Contaminated Ground Water and Sediment - Chapter 2 pps

2

The Role of Modeling in
Managing Contaminated
Sediments

prepared by Danny D. Reible

with contributions by
Sam Bentley, Mimi B. Dannel, Joseph V. DePinto,
James A. Dyer, Kevin J. Farley, Marcelo H. Garcia,
David Glaser, John M. Hamrick, Richard H. Jensen,
Wilbert J. Lick, Robert A. Pastorok, Richard F. Schwer,
C. Kirk Ziegler

CONTENTS

2.1Introduction
2.1.1SigniÞcance and Objectives
2.1.2Status of Contaminated Sediment Management
2.1.3Contaminated Sediment Modeling Applications
2.1.3.1Conceptual Site Model Development and Testing
2.1.3.2Baseline Risk Assessment
2.1.3.3Evaluation of Total Maximum Daily Loads
2.1.3.4Comparative Evaluation of Remedial Management Plans
2.2State of Knowledge and Practice
2.2.1Relation between Sediment and Common Contaminants
2.2.2Sediment Transport Model Components
2.2.2.1Sediment Erosion and Deposition Processes
2.2.2.2Sediment Transport Model Minimum Requirements

2.2.3Contaminant Fate and Transport Model Components
2.2.3.1Contaminant Fate and Transport Processes
in Unstable Sediments
2.2.3.2Contaminant Fate and Transport Processes
in Stable Sediments
2.2.3.3Contaminant Transference via Food Webs
2.2.3.4Human and Ecological Risk Evaluation
2.2.4Model Calibration and Uncertainty
2.2.4.1Sources of Model Uncertainty

L1667_book.fm Page 61 Tuesday, October 21, 2003 8:33 AM
©2004 CRC Press LLC

2.2.4.2Techniques for Calibrating a Model
and Evaluating Uncertainty
2.2.4.3Measures of Model Acceptability
2.3Challenges and Emerging Issues
2.3.1Cohesive Sediment Erosion and Transport
2.3.2Contaminant Release and Availability
2.3.3Advective Processes in the Hyperheic Zone
2.3.4Bioturbation as a Sediment and Contaminant
Transport Mechanism
2.3.5Contaminant Bioaccumulation and Effects in Benthic
and Higher Organisms
2.4Summary of Research and Data Needs
2.4.1Sediment Transport Process Modeling
2.4.2Contaminant Process Modeling
2.4.3Biological Process Modeling
2.4.4Metals Release and Availability
2.4.5Hydrophobic Organic Contaminant Release and Availability

Acknowledgments
References

2.1 INTRODUCTION
2.1.1 S

IGNIFICANCE



AND

O

BJECTIVES

Contaminated sediment management poses some of the most difÞcult site remediation
issues today. Contaminated sediments typically reside in spatially variable and dynamic
systems subject to seasonal ßow variations and episodic storm events. The volume of
sediments that must be managed often exceeds 1 million yd

3

, dwarÞng many contam-
inated soil sites. These sediments are also associated with equally daunting volumes
of water, and efforts to remove the contamination typically entrains even more water.
The National Sediment Quality Survey (United States Environmental Protection
Agency [USEPA], 1998) classiÞed 26% of 21,000 freshwater and estuarine sediment
sampling stations in the U.S. as Tier 1 (i.e., adverse effects on aquatic life or human
health are probable) and 49% as Tier 2 (i.e., adverse effects on aquatic life or human

health are possible but expected infrequently). The realization of these potential risks
depends in part on the degree of conservatism built into the toxicological assumptions
and in part on the processes controlling both contaminant release from the sediments
and the transfer to benthic, aquatic, and land-based organisms. Observations of
impairments in ecological or human health can indicate potential pollution problems;
however, linking these adverse effects to contaminated sediments requires an under-
standing of the processes leading to exposure and uptake. In addition, the selection
of cost-effective and environmentally protective remedial alternatives is dependent
upon the ability to predict the risks during implementation and into the future.
Conceptual models can establish hypotheses as to the links between current or
potential exposure to contaminated sediments and the risk to human and ecological
health. Testing these hypotheses, however, generally requires translation of the

