and loadings. That is, it needs
to be recognized that at the relatively fine temporal and spatial resolution of the EFDC model
application, even small differences in the timing and magnitude of HSPF simulated inputs could
60
-------
cause comparatively large differences between simulation results and measurements. For
example, Figure 4.2 shows the model-data comparisons for the April to June 1997 period. The
upper panel shows the comparison between the simulated flow (solid blue line) at Rising Pond
Dam and the measured flow at Division Street. As observed, the simulated flows are higher than
the measured flows for most of this three month period. While the TSS data are generally higher
than the simulated values, the vast majority of the differences are less than 5 mg/L, and thus
within the uncertainty associated with the measured values since the latter are based on samples
collected at one point in the cross-section at Division Street and the simulated values are the
cross-sectionally averaged values at the outflow from Rising Pond. Figure 4.3 shows the model-
data comparison for July - September 2004. Notwithstanding the extremely limited number of
measurements, the level of agreement seen in this figure between measured and simulated flows
and between the three measured values of TSS and tPCB and the simulated values is
representative of the model-data agreement achieved in the last six years of the 11-year
simulation.
Comparisons between simulated and measured tPCBs show similar behavior to that
described for TSS. Measured tPCB values vary narrowly from -0.01 to -0.05 |ig/L and do not
show pronounced temporal (e.g., seasonal) trends over time. Model results are consistently
within this range. However, the data are definitely impacted by changes in detection limits. For
example, model results overestimate measured tPCB concentrations when the detection limits are
lower, and at other times when detection limits are higher, the model results underestimate tPCB
concentrations. Overall, model results overestimate tPCB concentrations. However, as described
previously, this is likely a consequence of the differences between the actual watershed inputs
and those generated by HSPF.
4.3 Model Results for Sediment PCS Concentrations
Daily and spatially averaged (over each of the nine FCM reaches) organic carbon (OC)
normalized sediment tPCB concentrations over the 11-year simulation are shown in Figure 4.4.
In the majority of the modeling domain, sediment PCB exposure concentrations change by
relatively small amounts (less than +10%) over the 11-year simulation. Subreaches where
simulated changes were outside this range are limited to Reaches 7B (-17% change) and 7F (71%
change). The large increase in Reach 7F represents an increase in the tPCB concentration from
less than 40 to almost 70 mg/kg OC over the simulation period. This large relative percent
increase is attributed, at least in part, to uncertainty in the initial conditions for sediment tPCB
concentrations. Initial conditions in each of the flowing reaches (between impoundments) are
characterized by more uncertainty than impounded reaches because of varying sample density in
the different reaches. Sampling programs were targeted toward reaches that were more
depositional in character, e.g., impoundments. These included reaches 7B, 7E, 7G, and 8.
Because of the more limited number of samples collected in the flowing reaches, data from all of
these reaches were aggregated to develop an average initial tPCB concentration that was assigned
to each of the flowing reaches. Reach 7F has the smallest bed slope of the flowing reaches,
making it more conducive to deposition than the other similar reaches. The increase in sediment
tPCB concentrations simulated in Reach 7F reflects net deposition in that reach. This is further
discussed in Section 4.5.
61
-------
4.4 Evaluation of Model Performance
Figures 4.5 and 4.6 show plots (hereafter called cross-plots) of simulated versus measured
TSS and tPCB concentrations, respectively. The differences between measured and simulated
concentrations shown in these figures are attributed to, in part, phasing differences between the
simulated results at Rising Pond Dam and the data collected from Division Street. Phasing and
volumetric differences between actual flows (runoff and tributaries in Reach 7) and HSPF-
simulated flows from these sources contribute to the phasing difference between tPCB and TSS
water column data and EFDC simulated concentrations. This factor accounts for some
differences where the simulated tPCB concentrations are, for example, lower than the measured
values due to HSPF-simulated flows that are higher than actual flows, resulting in lower
simulated tPCB concentrations due to dilution.
As shown in Figure 4.5, a considerable number of model-data comparisons follow the
one-to-one line, with simulated values within a factor of 2 or 3 of the data. There is also a sizable
group of data that vary between 2 and 15 mg/L, with corresponding simulated TSS concentrations
falling within a narrow range of 2 to 4 mg/L (these appear as a dark band in Figure 4.5). The
regression of simulated versus measured TSS is strongly influenced by this little group of model-
data pairs because of the number of points falling within this group. Data within this group could
reflect variation from the data do to a variety do causes, but some of the comparisons are likely
influenced by the fact that the simulated hydrograph did not increase as quickly as the actual
flows. At the upper end of the range of simulated TSS concentrations, the simulated
concentrations exceed the measured concentrations, corresponding to transient flow events when
phase differences between simulated and measured hydrographs would be more pronounced.
The cross-plot of simulated and measured tPCB concentrations (Figure 4.6) include only
data for which sampling times were reported. The phasing issue discussed for TSS is further
exacerbated when the concentration data record does not include the time of sampling needed to
properly associate the data with the model output; therefore, these data were not included in the
cross-plot. For the majority of the detected concentrations, the simulated concentrations were
within a factor of two of the measured values, including elevated concentrations near 0.06 and
0.1|ig/L that are very close to the one-to-one line. Approximately 70% of the data were non-
detects (plotted at the detection limit) and the simulated concentrations at the sampling times
were typically near the detection limit. The one exception to this was a group of data collected in
1998 that had a relatively high detection limit of 0.125 |ig/L. Given the range of the detected
concentrations, it is unlikely that the actual tPCB concentrations were near this detection limit.
There are six model-data comparisons (with detected concentrations) for which the
simulated concentrations are not within a factor of two of the data. These points fall below the
one-to-one line and have a substantial effect on the regression of simulated versus measured
tPCB concentrations, that yields a slope of 0.27 and r2 = 0.17 if these points are excluded. The
highest tPCB concentration (0.26 |ig/L) was measured during the October 2003 high-flow event.
The tPCB concentration simulated at the exact time the sampling was conducted was 0.06|ig/L;
however, simulated concentrations increased to approximately 0.2|ig/L during that storm,
consistent with the magnitude of the measured concentration. This model-data comparison
highlights the significant effect that differences in the timing of actual and simulated hydrographs
have on the comparison of simulated and measured water column tPCB concentrations.
62
-------
Probability distributions of simulated and measured TSS and tPCB concentrations are
presented in Figures 4.7 and 4.8, respectively. In general, the simulated concentrations have
somewhat less variability than the measured concentrations, with the exception being the upper
5% of the TSS data. TSS concentrations above 20 mg/L reflect increased flow conditions when
phasing issues are more pronounced. The elevated tPCB detection limits for samples collected in
1998 complicate the comparison of the simulated and measured tPCB distributions shown in
Figure 4.8 because the actual concentrations could be substantially lower. This would shift the
detected concentrations above 0.04 |ig/L to the right on the plot, i.e., they would be less frequent
occurrences.
A statistical summary of the performance of the EFDC model for TSS and tPCB
concentrations is presented in Table 4.1. The Modeling Study QAPP (WESTON, 2000) presents
model performance measures for these and other model outputs. As shown in Table 4.1, the
relative bias at Rising Pond outlet is well within the model performance measure of ± 30% for the
TSS (-11.93%) and just outside the measure for PCB concentration (-31.97%). For median
relative error, the model performance measure is also ± 30%, and the EFDC model is within the
performance measure for both TSS (-27.12%) and PCB concentration (-3.32%).
4.5 Process-Based Flux Summaries
Process-based annual average mass fluxes for solids and PCBs calculated by the EFDC
model for Reaches 7 and 8 are presented in Figures 4.9 and 4.10 and summarized in Tables 4.2
and 4.3. The various mass fluxes in these summaries were not expected to balance exactly
because of rounding/truncation in the summation of the mass fluxes across each cell face at each
time step of the simulation. The residuals of the balance of the process-based annual average
mass flux terms represent a small fraction (0.1 to 6%) of the total annual mass flux into each
reach.
Solids mass fluxes are dominated by the water column advective fluxes, and increase
substantially in Reach 7D due to inputs from tributaries and direct runoff, both of which are
simulated by HSPF. Spatial patterns of erosion reflect the armored bed conditions in the steep
reaches (7 A, 7C, upper half of 7D, and 7H). Milder slopes in the lower half of Reach 7D and in
Reach 7F result in relatively high deposition and high erosion rates that produce a relatively small
net erosion flux in Reach 7D and a small net deposition flux in Reach 7F. Although these two
reaches have the smallest bottom slopes (i.e., gradients), relatively high erosion rates occur
during the rising limbs of runoff hydrographs, when previously deposited sediment is rapidly
eroded due to high bed shear stresses. Deposition on the floodplain is greatest in Reaches 7D and
7F, where the floodplain is considerably wider than in the other reaches domain. Bedload mass
fluxes of solids vary through the domain in response to the armoring conditions in the steeper
reaches and the physical barriers of the dams.
Water column advective PCB mass fluxes do not vary substantially through Reaches 7
and 8, decreasing from 29 kg/yr at the upstream boundary to 25.5 kg/yr at the downstream
boundary. PCB transport associated with bedload is a very small part of the overall PCB
transport balance, with annual average fluxes less than 0.005 kg/yr. PCB transport associated
with sediment erosion is higher in the impoundments of Columbia Mill Dam (Reach 7B), Willow
Mill Dam (Reach 7E), Glendale Dam (Reach 7G) and Rising Pond Dam (Reach 8) than in the
63
-------
steep-sloped flowing sections of the river. These higher PCB mass fluxes result from higher
tPCB concentrations in the impoundment sediment that are in turn related to higher fractions of
fine sediment with relatively higher organic carbon content. In Reaches 7D and 7F, PCB mass
fluxes associated with erosion are comparable to those in the impoundment reaches because of
the relatively higher rates of erosion during the rising limbs of runoff hydrographs.
As expected, depositional mass fluxes of PCBs are related to bed slope, with the highest
rates in impoundments and in the lower gradient reaches of 7D and 7F. Generally, deposition
rates on the floodplain are small, with the highest rates on the wider floodplains of the lower half
of Reach 7D and Reach 7F. Overall, volatilization losses represent approximately 11% of the
PCB inputs entering at the upstream boundary (Woods Pond Dam). Volatilization losses from
the river account for almost 90% of the total volatilization loss. The remainder is lost from water
transported across the floodplain.
The process-based annual average mass flux summaries highlight the processes
controlling solids and PCB transport through Reaches 7 and 8. The interactive effect of the
different bed slopes and solids composition characteristics of the flowing sections and
impoundments is seen in the relative magnitude of the various mass fluxes in these different
reaches. For example, although Reaches 7D and 7F have relatively high solids erosion rates, the
lower PCB concentrations in these long flowing sections result in PCB mass fluxes that are less
than the PCB mass fluxes in the relatively short impoundments, where higher fractions of fine-
grained sediment are found. These summaries provide an additional tool for understanding the
major processes controlling PCB fate in Reaches 7 and 8.
64
-------
Table 4.1
Statistical Evaluation of EFDC Model Performance,
January 1994 - December 2004
Statistical Summary of EFDC Model Performance
Station
No. of Data
(n)
TSS (mg/L)
Rising Pond Outlet
177
Average of
Data
Model Bias
Arithmetic
Model Bias
Geometric
Relative
Bias (%)
Mean
Absolute
Error
Median
Relative
Error (%)
Regression of Simulated vs.