L1667_book.fm Page 62 Tuesday, October 21, 2003 8:33 AM
©2004 CRC Press LLC

conceptual model into quantitative form. Quantitative models can be used to answer
such questions as the following:
•Is the observed exposure and risk consistent with identiÞed sources of
that risk?
•What are the most important source areas and exposure processes and
pathways?
•What data can be collected to characterize these important processes and
pathways most accurately?
•What interventions can be most effective in responding to these processes
and pathways?
•What are the future exposure and risks if the sediments are managed by
•Natural processes?
•

In situ

containment or treatment technologies?
•Removal and

ex situ

treatment or disposal?
•How does uncertainty in processes and pathways and the parameters that
characterize them translate into uncertainty in current and/or potential
future risks?
In making decisions about contaminated sites, the use of quantitative modeling
to answer these questions is a critical link between observing current exposure and
risk (i.e., deÞning baseline risk) and comparing and selecting management
approaches that effectively minimize or control that risk. This chapter summarizes
applications of quantitative prognostic models of contaminant processes in sedi-
ments, assesses the state-of-the-art of these models with respect to accuracy and
adequacy, and identiÞes research that can contribute to improvements in model
development and their use in resolving sediment management challenges.

2.1.2 S

TATUS



OF

C

ONTAMINATED

S

EDIMENT

M

ANAGEMENT

The goals for contaminated sediment management were identiÞed by the USEPA
(1998) to include the following:
• Prevent further contamination of sediments that may cause unacceptable
human health or ecological risks.
• When practical, clean up existing sediment contamination that adversely
affects the nation’s waterbodies or their uses or that causes other signiÞ-
cant effects on human health or the environment.
• Ensure that sediment dredging and the disposal of dredged material con-
tinue to be managed in an environmentally sound manner.
•Develop and consistently apply methodologies for analyzing contami-
nated sediments.
Sediment modeling can assist in achieving these goals by helping to quantify
the importance of potential sources of sediment contaminants and by predicting
sediment fate and transport processes that inßuence exposure and risk. SpeciÞcally,

L1667_book.fm Page 63 Tuesday, October 21, 2003 8:33 AM
©2004 CRC Press LLC

models can be used to evaluate the effect of extreme events, the likelihood that
existing sources can lead to sediment recontamination, and the contribution of

sediments to the pollutant burden faced by the ecosystem. In addition, models can
be used to compare the effectiveness of various sediment management approaches.
Contaminated sediment sites are often poorly controlled, dynamic systems con-
taining large volumes of moderately contaminated material. An analysis of Super-
fund Records of Decisions from 1982 to 1997 (USEPA, 1999) showed that the
average contaminated soil site considered for

ex situ

treatment contained 38,000 yd

3

of contaminated material; for

in situ

treatment, the total was approximately 105,000
yd

3

of contaminated material. Contaminated sediment sites, however, often contain
in excess of 1,000,000 yd

3

of contaminated material and generally are not directly
accessible. Soils can be removed in a relatively dry state for further processing,
whereas sediments are removed as slurries with a high proportion of water that must

be treated. The assessment and control of contaminant releases when removing
submerged sediments is also much more difÞcult than when removing soil for

ex
situ

treatment. This difÞculty is the result of limited control over the aquatic envi-
ronment as well as the chemical and physical changes the sediment undergoes during
removal (e.g., anaerobic to aerobic and wet to dry).
Many of the potential technologies for contaminated sediment management were
initially developed to manage contaminated soils. Unfortunately, many of these
technologies are either difÞcult to apply or impose potentially unacceptable risks
when applied to contaminated sediments. Identifying, comparing, and selecting
remedial options for contaminated sediment is also complicated by the multiple
technologies often involved. For example,