Measured Values
Slope
Y-
Intercept
Coefficient of
Determination
(r2)
6.82
-0.81
0.77
-11.93
4.88
-27.12
0.31
0.90
0.11
PCB (jig/L)
Rising Pond Outlet
28
0.055
-0.02
0.77
-31.97
0.02
-3.32
0.27
-2.52
0.17
65
-------
Table 4.2
Process-Based Annual Average Mass Flux Summary Tabulation for Solids
Channel/Floodplain Process
Channel Water Column Sources
Advection
Bed load
Import from Floodplain
HSPF Tributary and Surface Runoff
Erosion
Sum
Channel Water Column Sinks
Downstream Advection
Bed load
Export to Floodplain
Deposition
Sum
Floodplain Water Column Sources
Import from Channel
Overland Flow
Erosion
Sum
Floodplain Water Column Sinks
Export to Channel
Overland Flow
Deposition
Sum
Reach Residual
Solids Flux (MT/yr)
7A
2352.4
0.0
188.9
68.3
2609.7
2349.0
6.1
189.0
65.6
2609.7
189.0
10.2
199.2
189.9
9.1
199.0
0.2
7B
2349.0
6.1
141.8
10.9
263.0
2770.8
2520.0
0.0
250.8
2770.8
189.9
12.0
201.8
141.8
0.0
14.4
156.2
45.6
7C
2520.0
0.0
33.0
0.8
88.5
2642.3
2546.0
13.1
83.2
2642.3
0.0
3.4
3.4
279-433.0
12.8
9.1
54.8
-51.4
7D
2546.0
13.1
1928.3
2793.4
7280.8
4235.0
0.0
279.4
2766.4
7280.8
12.8
579.9
872.1
-4.4
714.5
710.1
162.1
7E
4235.0
0.0
1.8
512.0
4748.8
4315.0
0.0
21.5
412.3
4748.8
21.5
-4.4
0.0
17.1
0.0
0.0
0.0
17.1
7F
4315.0
0.0
272.0
563.9
3467.2
8618.1
4677.0
223.1
3718.0
8618.1
0.0
516.6
516.6
156-2572.0
0.2
529.9
802.1
-285.4
7G
4677.0
223.1
1.0
273.4
5174.5
4826.0
0.0
156.6
191.9
5174.5
0.2
33.2
189.9
0.0
76.8
76.8
113.2
7H
4826.0
0.0
2.6
410.7
5239.3
4769.0
27.8
44.5
398.1
5239.3
44.5
0.0
2.3
46.7
0.0
38.4
38.4
8.3
8
4769.0
27.8
1.9
1215.8
6014.4
4663.8
0.0
184.1
1166.6
6014.4
184.1
0.0
22.2
206.3
0.0
154.5
154.5
51.8
66
-------
Table 4.3
Process-Based Annual Average Mass Flux Summary Tabulation for PCBs
Channel / Floodplain Process
Channel Water Column Sources
Advection
Bed load
Import from Floodplain
Diffusion
Erosion
Sum
Channel Water Column Sinks
Downstream Advection
Bed load
Export to Floodplain
Volatilization
Deposition
Sum
Floodplain Water Column Sources
Import from Channel
Overland Flow
Diffusion
Erosion
Sum
Floodplain Water Column Sinks
Export to Channel
Overland Flow
Volatilization
Deposition
Sum
Reach Residual
PCB Flux (kg/yr)
7A
29.04
0.00
0.00
0.00
29.04
26.10
0.00
2.62
0.31
0.01
29.04
2.62
0.00
0.01
2.63
2.59
0.03
0.00
2.62
0.01
7B
26.10
0.00
2.68
0.09
0.24
29.11
28.80
0.00
0.15
0.16
29.11
2.59
0.01
0.03
2.62
2.68
0.00
0.01
0.01
2.70
-0.08
7C
28.80
0.00
0.02
0.02
28.83
28.40
0.00
0.23
0.15
0.05
28.83
0.23
0.00
0.00
0.00
0.23
0.20
0.00
0.03
0.22
0.00
7D
28.40
0.00
0.12
0.21
28.73
25.80
0.00
0.24
0.86
1.82
28.73
0.24
0.20
0.14
0.38
0.95
0.01
0.19
0.89
1.09
-0.14
7E
25.80
0.00
0.02
0.28
26.10
25.70
0.00
0.06
0.06
0.28
26.10
0.06
0.01
0.01
0.01
0.07
0.00
0.01
0.00
0.01
0.07
7F
25.70
0.00
0.06
0.12
0.44
26.32
23.50
0.00
0.62
2.20
26.32
0.00
0.10
0.47
0.57
0.06
0.00
0.18
0.54
0.78
-0.20
7G
23.50
0.00
0.18
0.02
0.70
24.39
24.30
0.00
0.03
0.06
24.39
0.00
0.00
0.04
0.04
0.18
0.00
0.01
0.03
0.22
-0.18
7H
24.30
0.00
0.02
0.01
0.01
24.34
23.80
0.00
0.38
0.16
24.34
0.00
0.00
0.00
0.00
0.02
0.03
0.00
0.02
0.07
-0.07
8
23.80
0.00
0.20
1.78
25.78
25.51
0.00
0.29
0.53
26.33
0.00
0.03
0.00
0.00
0.04
0.01
0.01
0.02
-0.53
67
-------
10000
Rising Pond Outlet
1000
° 100
JFMAMJJASONDJFMAMJJASOND
HMO
JFMAMJJASONDJFMAMJJASOND
0.001
JFMAMJJASONDJFMAMJJASOND
January 1994 to December 1995
Model
Observed above DL
< Observed at or below DL
O Model result at field sample time
Figure 4. la Comparison of Simulated (Solid Line) and Measured (Red Symbols) TSS and tPCB
Concentrations in the Water Column for 1994-1995. DL = detection limit
68
-------
10000
« 1000
o
T
2 100
10
1000
Rising Pond Outlet
J FMAMJJASONDJ FMAMJJASOND
100
O)
e
8
Mw«ws8f
JFMAMJJASONDJFMAMJJASOND
10
0.1
CO
o
2: 0.01
0.001
J FMAMJ JA80NDJ FMAMJ JASOND
January 1996 to December 1997
Model
Observed above DL
< Observed at or below DL
O Model result at field sample time
Figure 4. Ib Comparison of Simulated (Solid Line) and Measured (Red Symbols) TSS and tPCB
Concentrations in the Water Column for 1996-1997. DL = detection limit
69
-------
10000
W 1000
*»»
H
o 100
10
1000
Rising Pond Outlet
JFMAMJJASONDJFMAMJJASOND
JFMAMJJASONDJFMAMJJASOND
10
CQ
O
£ 0,01
0.001
J FMAMJJASONDJ FMAMJJASOND
January 1998 to December 1999
Model
Observed above DL
< Observed at or below DL
O Model result at field sample time
Figure 4. Ic Comparison of Simulated (Solid Line) and Measured (Red Symbols) TSS and tPCB
Concentrations in the Water Column for 1998-1999. DL = detection limit
70
-------
10000
Rising Pond Outlet
JFMAMJJASONDJFMAMJJASOND
1000
J FMAMJJASONDJ FMAMJJASOND
0,001
JFMAMJJASONDJFMAMJJASOND
January 2000 to December 2001
Model
Observed above DL
< Observed at or below DL
O Model result at field sample time
Figure 4. Id Comparison of Simulated (Solid Line) and Measured (Red Symbols) TSS and tPCB
Concentrations in the Water Column for 2000-2001. DL = detection limit
71
-------
10000
Rising Pond Outlet
JFMAMJJASONDJFMAMJJASOND
1000
J FMAMJ JASONDJ FMAMJ JASOND
10
0.1
O
S: 0.01
0.001
JFMAMJJASONDJFMAMJJASOND
January 2002 to December 2003
Model
Observed above DL
< Observed at or below DL
O Model result at field sample time
Figure 4. le Comparison of Simulated (Solid Line) and Measured (Red Symbols) TSS and tPCB
Concentrations in the Water Column for 2002-2003. DL = detection limit
72
-------
10000
W 1000
*»»
Q
o 100
10
1000
Rising Pond Outlet
JFMAMJJASONDJFMAMJJASOND
100
O)
W>
CO
J F" M A M J JASONtyj FMAMJ JASOND
10
CQ
O
-------
10000
1000
Jr 100
u_
Rising Pond Outlet
3. o.i
a
u
gj 0.01
0.001
1
01-Apr-1997 to 30-Jun-1997
Model
Observed a
Observed above DL
< Observed at or below DL
O Model result at field sample time
Figure 4.2 Comparison of Simulated (Solid Line) and Measured (Red Symbols) TSS and tPCB
Concentrations in the Water Column for April - June 1997. DL = detection limit
74
-------
10000
1000
.r 100
u.
10
1000
100
10
10
Rising Pond Outlet
=. 0.1
m
o
Q; 0.01
0.001
I I
I I
A;
1
1 k
^
/I!.:
01-Ju!-2004 to 30-Sep-2004
Model
Observed Q
Observed above CL
'' Observed at or below DL
O Model result at field sample time
Figure 4.3 Comparison of Simulated (Solid Line) and Measured (Red Symbols) TSS and tPCB
Concentrations in the Water Column for July - September 2004. DL = detection limit
75
-------
450
Jan-94 Jan-95 Jan-96 Jan-97 Jan-98 Jan-99 Jan-00 Jan-01 Jan-02 Jan-03 Jan-04
Reach 7A
Reach 7F
Reach 7B
Reach 7G
Reach 7C
Reach 7H
Reach 7D
Reach 8
Reach 7E
Note: Colored circle symbols indicate sediment initial conditions used at the beginning of EFDC spin-up.
Figure 4.4 Temporal Trend of tPCB Concentrations in Surface Sediments for Foodchain Reaches
76
-------
10
3 Rising Pond Outlet
O)
10
i 11 nil l i i 111 ill I i i i mil l i i i iii
mill
mill
= 0.11
= 6.29=
Intercept = o.94:
N= 183
i i
i mil
; 183-
i i i mill
10 " 10 10 10 10
TSS (mg/L) Measured
Note: Statistics evaluated on log-transformed values
Non-Detects Excluded from regression
N = Number of data points excluding non-detects
Figure 4.5 Cross-Plot of Simulated and Measured TSS Concentrations
77
-------
10
o Rising Pond Outlet
m
3
E
55
GO
o
10
r* = 0.01
- Slope = 0.06
Intercept = -3.23
N = 101
IIIU
X-
-2
-1
10 10 ' 10
tPCB (ug/L) Measured
Note: Statistics evaluated on log-transformed values
Non-Detects Excluded from regression
N = Number of data points excluding non-detects
Figure 4.6 Cross-Plot of Simulated and Measured tPCB Concentrations
78
-------
10
3 Rising Pond Outlet
10
1
Eio1
10
10
11 mill I f
dOv
N=183 -
iiiiiiil I i i I i i I
0.1 1 1020 50 8090 99 99.9
Probability (%)
Hollow Symbols - Data
Filled Circles - Model
N = Number of data points including non-detects
Figure 4.7 Probability Distributions of Simulated and Measured TSS Concentrations
79
-------
10
Rising Pond Outlet
D)
3
>*'
co
o
o.
10
10
D
D
n *
N=101
0.1 1 1020 50 8090 99 99.9
Probability (%)
Hollow Symbols - Data
Filled Circles - Model
N = Number of data points including non-detects
Figure 4.8 Probability Distributions of Simulated and Measured tPCB
Concentrations
80
-------
Solids Mass Flux (MT/year) Summary - Downstream Model Run (1994 - 2004)
Floodplaln
Import Item Channd
Export lo CHa<l
Main Channel
Expod to Roodpl*n
hrrpoit Irom Fioodp am
Adject on
Sedload
HSPF Thbutary ant Sudx» Riraff
Erosion
i T T
I t t
i t t
i T t
i 1 1
Numerical Residuals (MT/year)
neazh 7A Main Channel are Foodpair
R*ach 78 Main Chann»l and Floodplain
Reach 7C Main Channel and Floodplam
Reach 7D Main Channel and Rwd^ao
Reach 7E Main Channel and Roodplaini
Reach 7F Main Chamsl and Ftoodpiain
Reach 7G Main Channel and Roodptain
Reach 7H Main Channel and Raodplan
R»ach 8 Main Chamil and Floodpiair*
Oownstieam Model Domain
IK
41
162
17
2M
113
S
52
61
Figure 4.9 Process-Based Annual Average Mass Flux Summary for Solids
81
-------
PCB Mass Flux (Kg/year) Summary - Downstream Model Run (1994 - 2004}
Floodplain
Ifeatox 0.03 0.01
2.68
Main Channel
Import from FloodpJa n
Export to Roodpiain
Numerical Residuals (Kg/year)
Reach 7A Main Channel and Floodpiain
Reach 7E Main Channel and Roodpiain
R#«h 7C Mftr, Chann*! and Rwdpl&n
Reach 7D Main Cnannel and Roodpiain
Reach 7 E Main Channel and Roodpiain
Reach 7F MOT Channel and Roodpiain
R#ach 7G Main Chvirwl and Roodptairi
Reach 7H Mam Chan-el and Roodpiain
Reach 8 Man Channel and Roodpiain
Dowrislf*am Mod*) Domari
11 L
J i
O.lfl 0.01 0.00 0.01
0.06 0.18 0.02
0,00
LL
" "
Stomon
D.lluacn
^^^B 2J8
t t
1 1 1
0.01
OJOO
Ruch7A
0.00
t t
1 1 1
0.03
0.01
Reach 7B
11 JO
1 t t
1 1 1
0.00
0.00
Roach 7C
0.01
1 t t
t
0.38
0.14
RMCh 70
ana
\ t t
1
0.01
001
RfSi:l)7E
0.00
t t
1
0.47
0.10
Reach 7F
0,00
I t t
^L i
OJOO
RHCllTQ
0,03
1 t t
* i T
oaa
i!-"i.h7i:
1 t t
1
MO
OjOC
Reach 5
0.06 0.18
0.24 0.06
VoMfertM
Drtlusion
0.31
t t
^^^p .,..,
1 t t
1 1 1
0.00
0.00
Reach 7A
0.15
i t
U.UU
I t t
4 I I
0.24
0.03
R*ld<7B
0.15
t t
ujjg
I t T
i I
0.02
0.02
Roach 7C
0.86
' t
^piHi^Hm
i t t
i i i
021
0.12
HiMi.h7D
0X6
I t
UOJM
1 t T
i i i
OJS
002
B.-.nli7E
O.E2
I t
T^^^^^^^^
1 t t
4 1 1
OM
0.12
R«ach7F
003
1 t
^^^^^^^^^
I t t
1 1 1
0.70
0.02
RMChTQ
0.38
I t
QjQQ
1 f t
4 1
001
0.01
Reach 7H
029
*
1 t t
4 1
1.78
O.X
R«ache
-4±l!
0,00
»
0.010
-O.Gfl1
0X102
-0.140
0.068
-OJ03
-0.176
-0.074
-OJ32
Figure 4.10 Process-Based Annual Average Mass Flux Summary for PCBs
82
-------
5 CONCLUSION
As stated in Section 1, the purpose of this modeling effort was to evaluate the ability of
EFDC, the state-of-the-art contaminated sediment modeling system, to simulate the transport and
fate of a contaminant over a time period of at least 10 years. Evaluation of the results of an 11-
year simulation using the minimally calibrated EFDC Model indicate that this model is capable of
simulating the transport and resultant concentrations of TSS and PCBs in Reaches 7 and 8 of the
Housatonic River within the specified model performance measures. Specifically, a statistical
summary of the performance of the EFDC model for TSS and PCB concentrations found that the
relative bias at Rising Pond outlet is well within the model performance measure of ± 30% for
TSS (-11.93%) and just outside the measure for PCB concentrations (-31.97%). For median
relative error, the model performance measure is also ± 30%, and the EFDC model is within the
performance measure for both TSS (-27.12%) and PCB (-3.32%) concentrations. Considering the
fact that the model was only minimally calibrated, and that the system modeled had widely
varying hydraulic and morphologic regimes, the EFDC model's performance, as quantified by the
relative bias and the median relative error, is considered good. This demonstrates that EFDC is a
robust modeling system that can be successfully implemented at other contaminated sediment
sites.
83
-------
REFERENCES
Arega, F. W., and E. J. Hayter (2006). "Skill Assessment of EFDC - a 3D Hydrodynamic and
Transport Model - to Simulate Highly Stratified Estuarine Flows", submitted to Journal of
Estuarine, Coastal and Shelf Science in April 2006.
Ariathurai, R. (1974). "A Finite Element Model for Sediment Transport in Estuaries", Ph.D.
Dissertation., University of California, Davis, California.
Ariathurai, R., and K. Arulanandan (1978). "Erosion Rates of Cohesive Soils", Journal of the
Hydraulics Division, ASCE, 104(HY2), 279-283.
Aris, R. (1956). "On the Dispersion of a Solute in a Fluid Flowing Through a Tube", Proceedings
Royal Society of London, Series A, 235, 61-11.
Bain, A. J. (1981). "Erosion of Cohesive Muds", M.S. Thesis, University of Manchester,
United Kingdom, 1981.
Barnes, R. S. K., and J. Green, Editors. (1971). The Estuarine Environment, Applied Science
Publishers, London.
BBL (Blasland, Bouck & Lee, Inc.) (1994). MCP Interim Phase IIReport and Current
Assessment Summary for East Street Area 2/U.S. EPA Area 4. Volumes I, II, III, IV, V, VI,
VII, IX, X, and XII.
BBL (Blasland, Bouck & Lee, Inc) and QEA (Quantitative Environmental Analysis, LLC).
(2003). Housatonic River - Rest of River RCRA Facility Investigation Report. Prepared for
General Electric Company, Pittsfield, MA.
Been, K., and G. C. Sills. (1981). "Self-weight Consolidation of Soft Soils: An Experimental
and Theoretical Study," Geotechnique,3\(4), 519-535.
Bellessort, B. (1973). "Movement of Suspended Sediments in Estuaries-Flocculation and Rate of
Removal of Muddy Sediments," Tracer Techniques in Sediment Transport, International Atomic
Energy Agency, Technical Report Series No. 145, 31-40.
Burkhard, L.P. (2000). "Estimating dissolved organic carbon partition coefficients for
nonionic organic chemicals", Environmental Science and Technology 34(22):4663-4668.
Burt, T. N., and W. R. Parker. (1984). "Settlement and Density in Beds of Natural Mud During
Successive Sedimentation", Report IT 262, Hydraulics Research, Wallingford, United Kingdom.
84
-------
Christensen, B. A. (1965). "Discussion of'Erosion and Deposition of Cohesive Soils'," by
Partheniades, E., Journal of the Hydraulics Division., ASCE, 91(HY5), 301.
Christensen, R. W., and B. M. Das. (1973). "Hydraulic Erosion of Remolded Cohesive Soils,"
Special Report 135, Highway Research Board, Washington, D.C., 8-19.
Chang, F. M., D. B. Simons, and E. V. Richardson. (1965). "Total Bed-Material Discharge in
Alluvial Channels," U.S. Geological Survey Water-Supply Paper 1498-1.
Engelund, F. and E. Hansen (1967). A monograph on sediment transport in alluvial streams,
Teknisk Forlag, Copenhagen, Denmark. 62 pp.
Chapra, S. C. (1997). Surface Water-Quality Modeling, McGraw-Hill Companies, Inc.
Di Toro, D.M. (1984). "Probability model of stream quality due to runoff, Journal of
Environmental Engineering. ASCE 110(3):607-628.
Di Toro, D.M. (1985). "A particle interaction model of reversible organic chemical sorption",
Chemosphere 14(10): 1503-1538.
Di Toro, D.M., C.S. Zarba, D.J. Hansen, WJ. Berry, R.C. Swartz, C.E. Cowan, S.P. Pavlou,
H.E. Allen, N.A. Thomas, and P.R. Paquin. (1991). "Technical basis for establishing sediment
quality criteria for nonionic organic chemicals by using equilibrium partitioning",
Environmental Toxicology and Chemistry 10:1541-1583.
Di Toro, D.M. (2000). Sediment Flux Modeling. John Wiley & Sons, Inc, New York, NY.
Dixit, J. G. (1982). "Resuspension Potential of Deposited Kaolinite Beds," M.S. Thesis,
University of Florida, Gainesville, Florida.
Dyer, K. R. (1986). Coastal andEstuarine Sediment Dynamics, John Wiley and Sons,
London.
Dyer, K. R. (1989). "Sediment processes in estuaries: future research requirements," Journal of
Geophysical Research, 94(C10), 14327-14339.
Edzwald, J. K., J. B. Upchurch, and C. R. O'Melia. (1974). "Coagulation in Estuaries,"
Environmental Science and Technology, 8(1), 58-63.
Einstein, H. A. (1950). "The Bed-Load Function for Sediment Transportation in Open
Channel Flows," U.S. Department of Agriculture, Soil Conservation Service, Technical
Bulletin no. 1026.
Einstein, H. A., and N. L. Barbarossa. (1952). "River Channel Roughness," Transactions of
the ASCE, 117, 1121-1132.
85
-------
Einstein, H. A., and R. B. Krone. (1962). "Experiments to Determine Modes of Cohesive
Sediment Transport in Salt Water," Journal of Geophysical Research., 67(4), 1451-1464.
Fischer, H. B. (1966). "Longitudinal Dispersion in Laboratory and Natural Streams," Ph.D.
Dissertation, California Institute of Technology, Pasadena, California.
Fischer, H. B. (1972). "Mass Transport Mechanisms in Partially Stratified Estuaries," Journal
of Fluid Mechanics, 53(4), 671-687.
Fischer, H. B., J. Imberger, E. J. List, R. C. Y. Koh, andN. H. Brooks. (1919}. Mixing in Inland
and Coastal Waters, Academic Press, New York.
Galperin, B., L. H. Kantha, S. Hassid, and A. Rosati (1988). "A Quasi Equilibrium Turbulent
Energy Model for Geophysical Flows", Journal of Atmospheric Science, 45, pp. 55-62.
Garcia, M., and G. Parker (1991). "Entrainment of bed sediment into suspension", J. Hyd.
Engrg., 117, 414-435.
Gessler, J. (1965). "The Beginning of Bedload Movement of Mixtures Investigated as Natural
Armoring in Channels," W. M. Keck Laboratory of Hydraulics and Water Resources,
California Institute of Technology, Pasadena.
Gessler, J. (1970). "Self-stabilizing Tendencies of Alluvial Channels," Journal of the
Waterways and Harbors Division, ASCE, 96(WW2), 235-249.
Grim, R. E. (1968). Clay Mineralogy, McGraw-Hill, New York.
Grimshaw, R. W. (1971). The Chemistry and Physics of Clays, Wiley-Interscience, New
York.
Hamrick, J. M. (1992). "A Three-Dimensional Environmental Fluid Dynamics Computer Code:
Theoretical and Computational Aspects", Special Report No. 317 in Applied Marine Science and
Ocean Engineering, Virginia institute of Marine Science, Gloucester Point, VA., pp. 1-64.
Hamrick, J. M and W. B. Mills (2000). "Analysis of Water Temperatures in Conowingo Pond as
Influenced by the Peach Bottom Atomic Power Plant Thermal Discharge", Environmental
Science and Policy, 3(1), pp. 197-209.
Hanzawa, H., and T. Kishida. (1981). "Fundamental Considerations of Undrained Strength
Characteristics of Alluvial Marine Clays," Soils and Foundation, Japanese Society of Soil
Mechanics and Foundation Engineering, 21(1), 39-50.
Harza. (2001). Woods Pond Dam: Structural Integrity Assessment. Prepared for General
Electric Company, Pittsfield, MA.
86
-------
Hawker, D. W., and D. W. Connell. (1988). "Octanol-water partition coefficients of
polychlorinated biphenyl congeners", Environmental Science and Technology, 22:382-387.
Hayter, E. J. (1983). "Prediction of the Movement of Cohesive Sediments in Estuarial
Waters," Ph.D. Dissertation, University of Florida, Gainesville, Florida.
Hayter, E. J., Bergs, M., Gu, R., McCutcheon, S., Smith, S. J., and Whiteley, H. J. (1999).
"HSCTM-2D, A Finite Element Model for Depth-Averaged Hydrodynamics, Sediment and
Contaminant Transport", Technical Report, EPA Ecosystems Research Division, Athens,
Georgia.
Hayter, E. J., V. Paramygin, and C. V. John. (2003). "3D Modeling of Cohesive Sediment
Transport in a Partially Stratified Microtidal Estuary to Assess Effectiveness of Sediment Traps",
7th International Conference on Nearshore and Estuarine Cohesive Sediment Transport
Processes, Virginia Institute of Marine Science, Gloucester Point, VA, October 2003.
Holley, E. R. (1969). "Unified View of Diffusion and Dispersion," Journal of the Hydraulics
Division, ASCE, 95(HY2).
Holley, E. R., D. R. F. Harleman, and H. B. Fischer. (1970). "Dispersion in Homogeneous
Estuary Flow," Journal of the Hydraulics Division, ASCE, 96(HY8), 1691-1709.
Hunt, J. R. (1980). "Prediction of Ocean Particle Size Distribution from Coagulation and
Sedimentation Mechanisms," Advances in Chemistry Series No. 189 - Particles in Water, M. D.
Kavanaugh and J. O. Keckie, Editors, American Chemical Society, Washington, D.C., 243-257.
Imhoff, J. C., A. Stoddard, and E. M. Buchak (2003). "Evaluation of Contaminated Sediment
Fate and Transport Models", Final Report, U.S. EPA, National Exposure Research Laboratory,
Athens, GA.
Imhoff, J.C., J. Clough, R.A. Park, and A. Stoddard (2004). "Evaluation of Chemical
Bioaccumulation Models for Aquatic Ecosystems", Final Report, U.S. EPA, National Exposure
Research Laboratory, Athens, GA, 122 pp.
Jepsen, R., J. Roberts, and W. Lick. (1997). "Effects of Bulk Density on Sediment Erosion
Rates," Water, Air, and Soil Pollution, 99, 21-31.
Ji, Z.-G., M. R. Morton, and J. M. Hamrick (2000). "Modeling Hydrodynamic and Sediment
Processes in Morro Bay", Estuarine and Coastal Modeling: Proc., 6th Int. Conf, M. L. Spaulding
and H. L. Butler, eds., ASCE, New York, pp. 1035-1054.
Ji, Z.-G., M. R. Morton, and J. M. Hamrick (2001). "Wetting and Drying Simulation of Estuarine
Processes", Estuarine, Coastal Shelf Sci. 53( 5), pp. 683-700
Jin, K. R., Z.-G. Ji, and J. M. Hamrick (2002). "Modeling Winter Circulation in Lake
87
-------
Okeechobee, Florida," Journal of Waterways, Port, Coastal, Ocean Eng., 128(3), pp. 114-125.
Jin, K.-R., J. M. Hamrick, and T. Tisdale (2000). "Application of a Three Dimensional
Hydrodynamic Model for Lake Okeechobee," Journal of Hydraulic Engineering, 126(10), pp.
758-771.
Jobson, H. E., and W. W. Sayre. (1970). "Vertical Transfer in Open Channel Flow," Journal
of the Hydraulics Division, ASCE, 96(HY3), 703-724.
Kandiah, A. (1974). "Fundamental Aspects of Surface Erosion of Cohesive Soils," Ph.D.
Dissertation, University of California, Davis, California.
Karcz, I, and G. Shanmugam. (1974). "Decrease in Scour Rate of Fresh Deposited Muds,"
Journal of the Hydraulics Division, ASCE, lOO(HYll), 1735-1738.
Karickhoff, S. W., D. S. Brown, and T. A. Scott. (1979). "Sorption of hydrophobic pollutants
on natural sediments", Water Research 13:241-248.
Karickhoff, S. W. (1981). "Semi-empirical estimation of sorption of hydrophobic pollutants
on natural sediment and soils", Chemosphere 10(8):833-846.
Karickhoff, S. W. (1984). "Organic pollutant sorption in aquatic systems", ASCE, J.
Hydraulic Engineering 110(6):707-735.
Kelley, W. E., and R. C. Gularte. (1981). "Erosion Resistance of Cohesive Soils," Journal of
the Hydraulics Division, ASCE, 107(HY10).
Kim, S., D. L. Wright, J. Maa and J. Shen (1997). "Morphodynamic Responses to Extra Tropical
Meteorological Forcing on the Inner Shelf of the Middle Atlantic Bight: Wind Wave, Currents
and Suspended Sediment Transport", Proc. ofEstuarine and Coastal Modeling, pp. 456-66.
Kirby, R., and W. R. Parker. (1977). "The Physical Characteristics and Environmental
Significance of Fine Sediment Suspensions in Estuaries," Estuaries, Geophysics and the
Environment, National Academy of Sciences, Washington, D.C., 110-120.
Kranck, K. (1980). "Sedimentation Processes in the Sea," The Handbook of Environmental
Chemistry, 2(A), O. Hutzinger, Editor, Springer-Verlag, Berlin, 61-75.
Krone, R. B. (1962). "Flume Studies of the Transport of Sediment in Estuarial Shoaling
Processes," Final Report, Hydraulic Engineering Laboratory and Sanitary Engineering
Research Laboratory, University of California, Berkeley, California.
Krone, R. B. (1963). "Study of Rheological Properties of Estuarial Sediments," Technical
Bulletin No. 7, Committee of Tidal Hydraulics, U.S. Army Corps of Engineers, Vicksburg,
Mississippi.
88
-------
Krone, R. B. (1972). "Field Study of Flocculation as a Factor in Estuarial Shoaling
Processes," Technical Report No. 19, Committee on Tidal Hydraulics, U.S. Army Corps of
Engineers, Vicksburg, Mississippi.
Lambermont, J., and T. Lebon. (1978). "Erosion of Cohesive Soils," Journal of Hydraulic
Research, 16(1), 27-44.
Lee, P. T. (1974). "Deposition and Scour of Clay Particles Due to Currents," M.S. Thesis,
Asian Institute of Technology, Bangkok, Thailand.
Lick, W., J. Lick, and C. K. Ziegler (1994). "The resuspension and transport of fine-grained
sediments in Lake Erie", Journal of Great Lakes Research 20(4):599-612.
Mathew, R. M., E. J. Garland, and P. R. Paquin. (2002). "Partitioning of PCBs and
importance to fate and effects", Presented at the 2002 North American SET AC Conference,
Salt Lake City, UT, November 2002.
McNeil, J., C. Taylor, and W. Lick. (1996). "Measurements of the Erosion of Undisturbed
Bottom Sediments with Depth," Journal of Hydraulic Engineering, ASCE, 122(6), 316-324.
Mehta, A. J. (1981). "Review of Erosion Function for Cohesive Sediment Beds," Proceedings
of the First Indian Conference on Ocean Engineering, Indian Institute of Technology,
Madras, India, Vol. 1, 122-130.
Mehta, A. J., and E. Partheniades. (1973). "Depositional Behavior of Cohesive Sediments,"
Technical Report No. 16, Coastal and Oceanographic Engineering Laboratory, University of
Florida, Gainesville, Florida.
Mehta, A. J., and E. J. Hayter. (1981). "Preliminary Investigation of Fine Sediment Dynamics
of Cumbarjua Canal, Goa, India," UFL/COEL-81-012, Coastal and Oceanographic
Engineering Department, University of Florida, Gainesville, Florida.
Mehta, A. J., T. M. Parchure, J. G. Dixit, and R. Ariathurai. (1982a). "Resuspension Potential
of Deposited Cohesive Sediment Beds," Estuarine Comparisons, V. S. Kennedy, Editor,
Academic Press, New York, 591-609.
Mehta, A. J., E. Partheniades, J. Dixit, and W. H. McAnally. (1982b). "Properties of
Deposited Kaolinite in a Long Flume," Proceedings of the Hydraulics Division Conference
on: Applying Research to Hydraulic Practice, ASCE, Jackson, Mississippi.
Mehta, A. J., and J. W. Lott. (1987). "Sorting of fine sediment during deposition,"
Proceedings of Coastal Sediments '87, N. C. Kraus ed., ASCE, New York, 348-362.
Mellor, G. L., and T. Yamada (1982). "Development of a Turbulence Closure Model for
Geophysical Fluid Problems", Rev. Geophys. Space Phys, 8, pp. 851-875.
89
-------
Meyer-Peter, E. and R. Muller (1948). "Formulas for bed-load transport", Proc. Int. Assoc. Hydr.
Struct. Res., Report of Second Meeting, Stockholm, 39-64.