ex situ

treatment or sediment disposal
typically introduces a complete train of technologies, including removing material
by dredging, temporarily storing or pretreating to reduce water content or volume,
treating or disposing of Þnal dredged material, and managing any residually con-
taminated materials. Large contaminated sediment sites generally require applying
different options at different areas on-site, each containing multiple technologies.
Therefore, identifying sediment management and remediation options must recog-
nize the entire train of technologies that constitute each option so that a fair evalu-
ation and comparison of these options can be accomplished.
Risk reduction has been generally accepted as the metric by which various
options are judged and selected. Use of this metric, however, places a premium on
the quantitative modeling required to link the sediments to exposure and risk.

Evaluating management or remedial options requires deÞning remedial action goals
and objectives and developing a valid conceptual model of the sediment system to
be remediated. At all but the most trivial sites, a sophisticated quantitative model
can be helpful or necessary to develop and test the conceptual model and evaluate
the effectiveness of various management options in meeting the remedial action
goals and objectives. Large, complicated sites posing substantial risks and potentially
large cleanup costs generally require the development of an extensive database and
sophisticated prognostic models in order to compare management options and eval-
uate potential risk reduction adequately.
There is no generally accepted option for managing contaminated sediments at
all sites. Removal approaches typically have been focused on a portion of the

L1667_book.fm Page 64 Tuesday, October 21, 2003 8:33 AM
©2004 CRC Press LLC

contaminated sites where contaminant levels are the highest and the potential success
in terms of reduction of risk is limited to the extent that the “hot spot” contributes
to the overall risk of the site. Due to incomplete removal, resuspension during
removal, and the resulting residual contamination, the effectiveness of removal
options is closely linked to the natural setting and processes inßuencing the sediment
contaminants. Nonremoval options are also closely connected to natural fate and
transport processes. Linking these processes to exposure and risk is dependent on
modeling. Most sediment contaminants are relatively refractory at least below the
upper few inches of sediment and, therefore, are unlikely to exhibit substantial
attenuation by fate processes such as microbial degradation except over very long
time scales. Thus, quantitatively predicting exposure and risks far into the future is
often required. In addition, most sediment contaminants are strongly associated with
the solid phase, and sediment transport can control contaminant fate. Therefore,
modeling natural exposure and fate processes entails describing river hydraulics and
sediment migration as well as contaminant transport. The sections below discuss

some of the speciÞc applications of contaminated sediment modeling.

2.1.3 C

ONTAMINATED

S

EDIMENT

M

ODELING

A

PPLICATIONS

2.1.3.1 Conceptual Site Model Development and Testing

A conceptual model is necessary to deÞne the fundamental relationship between
contaminant levels in the sediment and levels of exposure and risk to human health
and the environment. A conceptual model identiÞes any ongoing sources that can
lead to sediment recontamination, mechanisms that can move contaminants from
sediments to receptor organisms, and fate processes that can reduce available
contaminant concentrations or their effects on receptor organisms. Some of the
processes that deÞne exposure within a conceptual model are depicted in Figure
2.1. A valid conceptual model is necessary to identify which remedial options
have the potential to address the most important contaminant processes effectively.
The conceptual model is the foundation on which site management actions are

identiÞed and implemented.
Although a conceptual model need not be quantitative, comparison with a quan-
titative model can help identify and test a conceptual model. For example, the
question as to whether all sources have been identiÞed can be answered by the ability
to quantitatively predict the extent of contamination in water and biota based on the
recognized sources. An inability to reproduce the observed patterns of contamination
can suggest that additional sources exist. Similarly, an inability to predict contami-
nant ßux from sediment to water based on presumed mechanisms and processes can
indicate that additional processes are operative.