Mitchell, J. K. (1961). "Fundamental Aspects of Thixotropy in Soils," Transactions of the
ASCE, 126(1), 1586-1620.
Odd, N. M. V., and A. J. Cooper. (1989). "A two-dimensional model of the movement of
fluid mud in a high energy turbid estuary," Journal of Coastal Research., SIS, 185-193.
Okubo, A. (1973). "Effect of Shoreline Irregularities on Streamwise Dispersion in Estuaries
and Other Embayments," Netherlands Journal of Sea Research, 6, 213-224.
Owen, M. W. (1971). "The Effect of Turbulence on the Settling Velocities of Silt Floes,"
Proceedings of the Fourteenth Congress ofl.A.H.R., 4, Paris, 27-32.
Owen, M. W. (1975). "Erosion of Avonmouth Mud," Report No. INT 150, Hydraulics
Research Station, Wallingford, United Kingdom.
Paaswell, R. E. (1973). "Causes and Mechanisms of Cohesive Soil Erosion: The State of the
Art," Special Report 135, Highway Research Board, Washington, D.C., 52-74.
Parchure, T. M. (1980). "Effect of Bed Shear Stress on the Erosional Characteristics of
Kaolinite/'M^. Thesis, University of Florida, Gainesville, Florida.
Parchure, T.M. (1984). "Erosional Behavior of Deposited Cohesive Sediments," Ph.D.
Dissertation, University of Florida, Gainesville, Florida.
Parker, W. R., and R. Kirby. (1982). "Time Dependent Properties of Cohesive Sediment
Relevant to Sedimentation Management-European Experience," Estuarine Comparisons, V.S.
Kennedy, Editor, Academic Press, New York.
Parker, W. R., and K. Lee. (1979). "The Behavior of Fine Sediment Relevant to the Dispersal
of Pollutants," ICES Workshop on Sediment and Pollutant Interchange in Shallow Seas,
Texel, United Kingdom.
Partheniades, E. (1962). "Study of Erosion and Deposition of Cohesive Soils in Salt Water,"
Ph.D. Dissertation, University of California, Berkeley, California.
Partheniades, E. (1964). "Summary of the Present Knowledge on the Behavior of Fine
Sediments in Estuaries," Technical Note No. 8, Hydrodynamics Lab, M.I.T., Cambridge,
Massachusetts.
Partheniades, E. (1965). "Erosion and Deposition of Cohesive Soils," Journal of the
Hydraulic Division, ASCE, 91(HY1), 105-138.
90
-------
Partheniades, E. (1971). "Erosion and Deposition of Cohesive Soils," River Mechanics, Vol.
II, H. W. Shen, Editor; Fort Collins, Colorado.
Postma, H. (1967). "Sediment Transport and Sedimentation in the Estuarine Environment,"
Estuaries, Publication 83, G. H. Lauff, Editor, American Association for Advancement of
Science, Washington, D.C., 158-179.
Richards, A. F., T. J. Hirst, and J. M. Parks. (1974). "Bulk Density - Water Content
Relationship in Marine Silts and Clays," Journal of Sedimentary Petrology., 44(4), 1004-1009.
Rosillon, R., and C. Volkenborn, (1964). "Sedimentacion de Material Cohesivo en Agua
Salada," Thesis, University of Zulia, Maracaibo, Venezuela.
Sayre, W. W. (1968). "Dispersion of Mass in Open Channel Flow," Hydraulics Papers, No.
3, Colorado State University, Fort Collins, Colorado.
Shen, J., J. D. Boon, and A. Y. Kuo (1999). "A modeling study of a tidal intrusion front and
its impact on larval dispersion in the James River estuary", Virginia. Estuaries, 22, 681-692.
Shields, A. (1936). "Application of Similarity Principles, and Turbulence Research to Bed-
Load Movement," California Institute of Technology, Pasadena (translated from German).
Shresta, P. L., and G. T. Orlob. (1996). "Multiphase distribution of cohesive sediments and
heavy metals in estuarine systems," Journal of Environmental Engineering, 122(8), 730-740.
Simons, D. B., and F. Sentiirk. (1992). Sediment Transport Technology, Water Resources
Publications, Littleton, CO.
Spangler, M. G., and R. L. Handy. (1982). Soil Engineering, Harper and Row, New York.
Teeter, A. M. (2000). "Clay-silt sediment modeling using multiple grain classes; part I:
settling and deposition," In: Coastal and Estuarine Fine Sediment Transport Processes, W.
H. McAnally and A. J. Mehta, eds., Elsevier, Oxford, UK, 157-171.
Terzaghi, K., and R. B. Peck. (1960). Soil Mechanics in Engineering Practice, John Wiley
and Sons, New York.
Tetra Tech. (1999). "Three-dimensional Hydrodynamic and Water Quality Model of Peconic Estuary",
Technical Rep. Prepared for Peconic Estuary Program, Suffolk County, NY, Tetra Tech, Inc., Fairfax, Va.
Thibodeaux, L. J., G. A. Canin and M. Cain. (2002). "Recent Advances in our Understanding
of Sediment-to-Water Contamination Fluxes: The Soluble Release Fraction", Keynote
presentation at the 5th International Symposium on Sediment Quality Assessment of Aquatic
Ecosystems and Public Health, Chicago, IL, October 16-18, 2002.
91
-------
Thomann, R. V. and J. A. Mueller. (1987). Principles of Surface Water Quality Modeling and
Control. Harper and Rowe, Publishers, New York, NY. 644 pp.
Thorn, M. F. C., and J. G. Parsons. (1977). "Properties of Grangemouth Mud," Report No.
EX781, Hydraulics Research Station, Wallingford, United Kingdom.
USEPA (U.S. Environmental Protection Agency) (1998). Ambient Water Quality Criteria
Derivation Methodology Human Health: Technical Support Document. Final Draft. EPA-
822-B-98-005.
USEPA (2005). Contaminated Sediment Remediation Guidance for Hazardous Waste Sites,
OSWER 9355.0-85, EPA-540-R-05-012.
van Olphen, H. (1963). An Introduction to Clay Colloid Chemistry, Interscience Publishers,
New York.
van Rijn, L. C. (1984a). "Sediment transport, Part I: Bed load transport", J. Hyd. Engrg., 110,
1431-1455.
van Rijn, L. C. (1984b). "Sediment transport, Part II: Suspended load transport", J. Hyd.
Engrg., 110, 1613-1641.
WESTON (Weston Solutions, Inc.). 2004a. Modeling Framework Design: Modeling Study of
PCB Contamination in the Housatonic River. Prepared for U.S. Army Corps of Engineers
and U.S. Environmental Protection Agency. DCN GE-042104-ACDP.
WESTON (Weston Solutions, Inc.). 2004b. Model Calibration: Modeling Study of PCB
Contamination in the Housatonic River. Prepared for U.S. Army Corps of Engineers and
U.S. Environmental Protection Agency. DCN GE-122304-ACMG.
WESTON (Weston Solutions, Inc.). 2006. Model Validation: Modeling Study of PCB
Contamination in the Housatonic River. Prepared for U.S. Army Corps of Engineers and
U.S. Environmental Protection Agency. DCN: GE-030706-ADBR.
Whitman, R. G. (1923). "A preliminary experimental confirmation of the two-film theory of
gas absorption", Chem. Metallurg. Eng. 29:146-14.
Wu, W., S. S. Y. Wang, and Y. Jia (2000). "Nonuniform sediment transport in alluvial
rivers", J. Hyd. Res., 38, 427-434.
Yang, C. T. (1973). "Incipient motion and sediment transport," Journal of Hydraulic
Engineering, 99(10), 1679-1704.
Yang, C. T. (1976). "Minimum Unit Stream Power and Fluvial Hydraulics," Journal of the
92
-------
Hydraulics Division, ASCE, 102(HY7), 919-934.
Yang, C. T. (1979). "Unit Stream Power Equations for Total Load," Journal of Hydrology,
40, 123-138.
Yang, C. T. (1984). "Unit Stream Power Equation for Gravel," Journal of Hydraulic
Engineering, ASCE, 110(HY12), 1783-1797.
Yang, C. T., and C. C. S. Song. (1979). "Theory of Minimum Rate of Energy Dissipation,"
Journal of the Hydraulics Division, American Society of Civil Engineers, 105(7).
93
-------
GLOSSARY
Adsorption
Aggradation
Alluvial channel
Angle of repose
Armoring:
Bank erosion
Process by which chemicals (e.g., inorganic phosphorous, metals,
organic chemicals) adhere via chemical bonds to the surface of fine-
grained (cohesive) sediment.
Process by which the bottoms of water bodies are raised due to
deposition of sediment.
Channel completely in alluvium; no bedrock is exposed in channel.
Angle between the horizontal and the maximum slope that a particular
soil assumes through natural processes. For dry noncohesive
sediments, the effect of the height of the slope is negligible, whereas
the angle of repose is meaningless for cohesive sediments due to their
particle-to-particle cohesion.
See bed armor ing.
Removal of soil from the exposed surface of a bank (i.e., bank face) by
high shear stresses during a high flow event, and by slumping.
Baroclinic circulation Vertical circulation in a water body generated by vertical density
gradients.
Baseflow
Bed armoring
Bed forms
Bedload
Bulk density
Groundwater inflow through the banks and bottom to the channel.
This is the portion of a channel's flow hydrograph that occurs between
precipitation induced runoff events.
Natural process by which finer grained bed material is removed from a
surficial channel bed by flow-induced erosion, leaving behind coarser,
more erosion resistant bed material. This layer of coarser bed material
essentially protects or armors the underlying bed material from being
exposed to flow-induced bed shear stresses.
Recognizable relief features on the bed of an alluvial channel, such as a
ripples, dunes and anti-dunes.
Sediment material moving on top of or near a channel bed by rolling,
sliding, and saltating, i.e., jumping.
Mass of sediment and pore water per unit volume of soil or bed
94
-------
Colloids
Critical flow
Degradation
Deposition
Discharge
Embeddedness
Energy grade line
Entrenchment
material.
Particles whose equivalent particle diameter is less than 1 um.
Flow at which the water depth is at critical depth, and when the inertial
and gravitational forces are equal, i.e., Froude number is equal to 1.0.
Process by which the bottoms of water bodies are lowered due to
erosion of bed sediment.
Process of suspended sediment settling and coming to rest on the
bed/bottom of the water body.
Volume of water that passes a given point (e.g., cross section of a
channel) within a given period of time. Typical units are ft3/s or m3/s.
Same as flow rate.
Degree to which fine grained sediments fill the spaces between coarse
sediments (e.g., cobbles, gravels, boulders) on the bed surface.
An inclined line that represents the total energy of a channel flowing
from a higher to a lower elevation. It is located a vertical distance
equal to the velocity head (U2/2g) above the water surface.
Geomorphological process by which a channel erodes downward
between relatively stable banks.
Equilibrium concentration Concentration of suspended noncohesive sediment immediately
above the channel bed (more specifically, at a distance equal to
the thickness of the bedload transport above the bed surface)
under steady flow conditions in an alluvial channel.
Erosion
Wearing away of soil particles on the land surface by detachment and
transport through the action of wind or moving water.
Evapotranspiration Quantity of water transpired by plants and evaporated from land and
water surfaces. Represents the combined processes of evaporation and
transpiration.
Floes
Floodplain
Aggregate of 100s to 1,000s of coagulated fine-grained sediment
particles.
Nearly level land thatl) is susceptible to flooding, 2) forms at the
bottom of a valley, and 3) is adjacent to a natural channel.
95
-------
Fluvial geomorphology
Geological study of the configuration, characteristics, origin
and evolution of channels.
Froude number
Geomorphology
Gradient
Groundwater
Hydrology
Hyporheic zone
Nonpoint source
Peclet number
Plasticity
Point Bar
Point source
Porosity
Reach
Dimensionless number equal to the ratio of inertial to gravitational
forces in channel flow. When the Froude number is less than 1.0, equal
to 1.0, and greater than 1.0, the flow is termed subcritical, critical, and
supercritical, respectively.
Geological study of the configuration, characteristics, origin and
evolution of the land surface.
As used in open channel hydraulics, the bottom slope of a channel.
Water supply in storage, usually in aquifers or crevices in bedrock,
beneath the land surface.
Study of the movement and storage of water in the environment.
Area immediately beneath the wetted perimeter of a channel that
groundwater moves through to recharge the channel in a gaining reach
and that water from the channel moves through to recharge the
groundwater in a losing reach.
Source of contaminant load that enters a water body from multiple
sources over a relatively large area.
Dimensionless number equal to the ratio of advective flux to the
turbulent diffusivity/dispersivity flux.
Property of a soil or rock that allows it to be deformed beyond the point
of recovery without cracking or appreciably changing volume.
An alluvial deposit of sand or gravel that occurs in a channel along the
inside bend of a meander loop, usually a short distance downstream
from the apex of the loop.
Source of contaminant load that is transported through a pipe, outfall or
conveyance channel. The load enters a water body at a specific
location, e.g., end of pipe.
Ratio of the volume of void space (i.e., pores) to the total volume of an
undisturbed soil sample.
Specified length of a channel.
96
-------
Receiving waters
Redispersion
Resuspension
Rheology
Riffle
Runoff
Saltation
Sediment
Sedimentation
Slumping
Spring
Sub critical flow
Supercritical flow
Water bodies into which surface water (i.e., runoff) and/or treated or
untreated waste are discharged.
Erosion or entrainment of a stationary suspension into the water
column.
Erosion of a sediment bed by flow-induced bed shear stresses.
The branch of physics that studies the deformation and flow of matter.
A natural shallow flow area extending across a channel bed in which
the surface of flowing water is broken by waves or ripples. Typically,
riffles alternate with pools along the length of a natural channel.
Portion of precipitation, snow melt, or irrigation that flows over the
land surface, eventually flowing into a water body.
Flow-induced movement of sediment in short jumps or bounces above
a channel bed.
Particles derived from rocks and/or biological materials that is
transported, suspended or deposited by flowing water.
Deposition of sediment.
Detachment of bank material due to combined action of gravitational
force and a pressure that acts from within the bank toward the face of
the channel bank. This pressure, that is caused by water that is stored
in the banks and floodplain soils during a high flow event seeping
towards the riverbank after the high flow event recedes, reduces the
ability of the bank material to stand as a vertical free face, and
sometimes leads to slumping of the bank.
Groundwater flowing or seeping out of the earth where the water table
intersects the land surface.
State of flow where the water depth is greater than the critical depth, in
which the influence of gravitational forces dominate the influences of
inertial forces, and for which the Froude number is less than 1.0.
State of flow where the water depth is less than the critical depth, in
which the influence of inertial forces dominate the influence of
gravitational forces, and for which the Froude number is greater than
1.0.
97
-------
Suspended load
Thalweg
Thixotropy
Turbidity
Void ratio
Watershed
Wetted Perimeter
Amount of sediment that is supported by the upward components of
turbulence in a channel and that stays in suspension for an appreciable
length of time.
Line extending down a channel that follows the lowest elevation of the
bed.
The property of a material that enables it to stiffen in a relatively short
time on standing, but upon agitation or manipulation, to change to a
very soft consistency or to a fluid of high viscosity, the process being
completely reversible.
Measure of the extent to which light passing through water is reduced
due to suspended matter in the water column.
Ratio of the volume of void space to the volume of solid particles in a
given soil mass.
Land area upon which water from direct precipitation, snowmelt, and
other storage collects in a (usually surface) channel and flows downhill
to a common outlet at which the water enters another water body such
as a stream, river, wetland, lake, or the ocean.
Length of wetted contact between water and the channel bottom,
measured in a direction normal to the flow.
98
-------
APPENDIX A
Sediment Properties and Transport
A.I Sediment Properties
Sediments are weathered rock material that are transported, suspended or deposited by
flowing water. All constituents of the parent rock material are usually found in the sediment.
Quartz, because of its greater stability, is by far the most common material found in sediments.
However, numerous other minerals (e.g., shale, carbonate particles, feldspar, igneous and
metamorphic rocks, magnetite) also usually present. Even when material other than quartz
particles are present in sediment, the average particle density of sediment is usually very close to
that of quartz - 2.65 gm/cm3. The specific gravity of sediment is defined as the ratio of the
sediment particle density to the density of water at 4°C (i.e., 1.0 gm/cm3), and thus has an average
value of 2.65.
Sediment diameter is denoted as D, and has dimensions of length. Since sediment
particles are rarely exactly spherical, the definition of diameter requires elaboration. For
sufficiently coarse particles, D is often defined to be the dimension of the smallest square mesh
opening through which the particle will pass. For finer particles, D usually denotes the diameter
of the equivalent sphere with the same fall (or settling) velocity as the actual particle. A sediment
gradation scale (given in Appendix B) has been established to classify sediment in size classes,
ranging from very fine clays to very large boulders. Sediment particles with diameters less than
63 |im are classified as fine-grained sediment, and are cohesive in nature. Sediment particles
with diameters greater than 63 jim are classified as noncohesive sediment.
Cohesive (or fine-grained) sediments are composed of clay and non-clay mineral
components, silt-sized particles, and organic material, including biochemicals (Grim, 1968).
Clays are defined as particles with an equivalent diameter of less than 4 jim, and generally consist
of one or more clay minerals such as kaolinite, bentonite, illite, chlorite, montmorillonite,
vermiculite and halloysite. The non-clay minerals consist of, among others, quartz, calcium
carbonate, feldspar, and mica. The organic matter often present in clay materials can be discrete
particles, adsorbed organic molecules, or constituents inserted between clay layers (Grim, 1968).
Additional possible components of clay materials are water-soluble salts and adsorbed
exchangeable ions and contaminants. Clays possess the properties of plasticity, thixotropy and
adsorption in water (van Olphen, 1963).
For clay-sized particles, surface physicochemical forces exert a distinct controlling
influence on the behavior of the particles due to the large specific area, i.e., ratio of surface area
to volume. In fact, the average surface force on one clay particle is several orders of magnitude
greater than the gravitational force (Partheniades, 1962).
99
-------
The relationships between clay particles and water molecules are governed by
interparticle electrochemical forces. Interparticle forces are both attractive and repulsive. The
attractive forces present are the London-van der Walls and are due to the nearly instantaneous
fluctuation of the dipoles that result from the electrostatic attraction of the nucleus of one atom
for the electron cloud of a neighboring atom (Grimshaw, 1971). These electrical attractive forces
are weak and are only significant when interacting atoms are very close together.
The electrical attractive forces are strong enough to cause structural build-up since they
are additive between pairs of atoms. The magnitude of these forces decreases with increasing
temperature; they are only slightly dependent on the salt concentration (i.e., salinity) of the
medium (van Olphen, 1963). The repulsive forces of clay materials, due to negatively charged
particle forces, increase in an exponential fashion with decreasing particle separation. An
increase in the salinity, however, causes a decrease in the magnitude of these repulsive forces.
The cation exchange capacity (CEC) is an important property of clays by which they
adsorb certain cations and anions in exchange for those already present and retain the new ones in
an exchangeable state. The CEC of different clays varies from 3 to 15 milliequivalents per 100
grams(meq/100 gm) for kaolinite to 100 to 150meq/100 gmforvermiculite. Higher CEC values
indicate greater capacity to adsorb/exchange cations. Some of the predominantly occurring
cations in cohesive sediments are Na, K, Ca, Al, Pb, Cu, Hg, Cr, Cd, and Zn.
In water with very low salinity (less than about 1 psu), individual cohesive sediment
particles are often found in a dispersed state. Small amounts of salts, however, are sufficient to
repress the electrochemical surface repulsive forces among the particles, with the result that the
particles coagulate to form floes. Depending primarily on the CEC of the clay minerals, floes can
form even in freshwater. Each floe can contain thousands or even millions of particles. The
transport properties of floes are affected by the hydrodynamic conditions and by the chemical
composition of the suspending fluid. Most estuaries and some freshwater water bodies contain
abundant quantities of cohesive sediments that usually occur in the coagulated form in various
degrees of flocculation. Therefore, an understanding of the transport properties of cohesive
sediments requires knowledge of the manner in which floes are formed.
Coagulation of suspended cohesive sediments depends on interparticle collision and
cohesion. Cohesion and collision, discussed in detail elsewhere [Einstein and Krone (1962),
Krone (1962), Partheniades (1964), Hunt (1980) and McAnally (1999)] are briefly reviewed here.
There are three principal mechanisms of interparticle collision in suspension, and these influence
the rate at which individual sediment particles coagulate. The first is due to Brownian motion that
results from the thermal motions of the molecules of the suspending water. Generally,
coagulation rates by this mechanism are too slow to be significant unless the suspended sediment
concentration exceeds 5-10 g/L as it sometimes does in fluid mud (a high density, near-bed
layer). Floes formed by this mechanism are weak, with a lace-like structure, and are easily
fractured by shearing, especially in the high shears found near the bed in rivers or estuaries, or are
crushed easily when deposited (Krone, 1962).
The second mechanism is due to internal shearing produced by local velocity gradients in
100
-------
the fluid. Collision will occur if the paths of the particles' centers in the velocity gradient are
displaced by a distance that is less than the sum of their radii. Floes produced by this mechanism
tend to be spherical, and are relatively dense and strong because only those bonds that are strong
enough to resist internal shearing can survive.
The third mechanism, differential sedimentation, results from particles of different sizes
having different settling velocities. A larger particle, due to its higher settling velocity, will
collide with smaller, more slowly settling particles and will have a tendency to pick-up these
particles. This mechanism produces relatively weak floes and contributes to the often observed
rapid clarification of estuarial waters at slack tide.
All three collision mechanisms operate in rivers and estuaries, with internal shearing and
differential sedimentation generally being predominant in the water column, excluding perhaps in
fluid mud where Brownian motion is likely to contribute significantly. The collision efficiency is
less than 100%, so not all collisions result in coagulation.
Cohesion of colliding colloidal particles is caused by the presence of net attractive
electrochemical surface forces on the particles. Particle cohesion is promoted by an increased
concentration of dissolved ions and/or an increased ratio of multivalent to monovalent ions
present in saline waters. The CEC, salinity and ratio of multivalent to monovalent ions all serve
to determine the net interparticle force and, thus, the potential for clay particles to become
cohesive. Kaolinite becomes cohesive at a salinity of 0.6 psu, illite at 1.1 psu and
montmorillonite at 2.4 psu (Ariathurai, 1974). Edzwald et al. (1974) reported that the
cohesiveness of clay particles develops quickly at the given salt concentrations, and that little
increase in coagulation occurs at higher salt concentrations, implying that the particles must have
attained their maximum degree of cohesion.
The rapid development of cohesion and the relatively low salinities at which clays
become cohesive indicate that cohesion is primarily affected by salinity variations near the
landward end of an estuary where salinities are often less than about 3 psu. However, it needs to
be noted that the effects of salinity on flocculation are controversial. Some researchers (e.g., Dr.
Wilbert Lick, University of California, Santa Barbara, CA) contend that salinity does not have
any appreciable effect, whereas the researchers cited here previously (and others as well) have
observed salinity effects on both the settling rates and on the erodibility of cohesive sediments.
The rate and degree of flocculation are important factors that govern the transport of
cohesive sediments. Factors, besides the water chemistry and magnitude of surface forces,
known to govern coagulation and flocculation include: sediment size grading, mineralogical
composition, particle density, organic content, suspension concentration, water temperature,
depth of water through which the floes have settled, and turbulence intensity (represented by the
rate of internal shearing) of the suspending flow (Owen, 1971).
The order of flocculation that characterizes the packing arrangement, density and shear
strength of floes is determined by: 1) sediment type, 2) fluid composition, 3) local shear field, and
101
-------
4) concentration of particles available for flocculation. Krone (1962) found that floe structure is
dependent on salinity for salinities less than about 10 psu. Primary, or 0-order floes, are highly
packed arrangements of clay particles, with each floe consisting of perhaps as many as a million
particles. Typical values of the void ratio (volume of pore water divided by volume of solids)
have been estimated to be on the order of 1.2. This is equivalent to a porosity of 0.55 and is a
more open structure than commonly occurs in noncohesive sediments (Krone, 1963).
Continued flocculation under favorable shear gradients can result in the formation of first
or higher order floes composed of loosely packed arrays of 0-order floes. Each succeeding order
consists of floes of lower density and lower shear strength. A range offices of different shear
strengths and densities are typically formed, with the highest order determined by the prevailing
shearing rate provided that a sufficient number of suspended clay particles are available for
promoting coagulation and flocculation.
A.2 Sediment Bed Properties
As rivers flow from mountains to coastal plains, noncohesive sediment tends to deposit
out, creating an upward concave, long profile of the bed and a pattern of downstream fining of
bed sediment. When the sediment transport capacity in a given reach of a river exceeds the total
sediment load being transported from upstream reaches, the difference between the capacity and
total load is supplied from the bed. This means that the river channel will undergo erosion, i.e.,
degradation. In a river with nonuniform bed material, the finer surficial bed sediment will be
eroded more rapidly than the coarser sediment. By this process, the median diameter of the
surficial bed sediment becomes coarser. If the degradation continues, the finer surficial bed
sediment will eventually be depleted, leaving a surficial layer of coarser sediment. This process
is called armoring and the surficial layer of coarser sediment is called the armor layer.
In response to varying flow conditions, and hence the rate of sediment transport in an
alluvial channel, the bed configuration of the water body will change. Simons and Sentiirk
(1992) defined bed configuration as any irregularity in the bed surface larger than the largest size
sediment particle forming the bed. Bed form is one of several synonyms used in the literature for
bed configuration. Any one who has ever swam in a sandy bottom river, lake, or ocean has no
doubt noticed ripples on the bottom. Ripples are one type of bed form that is created by a certain
range of flow conditions. Other types of bed forms include: plane bed, dunes, washed out dunes,
anti-dunes, and chutes andpools (Simons and Sentiirk, 1992). A plane bed does not have any bed
features. In other words, the bed is essentially flat or smooth. These will normally only be found
in channels with very low flows. With an increase in flow, ripples form in plane bed alluvial
channels. Ripples are small, asymmetric triangular shaped bed forms that are normally less than
5 cm in height and less than 30 cm in length. In general, ripples have long, gentle slopes on their
upstream sides and short, steep slopes on their downstream sides. Dunes are typically larger than
ripples but smaller than bars, and have similar longitudinal profiles as ripples. Dune formation
occurs near the upper end of the subcritical flow regime, and as such, dunes are out of phase with
the water surface; the water surface decreases slightly above the crest of the dune. Washed-out
dunes (also referred to as a transitional bed form) consist of intermixed, low amplitude dunes and
flat areas. These typically occur around the critical flow condition. Antidunes are usually more
102
-------
symmetrical (in their longitudinal profile) than dunes, and form under supercritical flows. Thus,
antidunes are in phase with the water surface elevation and move in the upstream direction.
Chutes and pools usually occur on relatively steep channel slopes, and as such, high velocities
and sediment discharges occur in the chutes.
Noncohesive sediment beds at a given location are characterized by vertical profiles in
grain size distribution and in porosity (or dry bed density). They do not undergo consolidation,
and thus their resistance to erosion does not change over time. A flow-deposited bed of cohesive
sediment floes possesses vertical density and bed shear strength profiles. The average values of
bed density and bed shear strength increase over time and their vertical profiles change with time,
primarily due to consolidation and secondarily due to thixotropy and associated physicochemical
changes affecting inter-particle forces. Consolidation is caused by the gravitational force of
overlying deposited floes (overburden) that crushes, and thereby decreases the order of
flocculation of the underlying sediment. Consolidation changes the erosive behavior of cohesive
sediment beds in two ways: (1) as the shear strength of the bed increases due to consolidation, the
susceptibility of the bed to erosion decreases, and (2) the vertical shear strength profile
determines the depth into the bed that a bed will erode when subjected to excess shear, i.e., an
applied bed shear stress in excess of the bed surface shear strength.
In rivers and other water bodies, sediment beds will sometimes be composed of a mixture
of fine-grained and noncohesive sediments. Researchers have found that if the percentage of
fine-grained sediment in the mixed bed is on the order of 10 to 20%, then that the bed will behave
as cohesive sediment when subjected to high flow velocities.
A.3 Sediment Erosion and Transport
A. 3.1 Noncohesive sediment transport
Incipient motion of a noncohesive sediment particle occurs when the flow-induced forces
are greater than the resistance forces and the particle begins to move across the surface of the
sediment bed. Figure A.I is a diagram of the forces acting on a single, spherical sediment
particle in the surface layer of a sediment bed. For simplicity, all the particles are assumed to
have the same diameter and to be arranged in the orderly fashion seen in this figure. The dashed
brown line in this figure represents the hypothetical bed surface where the mean flow velocity is
zero. The angle between the horizontal black line (on the right side of the figure) and the bed
surface is shown to be 6. The slope of the bed is equal to tan6. The forces shown in this diagram
are the following: Ws = submerged weight of the particle; FD = flow-induced drag force; FL =
flow-induced lift force; and FR = resistance force due to contact between adjacent particles.