2.1.3.2 Baseline Risk Assessment

Within a risk-based, decision-making framework, the existing or baseline risks deÞne
the signiÞcance of a contaminated sediment problem and the need for management
or remediation. Although Þeld measurements can identify contaminant levels in the
environment and body burdens in potentially affected species, establishing a cause-

L1667_book.fm Page 65 Tuesday, October 21, 2003 8:33 AM
©2004 CRC Press LLC

and-effect relationship between the two normally requires using a numerical model.
Generally, it is not possible to relate sediment concentrations to risks without iden-
tifying and quantifying fate and transport processes that lead to exposure, assimila-
tion, and effect.
Quantitative modeling can be used to assess important pathways and processes
and compare various contaminated sediment management approaches. The processes
and their relative importance vary widely in different sediment environments as
illustrated in Table 2.1. Current exposure and risk and the predicted attenuation of
contaminants as a result of these natural processes serves as a baseline with which
to compare active management approaches. Constitutive relationships and measure-

ments of the parameters within those relationships are critical to the quantitative
descriptions of these processes.

2.1.3.3 Evaluation of Total Maximum Daily Loads

Contaminated sediments fall under the purview of several regulatory programs,
including the total maximum daily load (TMDL) program established by 303(d)
of the Clean Water Act. A TMDL is a calculation of the maximum amount of a
pollutant that a waterbody can receive and still meet water quality standards. This
TMDL must be allocated to the various sources of the pollutant. A TMDL must

FIGURE 2.1

Potential water column and sediment processes inßuencing contaminant trans-
port and fate in a river segment.
Deep Sediments
Biologically
Active Zone
Overlying Water
Ground Water Exchange
Fate
Processes
Burial/Exposure
Source
Area
Vaporization/Deposition
Resuspension
Erosion
Particles Dissolved Colloidal
Advection

Dispersion
Advection
Dispersion
Nonparticle
Exchange
Processes

L1667_book.fm Page 66 Tuesday, October 21, 2003 8:33 AM
©2004 CRC Press LLC

also include a margin of safety to account for uncertainty and must consider
seasonal variation.
Of the approximately 20,000 waterbodies currently slated for TMDL develop-
ment, at least 300 are expected to be impaired speciÞcally by sediments. Many
additional bodies of water can be more impaired as a result of the contribution of
pollutants from contaminated sediments or through the effect of sediment processes
(e.g., sediment oxygen demand) on overlying water quality.
Predictive modeling is a key tool in establishing TMDLs because the model
provides the needed link between pollutant loadings to a waterbody and the effect
of these loadings on attaining water quality standards. Modeling is especially impor-
tant in situations such as contaminated sediment sites where it is difÞcult to directly

TABLE 2.1
Sediment Processes and Their Relationship to Various
Sediment Environments

Environment Environmental Characteristics Key Fate and Transport Processes

Lacustrine Low energy environment
Generally depositional environment

Ground water interaction decreasing
away from shore
Organic matter decreasing with distance
from shore
Often Þne-grained sediment
Sediment deposition
Water-side mass transfer limitations
Ground water advection in near-shore
area
Bioturbation (especially in near-shore
area)
Diffusion in quiescent settings
Metal sequestration
Aerobic and anaerobic
biotransformation of contaminants of
concern (COCs)
Biotransformation of organic matter
Riverine Low to high energy environment
Depositional or erosional environment
Potential for signiÞcant ground water
interaction
Variable sediment characteristics (Þne to
coarse grained)
Local and generalized ground water
advection
Sediment deposition and resuspension
Aerobic biotransformation processes in
surÞcial sediments (potentially
anaerobic at depth)
Bioturbation

Estuarine Generally low-energy environment
Generally depositional environment
Generally Þne-grained sediment
Bioturbation
Sediment deposition
Water-side mass transfer limitations
Aerobic and anaerobic
biotransformation of COCs
Biotransformation of organic matter
Uptake and biotransformation in plants
Coastal marine Relatively high-energy environment,
decreasing with depth and distance
from shore
Often coarse sediments
Bioturbation
Sediment erosion and deposition
Localized advection processes

L1667_book.fm Page 67 Tuesday, October 21, 2003 8:33 AM
©2004 CRC Press LLC

measure how much of the pollutant loading can be attributed to speciÞc sources.
However, most of the tools used to establish TMDLs, including the USEPA’s Better
Assessment Science Integrating Point and Nonpoint Sources (BASINS) system, do
not include state-of-the-art sediment modeling capabilities.
TMDLs can also be designed to eliminate sediment toxicity. Although sediment
toxicity can be determined based on biological assays, the allocation of point or
nonpoint contaminant sources can be related to sediment concentrations through
only a numerical model. Numerical models must be able to deÞne contaminant
access and availability and the assimilative capacity of the speciÞc receptor at risk.