103
-------
L
Figure A. 1 Diagram offerees acting on a sediment particle
Summing the forces in the direction perpendicular to the bed at the onset of incipient motion,
i.e., when the particle has not yet started to move, gives:
FL -
= 0
(A.1)
The lift force that acts on the particle is given by:
n
4
(A.2)
where CL = lift coefficient; D = particle diameter; p = water density; and VD = velocity at a
distance D above the bed. The submerged weight of the particle is given by:
Ws = (A - P)g
(A.3)
where g= gravitational acceleration; andps = sediment particle density.
Summing the forces in the direction parallel to the bed at the onset of incipient motion
gives:
FD-
= 0
(A.4)
The drag force that acts on the particle is given by:
104
(A.5)
-------
9 P 9
2
where CD = drag coefficient. Yang (1973) gives the following expression for the resistance
force:
-L (A6)
where i// = friction coefficient.
VD in Eqs. A.2 and A.5 can be determined using a logarithmic velocity distribution:
Vy J
^ = 5.751og^+£
D (A?)
where Vy = velocity at a distance^ above the bed; B = roughness function; and u* = (rb/p)0'5 =
shear velocity, with T& = bed shear stress. In the hydraulically smooth regime, as defined by
the shear velocity Reynolds number (given below in Eq. A. 8), B is given by:
u.D
£=5.5 + 5.751og - u.D
V for 0 < - < 5 (A.8)
V
In the hydraulically rough regime, B is given by:
D o c U*D
B = 8.5 for - > 70
V
Substituting y = D into Eqs. A.7 and A.8 gives VD = Bu*.
The depth-averaged velocity, V, can be obtained by integrating Eq. A.7 over the flow
depth:
V ( d \
=5.75 log -1+5 (A.9)
u* \ D J
Three different approaches have been used to develop criteria for incipient motion. These are
the shear stress, velocity, and probabilistic approaches. The shear stress approach by Shields
(1936) for determining the critical shear stress at the onset of incipient motion, TCS, is probably
the most well known of all the approaches. An example of a probabilistic approach is that
developed by Gessler (1965, 1970). The Shield's shear stress approach, further developed by
van Rijn (1984a), and the velocity approach used by Yang (1973) are summarized below.
The basis of the shear stress approach is that incipient motion of noncohesive
sediment occurs when the bed shear stress exceeds a critical shear stress referred to as the
105
-------
Shield's shear stress, TCS. The latter can be defined by the following nondimensional
relationship:
6L =
CS I
g'D
(A. 10)
where g '= reduced gravitational acceleration, given by:
(A. 11)
and Rd = sediment particle densimetric Reynolds number, given by:
(A. 12)
where v = kinematic viscosity, van Rijn (1984b) gives the following expressions iorf(Rd) on
the right hand side of Eq. A. 10:
0.24(O
~'
0.055
forRT <4
for 4 < R2d13 < 10
for 10 < R j13 < 20
for2Q 150
(A.13)
In his velocity approach, Yang (1973) first assumed that the channel slope was small
enough to neglect the component of the sediment particle's weight in the flow direction in Eq.
A.4, i.e., Wssin6 = 0. Assuming that incipient motion occurs when the two remaining terms
in Eq. A.4 are equal, i.e., FD = FR, he then equated Eqs. A.5 and A.6, substituted Eq. A.9 into
both sides of the resulting equation, and then solved for the dimensionless parameter Vcr/ws,
where Vcr = depth-averaged critical velocity at the onset of incipient motion, and ws = particle
settling velocity (i.e., terminal fall velocity). He also assumed that the drag coefficient was
linearly proportional to the lift coefficient. Yang then used laboratory data sets collected by
several researchers to determine the values of the friction coefficient in Eq. A.6 and the
proportionality coefficient between the drag and lift coefficient to obtain the following
expressions for Vcr/wi.
106
-------
2.5
+ 066
'
log(«.Z)/i}-0.06 ' for
= 2.05 ./)
for 70 < - (A.15)
The friction force exerted along the wetted perimeter of an open channel on the flow is
usually quantified using a resistance formula that contains a roughness coefficient. The
Manning's roughness coefficient is the one most commonly used for open channels with rigid
boundaries. This coefficient is normally used as a calibration parameter in hydraulic models
to achieve optimum agreement between measured and predicted stages (i.e., water surface
elevations) or discharges. Once the model is calibrated, the Manning coefficient is treated as
being temporally constant. For movable boundary problems, i.e., when sediment transport is
involved, the resistance coefficient 1) will change with time due to changes in the movable
bed that result from aggradation and degradation, and 2) can be attributable to two resisting
forces; one force is due to the roughness of the bed surface (this is called grain roughness or
skin friction), and the other force is due to the presence of bed forms in alluvial (i.e., movable
boundary) channels (this is called form roughness or form drag). Einstein and Barbarossa,
(1952) and other researchers have developed procedures for calculating both forms of
movable boundary resistance.
The approach by Yang (1976) for estimating the grain- and form-related flow
resistance in movable boundary open channels does not involve predicting what type of bed
form occurs for a given flow regime (Yang, 1976). The basis for his formulation is the theory
of minimum rate of energy dissipation that states that when a dynamic system (e.g., alluvial
channel) reaches an equilibrium condition, its energy dissipation rate is minimum. This
theory was derived from the second law of thermodynamics. The basic assumption made in
this approach is that the rate of energy dissipation due to sediment transport can be neglected.
For an open channel, the energy dissipation rate per unit weight of water is equal to the unit
stream power VS, where I7is the average flow velocity in the open channel and S is the slope
of the energy grade line. Therefore, the theory of minimum energy dissipation rate requires
that (Yang, 1976):
VS=VmSm (A. 16)
where the subscript m indicates the value of V and S when the unit stream power is
minimized. Yang's approach involves using Eq. A. 14 or A.15 to determine the value of Vcr,
and then using the following sediment transport equation developed by Yang (1973) to
determine the total sediment transport:
107
-------
Cte = 5.435 -0.2861og^ -- 0.4571og
v w,
\ f
-
+ l.799-0.4091og-0.3141og- log -
I v wj ^ wx
where Cte = total sediment concentration being transported by the flow (in ppm by weight), D
= median sieve diameter of the sediment, and VcrS= critical unit stream power required at
incipient motion. The iterative procedure developed by Yang (1976) to determine the value
of the Manning's coefficient in an alluvial open channel uses known values for Q, D, ws, Cts,
and A(d), where the latter is the functional relationship between the cross-sectional area, A, of
the open channel and the flow depth, d. The Yang iterative procedure consists of the
following six steps:
1 . Assume a value for d = flow depth.
2. Solve the 1-D continuity equation (Q=AV) and Eq. A. 17 for Fand S.
3. Compute the unit stream power, i.e., VS.
4. Select another value for d and repeat steps 2 and 3 .
5. Step 4 should be repeated a sufficient number of times to allow for an accurate
determination of the minimum value of VS.
6. Once the minimum value of VS has been determined, the corresponding values
of V, S and d can be calculated using the 1-D continuity equation and Eq. A. 17. The
Manning equation (given below) can then be used to calculate the value of the
Manning's coefficient, n.
V = -R2/\Sl/2 (A.18)
n
where R = hydraulic radius, which is equal to the ratio A/P, where P is the wetted perimeter.
Equation A.18 is the Manning's equation form to use with metric units. Using the theory of
minimum unit stream power, Yang and Song (1979) found good agreement between the
following measured and computed parameters: S, V, d, VS, and n. Parker (1977) also found
good agreement for flows where the sediment transport rate was not too high, thus justifying
Yang's assumption, mentioned previously, under such conditions. However, the method by
Yang (1976) should not be used for critical or supercritical flows, or when the sediment
transport rate is high, since the assumption is invalid under these conditions.
Immediately after onset of incipient motion, the sediment generally moves as bedload.
Bedload transport occurs when noncohesive sediment rolls, slides, or jumps (i.e., saltates)
along the bed. If the flow continues to increase, then some of the sediment moving as
bedload will usually be entrained by vertical turbulent velocity components into the water
column and be transported for extended periods of time in suspension. Thus, it takes more
energy for the flow to transport sediment in suspension than as bedload. The sediment that is
transported in suspension is referred to as suspended load. The total load is the sum of the
bedload and suspended load. Bedload is typically between 10-25 percent of the total load,
108
-------
though for beds with a high fraction of coarse sediment, the percentage will normally be
higher. Many different methods have been developed for calculating the bedload transport
rate in open channels. Some of these methods (along with their references) are listed next.
The specific shear stress approach of van Rijn (1984a) is also described in some detail in the
following.
1. Shear Stress Method: Shields (1936); Chang et al. (1967), and van Rijn (1984a).
2. Energy Slope Method: Meyer-Peter and Muller (1948).
3. Probabilistic Method: Einstein (1950)
Utilizing a shear stress approach, the dimensionless form of the bedload transport rate
is given by van Rijn (1984a) as:
(A. 19)
where O= ~~ , q^ = bedload transport rate (with units of mass per unit time per unit
width), and 6CS is defined in Eq. A. 10. Sediment is transported as bedload in the direction of
the mean flow.
The settling velocity for individual noncohesive sediment particles, ws, is given by van
Rijn (1984b) as the following functions of/), g' and Rd'.
,
forD< 100/jm
18
w
Ul+ 0.0\R2d - \\ for\00jum< D< 1000jam (A.20)
1.1 D> 1000 jum
To predict the noncohesive suspended sediment load in a water body, it is necessary to
determine whether, for a given particle size and flow regime, the sediment is transported as
bedload or as suspended load, van Rijn (1984a) presented the following approach for
distinguishing between bedload and suspended load. When the bed shear velocity, u*, is less
than the critical shear velocity, u*cs, no erosion is assumed to occur, and, therefore, no bedload
transport occurs. Under this latter flow condition, any sediment in suspension whose critical
shear velocity is greater than the bed shear velocity will deposit. When the bed shear velocity
exceeds the critical shear velocity for a given particle size, erosion of that size (and smaller)
sediment from the bed surface is assumed to occur. Therefore, if the following inequality is
true, sediment will be transported as bedload (and not as suspended load):
109
-------
(A21)
Under this inequality condition, any suspended sediment whose critical shear velocity is
greater than the bed shear velocity is assumed to deposit. If the bed shear velocity exceeds
both the critical shear velocity and settling velocity for a given particle size, then that size
sediment (and any smaller) is assumed to be eroded from the bed and transported as
suspended load, and any sediment of that particle size (and smaller) already moving as
bedload is assumed to be subsequently transported in suspension.
The rate of suspended load transport can be calculated as:
(A.22)
ucdz
where qs = suspended load transport rate per unit width of the open channel (with units of
kg/s), u = time-averaged velocity at a distance z above the bed, c = time-averaged
suspended sediment concentration (by volume) at a distance z above the bed, and a =
thickness of the bedload transport zone. Though not described in this report, Lane and
Kalinske (1941), Einstein (1950), Brooks (1963), and Chang etal. (1965), among others,
developed alternative methods to calculate qs.
The two general approaches used to calculate the total noncohesive sediment load in
an open channel consist of: 1) adding the separately estimated bedload and suspended load,
and 2) using a total load function that directly estimates the total amount of bedload and
suspended load transport. Various formulations of the latter are briefly reviewed in this
section. The advantage of using a total load approach is that sediment particles can be
transported in suspension in one reach of an open channel and as bedload in another reach. In
this section, only the unit stream power methods developed by Yang (1973) for estimating the
total load will be presented.
The total sediment load function given by Eq. A. 17 is valid for total sand
concentrations less than about 100 ppm by weight. For higher sediment concentrations, Yang
(1979) presented the following total load equation, again based on the unit stream power
concept:
logCte =5.165-0.1531og-^ 0.2971og
v w
, , (A.23)
+ 1.780 - 0.3601og^ 0.4801og log
v w, } w.
Yang (1984) also presented the following unit stream power based total load equation that is
applicable for gravel sized sediment with median particle sizes between 2 and 10 mm:
110
-------
w D u, (A.24)
logC, = 6.681- 0.633 log^- 4.8161og
g v ws
+ 2.784 -0.305 log - 0.282 log log --M ......
V v wj \ws ws ) For open channels that have bed
sediments in the sand to medium
gravel size range, i.e., between 0.063 to 10 mm, the total load would be the sum, depending
on the value of Cts, of either Eqs. A. 17 and A.24 or Eqs. A. 23 and A.24.
When the sediment transport capacity in a given reach of an open channel exceeds the
total sediment load being transported from upstream reaches, the difference between the
capacity and total load is supplied from the bed. This means that the channel will undergo
erosion, i.e., degradation. In a natural open channel with nonuniform bed material, the finer
surficial bed sediment will be eroded more rapidly than the coarser sediment. By this process,
the median diameter of the surficial bed sediment becomes coarser. If the degradation
continues, the finer surficial bed sediment will eventually be depleted, leaving a layer of
coarser sediment on the bed surface. This process is called armoring, and the surficial layer
of coarser sediment is called the armor layer.
Garcia and Parker (1991) developed the following approach that accounts for the
effect of armoring to estimate the near-bed equilibrium concentration, Ceq, for bed material
that consists of multiple, noncohesive sediment size classes:
c . (A.25)
^
jeq
"7
where Cj-eq = near-bed equilibrium concentration for thej-th sediment size class, A = 1.3*10",
and
(A.26)
&*,
V^FB (A.27)
(A.28)
111
-------
where D50 = median particle size of the noncohesive bed sediments, aff = standard deviation
on the sedimentological phi scale of the bed sediment size distribution, X0 = 0.81, and aff0 =
0.67 (Garcia and Parker, 1991). FH is referred to as a hiding factor.
The near-bed equilibrium concentration is the suspended sediment concentration at a
reference height, zeq, above the bed surface. It represents the maximum suspended sediment
concentration. Some researchers take zeq to be equal to a, i.e., thickness of the bedload
transport zone, in Eq. A.22. Einstein (1950) assumed that zeq = a = 2Db, where Db was
defined as the representative bed sediment grain size, van Rijn (1984b) assumed zeq was
equal to three grain diameters. DuBoy (1879) derived the following expression for the
thickness of the bedload zone:
W(T-TC)
gl-A/>^an^ (A29)
where X = porosity of bed material, and
-------
cause floes to form and grow in size as a result of frequent interparticle collisions and
increased cohesion. The large floes will settle to the lower portion of the water column
because of their high settling velocities. Results from laboratory experiments show that floe
settling velocities can be up to four orders of magnitude larger than the settling velocities of
Upstream Sea
« V »
Rlverborne Sedjnent " Transport to Sea ^
_^^ ^x N
Upward ^=^\ jf^ ~ - H ^safie Wedge
Entralnment r~, ^ v L ^^ x/
x>
Net Upstream /L
] <^| 1 Transport \j-
n
/ /
^y Transport from Sea
^-J
«^ ^* -
Shoalig Mfchg arid Enhanced Aggregation
Turb|d|ty Maximum
Figure A. 3 Schematic representation of transport and sedimentation processes in the
mixing zone of a stratified estuary (after Mehta and Hayter, 1981)
the individual particles (Bellessort 1973). Some of the sediment/floes will deposit; the
remainder will be carried upstream near the bottom until periods close to slack water when
the bed shear stresses decrease sufficiently to permit deposition in the so called turbidity
maximum, after which the sediment starts to undergo self-weight consolidation. The depth to
which the new deposit scours when the currents increase after slack will depend on the bed
shear stresses imposed by the flow and the shear strength of the deposit. Net deposition, i.e.,
sedimentation, will occur when the bed shear during flood, as well as during ebb, is
insufficient to resuspend, i.e., erode, all of the material deposited during preceding slack
periods. Some of the sediment that is resuspended may be re-entrained throughout most of
the length of the mixing zone to levels above the sea water-fresh water interface, and
subsequently transported downstream. At the seaward end, some material may be transported
out of the estuary, a portion of which could ultimately return with the net upstream bottom
current.