In addition, numerical modeling is required to determine the ability of management
approaches (including natural attenuation and recovery) to eliminate sediment tox-
icity. Numerical models can also be used as a foundation for allocating allowable
pollutant discharges to point and nonpoint sources.

2.1.3.4 Comparative Evaluation of Remedial
Management Plans

The primary goal of contaminated sediment management is to protect resources
at risk such as human or ecological health, commercial or recreational Þshing
stocks, or a particular endangered species. Ideally, management and remedial
options that best protect affected resources or lead to resource recovery should be
selected. Because measurements can only hope to indicate the current state, quan-
titative models must be used to allow comparison of the future effect of various
scenarios. SigniÞcant questions remain regarding how best to use the forward
projections in time. In principle, models provide concentrations or rates of expo-
sure as a function of time and place. This information provides a basis on which
risk and effects can be estimated. However, both the assessment of future concen-
trations and the rates of exposure, along with the assessment of future risks and
effects, are subject to great uncertainty.
An alternative to evaluating and comparing management options is to employ
contaminant mass ßows as a surrogate measure of exposure and, ultimately, risk.
That is, a technology can generally be assumed to pose less exposure and risk if
it leaves less residual contamination and loses fewer contaminants to the air and
water than does an alternative technology. The evaluation of contaminant mass
ßows for each management option can be most useful in the comparative evaluation
by providing a systematic screening tool. A comparative analysis of mass ßows
can also help identify those components of an overall management strategy that
largely control the overall exposure or risk and, therefore, should receive the most
resources and effort for detailed evaluation. In this manner, screening sediment

management alternatives can ensure that all needed components of any given
option are included for subsequent evaluation. Even a comparative analysis of
mass ßow generally requires sophisticated modeling of contaminated sediment
fate and transport processes.
Although contaminant mass ßow can be useful, exposure and risk to human
and ecological health ultimately drives the need for remediation and the success
or failure of any management or remedial option. In particular, contaminant mass

L1667_book.fm Page 68 Tuesday, October 21, 2003 8:33 AM
©2004 CRC Press LLC

ßow is not very helpful in balancing short-term acute risks with long-term risks.
Contaminated sediment removal, for example, tends to lead to increased risks in
the short term in exchange for potential reduced long-term risk. It is important to
note that

in situ

sediment management approaches are always subject to potential
failure because of future events. Modeling can provide a basis for identifying the
magnitude of the potential future exposures and risk that the various management
options can pose.

2.2 STATE OF KNOWLEDGE AND PRACTICE

A variety of models have been used to predict sediment and contaminant behavior,
fate, and effects in ecosystems. Thoms et al. (1995) summarized the capabilities of
some of these models and showed the variety of transport and fate pathways that
they describe. This section summarizes some of the key processes that characterize
the fate and transport pathways.

2.2.1 R

ELATION



BETWEEN

S

EDIMENT



AND

C

OMMON


C

ONTAMINANTS

Most priority pollutants and other contaminants of concern (COCs) are strongly
associated with solids and therefore tend to accumulate in sediments. Contaminants
that do not strongly associate with solids (e.g., polar organic compounds, soluble
metals) rarely represent sediment contaminants in that they are efÞciently released

to the overlying water. Historical industrial and municipal efßuents and runoff are
often responsible for sediment contamination because only those contaminants that
tend to partition strongly to the solid phase remain in their historical location. Soluble
and volatile contaminants tend to be transported away by water movement or are
released to the air via evaporative processes. Similarly, when more soluble and
volatile contaminants initially contaminate sediments, their mobility ultimately
allows them to migrate into more mobile phases, effectively eliminating them from
the sediments. There are exceptions to these general rules when contaminants are
continuing to be introduced to the sediments from ground waters or from active
sources, but the bulk of sediment contaminants are those that strongly associate with
the solid phase. The extent to which sediments are associated with the solid phase
is deÞned by the effective sediment–water partition coefÞcient,