In the mixing zone of a typical estuary, the sediment transport rates often are an order
of magnitude greater than the rate of inflow of new sediment derived from upland or oceanic
sources. The estuarial sedimentary regime is characterized by several periodic (or quasi-
periodic) macro-time-scales, the most important of which are the tidal period (diurnal, semi-
diurnal, or mixed) and one-half the lunar month (spring-neap-spring cycle). The tidal period
is the most important since it is the fundamental period that characterizes the basic mode of
sediment transport in an estuary. The lunar month is often significant in determining net
sedimentation rates.
113
-------
From an Eulerian point of view, the superposition of oscillating tidal flows on the
quasi-steady state transport phenomenon depicted in Figure A.3 results in corresponding
oscillations of the suspended sediment concentration with time as shown in Figure A.4. Such
a variation of the suspended load ultimately results from a combination of advective and
dispersive transport, erosion, and deposition. Because of the complexity of the phenomenon,
more than one interpretation is possible as far as any schematic representation of these
phenomena is concerned. One such representation is shown in Figure A.5.
According to this description, cohesive sediments can exist in four different physical
states in an estuary: mobile suspension, stationary suspension, partially consolidated bed, and
settled bed. The last two are formed as a result of consolidation of a stationary suspension.
Stationary here implies little horizontal movement. A stationary suspension, a partially
consolidated bed and a settled bed can erode if the shear stress exceeds a certain critical
value. Erosion of a stationary suspension is referred to as redispersion or mass erosion,
whereas erosion of a partially consolidated bed or a settled bed is termed either resuspension
or surface erosion.
To summarize, the sediment transport regime is controlled by the hydrodynamics, the
chemical composition of the fluid, and the physicochemical properties of the cohesive
sediments. These factors affect the processes of erosion, advection, dispersion, flocculation,
settling, deposition, and consolidation. A brief description of these processes follows that of
cohesive sediment beds.
A flow-deposited bed of cohesive sediment floes possesses a vertical density and bed
shear, i.e., yield, strength profile. The average values of bed density and bed shear strength
increase and their vertical profiles change with time, primarily due to consolidation and
secondarily due to thixotropy and associated physicochemical changes affecting inter-particle
forces. Consolidation is caused by the gravitational force of overlying deposited floes
(overburden) that crushes, and thereby decreases the order of flocculation of the underlying
sediment. Consolidation changes the erosive behavior of cohesive sediment beds in two
ways: (1) as the shear strength of the bed increases due to consolidation, the susceptibility of
the bed to erosion decreases, and (2) the vertical shear strength profile determines the depth
into the bed that a bed will erode when subjected to excess shear, i.e., an applied bed shear
stress in excess of the bed surface shear strength.
114
-------
Mobile Suspension
O)
O
c"
O
"ro
I
o
O
4_i
0)
CD
CO
Pgpostion Reentramment
uspensor
0 2.4m
4 4.9m
7.6m
o Surface
Stationary Suspension
Consolidating Bed
15 16 17 18 19 20 21 22 23 24 1
9/24/68 Time of Measurement
345
9/25/68
Figure A.4 Time and depth variation of suspended sediment concentration in the
Savannah River estuary (after Krone, 1972)
Estuarial sediment beds, typically composed of flow-deposited cohesive sediments,
can be assumed to occur in three different states: stationary suspensions, partially
consolidated beds, and settled (or fully consolidated) beds (see Figure A. 5). Stationary
suspensions are defined by Parker and Lee (1979) as assemblages of high concentrations of
sediment particles that are supported jointly by the water and developing skeletal soil
framework and have no horizontal movement. These suspensions develop whenever the
settling rate of concentrated mobile suspensions exceeds the rate of self-weight consolidation
(Parker and Kirby, 1982). They tend to have a high water content (therefore low bulk
density) and a very low shear strength that must be at least as high as the bed shear that
existed during the deposition period (Mehta et a/., 1982a). Thus, they exhibit a definite non-
Newtonian rheology. Kirby and Parker (1977) found that the stationary suspensions they
investigated had a surface bulk density of approximately 1050 kg/m3 and a layered structure.
Whether redispersion of these suspensions occurs during periods of erosion depends
upon the mechanical shear strength of the floe network. That portion of the floes remaining
on the bed undergoes: 1) self-weight consolidation, and 2) thixotropic effects, defined as the
slow rearrangement of deposited floes attributed to internal energy and unbalanced internal
Figure A. 5 Schematic representation of the physical states of cohesive sediment in an
estuarial mixing zone (after Mehta et a/., 1982a)
115
-------
stresses (Mitchell, 1961), both of which reduce the order of flocculation of sub-surface bed
layers. This implies that the bed becomes stratified with respect to density and shear strength,
with both properties typically increasing monotonically with depth, at least under laboratory
conditions (Mehta etal., 1982a).
Continued consolidation eventually results in the formation of settled mud, defined by
Parker and Lee (1979) as "assemblages of particles predominantly supported by the effective
contact stresses between particles as well as any excess pore water pressure". This portion of
the bed has a lower water content, lower order of flocculation, and higher shear strength. The
settled mud in the Severn Estuary and Inner Bristol Channel, United Kingdom, was found to
posses a bulk density ranging from 1300 to 1700 kg/m3 (Kirby and Parker, 1983). The nature
of the density and shear strength profiles typically found in cohesive sediment beds has been
revealed in laboratory tests by, among others, Richards etal. (1974), Owen (1975), Thorn and
Parsons (1980), Parchure (1980), Bain (1981), Dixit (1982), and Burt and Parker (1984). A
review of this subject is given by Hayter (1983).
Erosion of cohesive sediments occurs whenever the shear stress induced by water
flowing over the sediment bed is great enough to break the electrochemical interparticle
bonds (Partheniades, 1965; Paaswell, 1973). When this happens, erosion takes place by the
removal of individual sediment particles and/or floes. This type of erosion is time dependent
and is defined as surface erosion or resuspension. In contrast, another type of erosion occurs
more or less instantaneously by the removal or entrainment of relatively large pieces of the
bed. This process is referred to as mass erosion or redispersion, and occurs when the flow-
induced shear stresses on the bed exceed the sediment bed bulk strength along some deep-
seated plane.
A number of laboratory investigations were carried out in the 1960's and 1970's in
order to determine the rate of resuspension, £, defined as the mass of sediment eroded per unit
bed surface area per unit time as a function of bed shear in steady, turbulent flows. An
important conclusion from those tests was that the usual soil indices, such as liquid and
plastic limit, do not adequately describe the erosive behavior of these sediments (Mehta,
1981). For example, Partheniades (1962) concluded that the bed shear strength as measured
by standard tests, e.g., the direct-shear test (Terzaghi and Peck, 1960), has no direct
relationship to the sediment's resistance to erosion that is essentially governed by the strength
of the interparticle and inter-floe bonds.
The sediment composition, pore and eroding fluid compositions, and structure of the
flow-deposited bed at the onset of erosion must be determined in order to properly define the
erosion resistance of the bed. Sediment composition is specified by the grain size distribution
of the bed material (i.e., weight fraction of clays, silts), the type of clay minerals present, and
the amount and type of organic matter. The compositions of the pore and eroding fluids are
specified by the temperature, pH, total amounts of salts and type and abundance of ions
present, principally Cl", Na+, Ca2+, and Mg2+. Cementing agents, such as iron oxide, can
significantly increase the resistance of a sediment bed to erosion. Measurement of the
116
-------
electrical conductivity is used to determine the total salt concentration in the pore and eroding
fluids. The effect of the bed structure, specifically the vertical sediment density and shear
strength profiles, on the rate of erosion is discussed by Lambermont and Lebon (1978) and
Mehtaetal. (1982a).
The erosive forces, characterized by the flow-induced instantaneous bed shear stress,
are determined by the flow characteristics and the surface roughness of the fluid-bed
interface. Several different types of relationships between the rate of erosion, e, and the time-
mean value of the flow-induced bed shear stress, T&, have been reported for non-stratified
beds. These include statistical-mechanical models (Partheniades, 1965; Christensen, 1965), a
rate process model (Paaswell, 1973; Kelley and Gularte, 1981), and empirical relationships
(Ariathurai and Arulanandan, 1978).
re
~o "rc
O
0.12 -
0.10 -
0.08 -,
SgO 0.06
^ g 0.04
0) E
4> 0.02 -
Th=0.207 N/m2
0
i I i
i i i i i I i
0 2 4 6 8 10 12 14 16 18 20 22 24
Time (Hours)
Figure A.6 Relative suspended sediment concentration versus time for a stratified bed (after
Mehta and Partheniades, 1979)
117
-------
Figure A.6 shows the measured variation of C, expressed as a relative concentration
by dividing the measured suspended sediment concentration by the initial suspended sediment
concentration before a flow-deposited bed was formed, with time typically found by several
investigators (Partheniades, 1962; Mehta and Partheniades, 1979; Mehta etal., 1982a) in
laboratory resuspension tests with flow-deposited (i.e., stratified) beds under a constant T&. As
observed, dC/dt is high initially, decreases monotonically with time, and appears to approach
zero. The value of rc at the depth of erosion at which dC/dt, and therefore fthat is
proportional to dC/dt, becomes essentially zero has been interpreted to be equal to t\, (Mehta
et a/., 1982a). This interpretation is based on the hypothesis that erosion continues as long as
n > TC. Erosion is arrested at the bed level at which T& - TC = 0. This interpretation, coupled
with measurement ofps(zb\ i.e., the dry bed density profile, and the variation of C with time
resulted in an empirical relationship for the rate of erosion of stratified beds. Utilizing this
above approach, resuspension experiments with deposited beds were performed by Parchure
(1980) in a rotating annular flume and by Dixit (1982) in a recirculating straight flume. The
following empirical relationship between fand n - rc(zb) was derived from these experiments:
s = s0 exp
a-
(A.31)
where £0 and a are empirical resuspension coefficients. This relationship is analogous to the
rate expression that results from a heuristic interpretation of rate process theory for chemical
reactions (Mehta etal., 1982a). Christensen and Das (1973), Paaswell (1973) and Kelley and
Gularte (1981) have used the rate process theory in explaining the erosional behavior of
cohesive sediment beds. By analogy, fis a quantitative measure of the work done by Tb on
the system, i.e., the bed, and e0 and a/Tc(zb) are measures of the system's internal energy, i.e.,
bed resistance to an applied external force.
An important conclusion reached from these experiments was that new deposits
should be treated differently from consolidated beds (Mehta et a/., 1982a). The rate of
surface erosion of new deposits is best evaluated using Eq. A.31, while the erosion rate for
settled beds is best determined using Eq. A.30, in which ovaries linearly with the normalized
excess bed shear stress. The reasons for this differentiation in determining fare twofold: 1)
Typical rc and ps profiles in settled beds vary less significantly with depth than in new
deposits, and may even be nearly invariant. Therefore, the value of \Tb / Tc)\ = &Tb
A *
will be relatively small. For A Tb « 1, the exponential function in Eq. A.31 can be
approximated by OC- (1 + A Tb ) that represents the first two terms in the Taylor series
expansion of expl CX- A Tb ). Thus, for small values of A Tb both expressions for s vary
*
linearly with A Tb and, therefore, the variation of £ with depth in settled beds can be just as
accurately and more simply determined using Eq. A.30; and 2) The laboratory resuspension
tests required to evaluate the coefficients e0 and a for each partially consolidated bed layer
118
-------
cannot be practically or easily performed using vertical sections of an original settled bed
(obtained from cores). A simpler laboratory test has been described by Ariathurai and
Arulanandan (1978) to evaluate the variability of M with depth.
Parchure and Mehta (1985) developed the following relationship for ethat is
applicable for soft, cohesive sediment deposits such as the top, active layer of sediment beds
in estuaries:
(A.32)
where e/= floe erosion rate (gm/m2-s), TS = bed shear strength (Pa), and a = a factor that can
be shown to be inversely proportional to the absolute temperature (Parchure, 1984). ^is
defined to be the erosion rate when the time-averaged bed shear stress is equal to the bed
shear strength, i.e., t\, = TS. Even under this condition, some erosion of particles or floes will
occur due to the stochastic nature of turbulence and therefore in the instantaneous value of T&.
Sea salt is a mixture of salts, with monovalent sodium ions and divalent calcium and
magnesium ions prevalent as natural electrolytes. The sodium adsorption ratio (SAR), defined
as,
Na+
SAR=-
(A.33)
is a measure of the relative abundance of the three mentioned salts (cations). The cation
concentrations in this equation are in milliequivalents per liter (Arulanandan, 1975). Sherard
et al. (1972) have shown that the susceptibility of a cohesive sediment bed to erosion depends
on two factors: 1) the pore fluid composition, as characterized by the SAR; and 2) the salinity
of the eroding fluid. It was found that, as the eroding fluid salinity decreases, soil resistivity
to resuspension decreases. In addition, Kandiah (1974) and Arulanandan et al. (1975) found
that erosion resistance decreased and the rate of resuspension increased with increasing SAR
(and therefore decreasing valency of the salt cations) of the pore fluid.
Once eroded from the bed, cohesive sediment is transported mostly as suspended load,
though the author has observed clumps of cohesive sediments (i.e., mud) rolling along the
bottom of both laboratory flumes and shallow rivers. The latter form of transport cannot be
predicted at present. The transport of both unflocculated and flocculated cohesive sediments
in suspension is the result of three processes: 1) advection - the sediment is assumed to be
transported at the speed of the local mean flow; 2) turbulent diffusion - driven by spatial
suspended sediment concentration gradients, the material is diffused laterally across the width
of the flow channel, vertically over the depth of flow, and longitudinally in the direction of
the transport; and 3) longitudinal dispersion - the suspended sediment is dispersed in the flow
119
-------
direction by spatial velocity gradients (Ippen, 1966).