K

sw

. This coefÞcient
is deÞned as the ratio of the concentration of contaminant on solids,

W

s

(milligrams
per kilogram [mg/kg]), to the water concentration,

C

w

(milligrams per liter [mg/l]).
Given a density of solids,

r

s

(kg/l), the fraction of contaminant associated with the
solid phase,

f

s,

is given by the following equation:
(2.1)
For large

K

sw

values, as would be expected for most sediment contaminants, the
fraction associated with solids approaches unity unless the density of solids is small.
f
K
K
s
ssw

ssw
=
+
r
r1

L1667_book.fm Page 69 Tuesday, October 21, 2003 8:33 AM
©2004 CRC Press LLC

For hydrophobic organic compounds, the commonly employed linear, reversible
model is that the sediment–water partition coefÞcient is given by (

K

oc

)(

f

oc

) where

K

oc

is the organic carbon–based partition coefÞcient and a measure of the compound
hydrophobicity, and f

oc

is the fraction organic carbon that serves as the solution
phase in the solid. A moderately hydrophobic sediment contaminant might be pyrene,
a polycyclic aromatic hydrocarbon (PAH) with a

K

oc



of approximately 10

5

l/kg. A
typical sediment organic carbon is of the order of 1%, suggesting a

K

sw



of about
1,000 by this model. Thus, in the sediment bed, where

r

s

is large (of the order of 1
kg/l), a

K

sw

of 1,000 l/kg suggests that about 99.9% of the contaminant would be
associated with the solid phase. In the overlying water, however, if the suspended
sediment concentration is low (perhaps 100 mg/l or less), the majority of the con-
taminant can be dissolved and not associated with the solid phase. Similarly, changes
in redox conditions for resuspended sediment can cause metal releases that might
normally be associated with sediments in a stable bed.
The strong association of contaminants with the solid phase in a sediment bed,
however, suggests that contaminant fate often is deÞned largely by sediment mobility
and fate. This section examines some of the most important sediment contaminants
and their physical and chemical characteristics that relate to fate and mobility in the
environment. These contaminants are described below and include heavy metals,
oxygen-demanding contaminants in sediments, undifferentiated oil and grease, pes-
ticides, polychlorinated biphenyls (PCBs), and PAHs.
• Heavy Metals
The toxic elements include antimony, arsenic, beryllium, cadmium,
copper, lead, mercury, nickel, silver, thallium, and zinc. These pollut-
ants are important in that they are nonbiodegradable, toxic in solution,
and subject to biomagniÞcation. The chemistry of many of these com-
pounds is complex in sediments. A portion is generally chemically Þxed
and largely unavailable to Þsh and higher organisms without chemical

changes in the sediment. Often a portion is ion exchangeable that can
become available simply with the addition of a more strongly held con-
taminant. Finally, a portion is soluble, mobile, and directly available for
uptake by organisms.



Myers et al. (1996) indicate that the partition co-
efÞcient between the leachable fraction and the water is typically be-
tween 3 and 10, resulting in a leachable fraction of metals that is
typically less than 10% and sometimes much less.
The equilibrium state for metals and other elemental species depends on
the chemical state of the water and sediment, particularly the pH and ox-
idation–reduction conditions. The ratio of sediment loading to equilibri-
um water concentration is often very large for metals, but only a small
fraction of the metals are typically available. As a result of this variability,
a site-speciÞc measurement of the sediment–water partition coefÞcient is
preferred over any predictive approach. The ultimate or potential avail-
ability of many metals appears to be controlled by the presence of acid
volatile sulÞdes (AVS) in the sediments in which they reside. The term
“AVS” refers to the manner in which sulÞde presence is measured. If the

L1667_book.fm Page 70 Tuesday, October 21, 2003 8:33 AM
©2004 CRC Press LLC

ratio of AVS is greater than the simultaneously extractable metal (SEM)
content, most of the key metals of concern are bound to the sediment as
sulÞdes. In this form, the metals appear to be largely unavailable to recep-
tors from microorganisms up through humans. Because many marine
sediments contain signiÞcant quantities of AVS, much of the metals may

not be available to provide exposure and risk to organisms. Although high
AVS/SEM ratios indicate that certain metals are bound in unavailable
forms, low AVS/SEM ratios do not necessarily indicate available metals.
Historically, metals-processing industries as well as urban and rural
runoff provided the most signiÞcant sources of these elements. In addi-
tion, lead was widely distributed in the environment as a result of the
use of tetraethyl lead in gasoline to control premature ignition (i.e.,
knocking). Although ongoing sources have been largely controlled,
some sources of metals are controlled poorly and represent a continuing
source (e.g., leaching from abandoned mining sites, urban runoff). Un-
like oxygen-demanding contaminants, these pollutants are not easily
neutralized by natural processes.
• Oxygen-Demanding Contaminants
A variety of organic and inorganic compounds in sediments consume ox-
ygen during chemical fate reactions. The cumulative effect of their pres-
ence is measured by sediment oxygen demand, a parameter similar in
signiÞcance to oxygen-demanding measures in the overlying water. Sed-
iment oxygen demand serves to reduce available oxygen in the sediment
and encourage anaerobic conditions within the sediment. This can inßu-
ence the rate of fate processes (e.g., biological contaminant degradation)
and the chemical state of metals, inßuencing their mobility. In slow-mov-
ing water, the sediment oxygen demand can also impact oxygen levels in
the overlying water. No speciÞc levels of oxygen-demanding constituents
are considered problematic. Rather, the impact of these contaminants de-
pends on the dynamics of the sediment and overlying water column.
• Undifferentiated Oil and Grease
Long-chain nonpolar organic compounds, such as oil and grease, asso-
ciate strongly with solids and sediments. Their presence in sediments is
measured by oil and grease and total petroleum hydrocarbon (TPH)
concentration. The sources of these compounds are generally TPH pro-

duction or processing facilities or facilities that use or process signiÞ-
cant amounts of these compounds. In addition, municipal and industrial
wastewater treatment efßuents can lead to signiÞcant accumulation in
sediments over time. Because many of these sources are much more
carefully controlled than in the past, oil and grease or TPH levels in sed-
iments often represent excellent indicators of historical pollution.
• Pesticides
The pesticide priority pollutants are generally chlorinated hydrocar-
bons. They include compounds such as aldrin; dieldrin; 1,1,1-trichloro-
2,2-bis(

p

-chlorophenyl)ethane (DDT); 1,1-dichloro-2,2-bis(

p

-chloro-
phenyl)ethane (DDD); endosulfan; endrin; heptachlor; lindane; and

L1667_book.fm Page 71 Tuesday, October 21, 2003 8:33 AM
©2004 CRC Press LLC

chlordane. They are readily assimilated by aquatic animals and bioac-
cumulate in body fats and are subject to biomagniÞcation. BiomagniÞ-
cation refers to the tendency of these pollutants to increase in
concentration at higher trophic levels.
DDT is an excellent example of the potential problem associated with
these compounds. Although it is not considered extremely toxic to peo-
ple and its usage was decreasing during the 1960s, the use of DDT was

effectively banned in 1972 as a result of its persistence, its potential for
biomagniÞcation, and its effects on wildlife. Braune and Nordstrom
(1989) measured a herring gull/alewife trophic transfer factor of 85 for
DDE, the dominant persistent metabolite of DDT. Oliver and Niimi
(1988) measured a biota–sediment accumulation factor of 3.7 (