The principle of conservation of mass with appropriate source and sink terms
describes the advective and dispersive transport of suspended sediment in a turbulent flow
field. This principle, expressed by the advection-dispersion equation, says that the time-rate
of change of mass of sediment in a stationary control volume is equated to the spatial rate of
change of mass due to advection by an external flow field plus the spatial rate of change of
mass due to turbulent diffusion and dispersion processes. The three-dimensional form of the
advection-dispersion transport equation is:
dC dC dC . ,dC d
h u h v \-(w - wsc) =
dt dx dy dz dx
dy
dC dC dC
Ky*~^ + Kyy~^ + Kyz~^~
dx dy dz
dz
dC
dx
dC_
dy
'~ty
dC
dz
(A.34)
ST
where Ky = effective sediment dispersivity tensor, and ST = the net source/sink term that
accounts for source(s) (i.e., addition) of sediment to the water column due to erosion and
other inputs, and sink(s) (i.e., loss) of sediment due to deposition and other removals.
Implicit in this equation is the assumption that suspended material has the same velocity as
the water. Sayre (1968) verified the reasonableness of this assumption for sediment particles
less than about 100 um in diameter. Rolling and saltation of sediment that occur during bed
load transport can result in a significant difference between the water and sediment velocities.
Therefore, the assumption of equal velocity is not applicable to bed load. The net
source/sink term in Eq. A.34 can be expressed as:
S =
dC
dt
dC
dt
S,
(A.35)
where
dC
dt
is the rate of sediment addition (source) due to erosion from the bed, and
dC
dt
is the rate of sediment removal (sink) due to deposition of sediment. SL accounts for removal
(sink) of a certain mass of sediment, for example, by dredging in one area (e.g., a navigational
channel) of a water body, and/or dumping (source) of sediment as dredge spoil in another
location.
The dispersive transport terms in Eq. A.34 include the effects of spatial velocity
variations in bounded shear flows and turbulent diffusion. Thus, the effective sediment
dispersivity tensor in Eq. A.34 must include the effect of all processes whose scale is less than
the grid size of the model, or, in other words, what has been averaged over time and/or space
(Fischer etal, 1979).
120
-------
Turbulent diffusion is defined as "the transport in a given direction at a point in the
flow due to the difference between the true advection in that direction and the time average of
the advection in that direction," and dispersion is defined as "the transport in a given direction
due to the difference between the true advection in that direction and the spatial average of
the advection in that direction" (Holley, 1969). Holley delineates the fact that diffusion and
dispersion are both actually advective transport mechanisms, and that in a given flow field,
the relative importance of one mechanism over the other depends on the magnitude of the
concentration gradient. In Eq. A.34, the effective sediment dispersion coefficients are equal
to the sum of the turbulent diffusion and dispersion coefficients. This approach follows the
analysis of Aris (1956) that showed that the coefficients due to turbulent diffusion and shear
flow (dispersion) were additive. Thus, analytical expressions used for the effective sediment
dispersion tensor should represent both diffusion and dispersion.
Fischer (1966) showed that the dispersion of a given quantity of tracer injected into a
natural stream is divided into two separate phases. The first is the convective period in which
the tracer mixes vertically, laterally, and longitudinally until it is completely distributed
across the stream. The second phase is the diffusive period during which the lateral, and
possibly the vertical (depending on the nature of the tracer), concentration gradient is small,
and the longitudinal concentration profile is highly skewed. Equation A.34 is strictly valid
only in the diffusive period. The criterion for determining whether the dispersing tracer is in
the diffusive period is if it has been in the flow longer than the Lagrangian time scale and has
spread over a distance wider than the Lagrangian length scale (Fischer et a/., 1979). The
latter scale is a measure of the distance a particle travels before it forgets its initial conditions
(i.e., initial position and velocity).
Analytical expressions for the sediment (mass) diffusion coefficients can be obtained
by analogy with the kinematic eddy viscosity. The Reynolds analogy assumes that the
processes of momentum and mass transfer are similar, and that the turbulent diffusion
coefficient and the kinematic eddy viscosity, £v, are linearly proportional. Jobson and Sayre
(1970) verified the Reynolds analogy for sediment particles in the Stokes range (less than
about 100 um in diameter). They found that the "portion of the turbulent mass transfer
coefficient for sediment particles that is directly attributable to tangential components of
turbulent velocity fluctuations: (a) is approximately proportional to the momentum transfer
coefficient and the proportionality constant is less than or equal to 1; and (b) decreases with
increasing particle size". Therefore, the effective sediment mass dispersion coefficients for
cohesive sediments may be justifiably assumed to be equal to those for the water itself.
Fischer et al. (1979) define four primary mechanisms of dispersion in estuaries:
1) gravitational circulation, 2) shear-flow dispersion, 3) bathymetry-induced dispersion and
4) wind-induced circulations. The last three mechanisms occur in freshwater water bodies as
well. Gravitational or baroclinic circulation in estuaries is the flow induced by the density
difference between freshwater at the landward end and sea water at the ocean end. There are
two types of gravitational circulation. Transverse gravitational circulation is depth-averaged
flow that is predominantly seaward in the shallow regions of a cross-section and landward in
the deeper parts. Figure A.7a depicts this net depth-averaged upstream (landward) and
121
-------
downstream (seaward) transport and the resulting transverse flow from the deeper to
shallower parts of the cross-section. The interaction between the cross-sectional bathymetry
and baroclinic flow causes the transverse circulation. Vertical gravitational circulation is
schematically illustrated in Figure A.7b that also shows predominantly seaward flow in the
upper part of the water column and landward flow in the lower part of the water column.
Fischer (1972) said that vertical gravitational circulation is more important than transverse
circulation only in highly stratified estuaries.
The mechanism of shear-flow dispersion is thought to be the dominant mechanism in
long, fairly uniform sections of well-mixed and partially stratified estuaries (Fischer etal.,
1979). Holley et al. (1970) concluded that for wide estuaries, the effect of the vertical
velocity distribution on shear-flow dispersion is dominant over that of the transverse velocity
distribution. The exact opposite situation was found for relatively narrow estuaries.
The joint influence of bathymetry and density differences on dispersion has already
been mentioned in reference to baroclinic circulation. Other examples of bathymetry-induced
dispersion include: intrusion of salinity or sediment into certain parts of a cross-section
caused
122
-------
Net Downstream Transport
h the
Transverse Advect|on and D|ffus|?n
Net Upstream Transport
|i the Deeps
(a) A Transverse Section
Net Downstream Transport
Near the Surface
Vertical Advect |on
and Diffusion
Net Upstream Transport
Near the Bottom
(b) A Vertical Section
Figure A. 10 Internal circulation in a partially stratified estuary (after Fischer et al.,
1979)
123
-------
by channelization of flood and ebb tides in tidal inlets or narrow estuaries (Fischer etal.,
1979), and enhanced dispersion of dissolved substances of concern (e.g., a contaminant) or
intrusion of salinity into tidal flats and side embayments that then serve as storage areas for
these substances, caused by the out-of-phase flow that occurs between the main channel and
such features (Okubo, 1973).
An example of wind-induced circulation is shown in Figure A.8. Here, the steady
onshore wind causes circulation in the wind direction in a shallow bay, where the smaller
water mass per unit surface area results in a higher acceleration and, therefore, quicker
response to the wind-induced surface stresses, and circulation in the opposite direction in the
deeper sections of the channel. Such a circulation can cause significant dispersion (Fischer et
al, 1979).
The settling rate of coagulated sediment particles depends on, in part, the size and
density of the floes, and as such is a function of the processes of coagulation and flocculation
(Owen, 1970). Therefore, the factors that govern these two processes also affect the settling
rate of the resulting floes. The settling velocities of floes can be several orders of magnitude
larger than those of individual clay particles (Bellessort, 1973). For floes from 10 to 1,000
um in size, settling velocities have been found to range from 10"5 to 10"1 m/s (Dyer, 1989).
The following four settling zones have been identified for floes: free settling,
flocculation settling, hindered settling, and negligible settling. In the free settling zone, the
settling velocities are independent of the suspension concentration. In the flocculation zone,
the settling velocities increase with increasing suspension concentration due to increased
interparticle collisions that result in the formation of larger and denser floes. In the hindered
settling zone, the upward transport of interstitial water is inhibited (or hindered) by the high
suspension concentration. This, in turn, results in a decrease in the floe settling velocity with
increasing suspension concentration. At the upper end of the hindered settling zone, the
suspension concentration near the bed is so high that no settling of floes occurs. Hwang
(1989) proposed the following expressions for the floe settling velocity:
wsf forCC3
where wsf= free settling velocity, aw =
velocity scaling coefficient, nw = flocculation settling exponent, bw = hindered settling
coefficient, mw = hindered settling exponent, Cj = concentration between the free settling and
flocculation settling zones, 3 = concentration at the upper limit of the hindered settling zone,
and though not included in Eq. A.36, 2 = concentration between the flocculation and
hindered settling zones (where wsfis maximum). Ranges of values for Ci, 2, and 3 are 100
- 300 mg/L, 1,000 - 15,000 mg/L, and on the order of 75,000 mg/L, respectively (Krone,
1962; Odd and Cooper, 1989).
124
-------
Wind - Induced
Circulation
Figure A.8 Illustration of wind-induced circulation (adapted from Fischer et a/., 1979)
Shrestha and Orlob (1996) developed the following expression for the settling velocity
that accounts for the effect of both the suspension concentration and flow shear:
(A.37)
where OC= 0.11 + 0.039G , and G = J[ ] +[] ; i.e., G is the magnitude of the
vertical shear of the horizontal velocity.
Deposition of floes occurs relatively quickly during slack water. Settling and
deposition also occurs in slowly moving and decelerating flows, as was observed in the
Savannah River Estuary (refer back to Figure A.4) during the second half of flood and ebb
flows (Krone, 1972). Under these flow conditions, only those floes with shear strengths of
sufficient magnitude to withstand the highly disruptive shear stresses in the near bed region
will actually deposit and adhere to the bed. Thus, deposition is governed by the bed shear
stresses, turbulence structure above the bed, settling velocity, type of sediment, depth of flow,
suspension concentration, and ionic constitution of the suspending fluid (Mehta and
Partheniades, 1973). Specifically, deposition has been defined to occur when n is not high
enough to resuspend sediment material that settles onto and bonds with the bed surface. This
process, therefore, involves two other processes, settling and bonding.
Laboratory studies on the deposit!onal behavior of cohesive sediment in steady
turbulent flows have been conducted by, among others, Krone (1962), Rosillon and
Volkenborn (1964), Partheniades (1965), Lee (1974), Mehta and Partheniades (1975), Mehta
et al. (1982b), Mehta and Lott (1987), Shrestha and Orlob (1996), and Teeter (2000).
125
-------
The most commonly used expression for the sediment mass deposition rate, given
initially by Einstein and Krone (1962), is:
d£_
dt ~ d
(A.38)
where rcj = critical shear stress for deposition, above which no deposition occurs. The value
of Tcd was found to be equal to 0.06 Pa for San Francisco Bay mud with C < 300 mg/L
(Krone, 1963), and values from 0.02 to 0.2 Pa have been reported in the literature. Mehta and
Lott (1987) found Eq. A.38 to agree reasonably well with laboratory data for suspended
sediment concentrations up to approximately 1,000 mg/L.
A cohesive sediment bed is formed when deposited sediment particles and/or floes
comprising a stationary suspension begin to interact and form a soil that transmits an effective
stress by virtue of particle-to-particle contacts. The self-weight of the particles, as well as
deposition of additional material, brings the particles closer together by expulsion of pore
water between the particles. A soil is formed when the water content of the sediment-water
suspension decreases to the fluid limit. Unfortunately, there is not a unique water content
value for cohesive soils at which the suspension changes into a soil (Been and Sills, 1981).
During the transition from suspension to soil, an extremely compressible soil
framework or skeleton develops (Been and Sills, 1981). The strains involved in this first
stage of consolidation are relatively large and can continue for several days or even months.
The straining and upward expulsion of pore water gradually decreases as the soil skeleton
continues to develop. Eventually, this skeleton reaches a state of equilibrium with the normal
stress component of the overlying sediment (Parker and Lee, 1979).
During the early stages of consolidation, the self-weight of the soil mass near the bed
surface is balanced by the seepage force induced by the upward flow of pore water from the
underlying sediment. As the soil continues to undergo self-weight consolidation and the
upward flux of pore water lessens, the self-weight of this near surface soil gradually turns into
an effective stress. This surface stress and the stress throughout the soil will first crush the
soil floe structure and then the floes themselves. Primary consolidation is defined to end
when the excessive pore water pressure has completely dissipated (Spangler and Handy,
1982). Secondary consolidation, that can continue for many weeks or months, is the result of
plastic deformation of the soil under its overburden.
126
-------
The shear strength of clays is due to the frictional resistance and interlocking between
particles (physical component), and interparticle forces (physicochemical component) (Karcz
and Shanmugam, 1974; Parchure, 1980). Consolidation results in increasing bed density and
shear strength (Hanzawa and Kishida, 1981). Figure A.9 shows the increase in the shear
strength profile with consolidation time and bed depth for flow-deposited kaolinite beds in tap
water.
0
0.4
0.8 -
^
N 1.2
1.6 -
2.0 -
24
Symbol J (hrs
0
Figure A.9
0.2
0.4
0.6
Tc(Z)(N/m2)
Bed shear strength versus distance below the initial bed surface for
various consolidation oeriods (after Dixit 1982")
127
-------
APPENDIX B
Sediment Gradation Scale (adapted from ASCE, 1975)
Sediment Class Name
Very large boulders
Large boulders
Medium boulders
Small boulders
Large cobbles
Small cobbles
Very coarse gravel
Coarse gravel
Medium gravel
Fine gravel
Very fine gravel
Very coarse sand
Coarse sand
Medium sand
Fine sand
Very fine sand
Coarse silt
Medium silt
Fine silt
Very fine silt
Coarse clay
Medium clay
Fine clay
Very fine clay
Size Range
(mm)
4096 - 2048
2048 - 1024
1024-512
512-256
256- 128
128 - 64
64-32
32- 16
16-8
8-4
4-2
2- 1
1 -0.5
0.5-0.25
0.25-0.125
0.125-0.063
0.063 -0.031
0.031 -0.016
0.016-0.008
0.008 - 0.004
0.004 - 0.002
0.002-0.001
0.001 -0.0005
0.0005 - 0.00024
Size Range
(urn)
2000 - 1000
1000-500
500-250
250- 125
125 - 63
63 -31
31 - 16
16-8
8-4
4-2
2- 1
1 -0.5
0.5-0.24
128
------- |