m

g/g
wet)/(

m

g/g dry)



for DDE in sculpins in a Lake Ontario ecosystem. The
biota–sediment accumulation factor is the accumulation in the organ-
ism normalized by the organism’s lipid content divided by the organic
carbon–normalized sediment concentration. The magniÞcation at each
level is dependent on the feeding habits and animal metabolism and, be-
cause the organochlorines tend to build up in the lipid or fat fraction of
the body, the proportion of body fat.
• PCBs
PCBs are complex mixtures of organochlorines that are extremely stable
and are used widely in industry, especially as electrical capacitor and
transformer oils. Unfortunately, the stability of PCBs also means that
they are persistent in the environment. As with organochlorine pesticides,
PCBs are readily assimilated by aquatic animals, soluble in body fats, and

biomagnify in the food chain. Although the toxicity of many individual
PCBs is relatively low, speciÞc isomers plus trace contamination with
other chlorinated species raised signiÞcant health concerns. As a result,
PCB production was banned in the U.S. in 1979. It should be emphasized
that PCBs are a complex mixture of compounds and, in fact, are generally
named only by the total percentage of chlorine in the mixture. SpeciÞc
PCB mixtures are referred to as Aroclors. For example, Aroclor 1254
contains 54% chlorine and Aroclor 1260 contains 60% chlorine.
Again, as a result of the material’s persistence, signiÞcant quantities re-
main in the environment. Industrialized harbor areas in the Great Lakes
and northeastern U.S. represent the most signiÞcant repository of
PCBs. Fish advisories exist in many of the Great Lakes as a result of
health concerns from eating PCB-contaminated Þsh. Because of the po-
tential for PCBs to sorb onto organic materials in sediments and Þsh lip-
ids, such advisories are aimed primarily at fatty, bottom-feeding Þsh
and top predators in which PCB concentrations are the highest.
•PAHs
PAHs (e.g., naphthalene, ßuoranthene, pyrene, chrysene) are used as
chemical intermediates and are present in fossil fuels. PAHs composed
of two aromatic rings (e.g., naphthalene) tend to be the most volatile,
soluble, and mobile. Solubility and volatility as well as degradability

L1667_book.fm Page 72 Tuesday, October 21, 2003 8:33 AM
©2004 CRC Press LLC

under natural conditions tend to decrease as the number of rings in-
creases. Although the monocyclic aromatics tend to be present in the
lighter oil stocks, PAHs tend to be present in coal liquids and the heavi-
er oil stocks as a result of their lesser volatility. Some PAHs have been
found to be carcinogenic in animals and are assumed to be carcinogen-

ic in humans. PAHs tend to be intermediate in persistence and have a
bioaccumulation potential between the monocyclic aromatics/haloge-
nated aliphatics and the PCBs. The use of PAHs during the combustion
of industrial fuels and oils (e.g., diesel, coal liquids, heavy fuel oils)
has resulted in their presence at old industrial sites where contamina-
tion levels can be especially high.

2.2.2 S

EDIMENT

T

RANSPORT

M

ODEL

C

OMPONENTS



2.2.2.1 Sediment Erosion and Deposition Processes

As outlined above, a critical component of any attempt to describe the behavior of
sediment contaminants involves a need to describe the migration and fate of the
sediments with which they are associated. The dominant characteristics that control

the direct exposure of Þsh and higher animals to contaminated sediment are burial,
vertical mixing of the sediment bed, and resuspension of particles from the sediment
bed. Because most persistent sediment contaminants are associated with the solid
phase, any mobilization of this phase dramatically increases contaminant mobility.
As a result, contaminants can be distributed over large areas, and signiÞcantly
increased water column concentrations can be observed relative to less active sedi-
ment–water transport. Erosion and resuspension conditions also eliminate natural
recovery that might occur in less active environments due to deposition and burial
of the contaminated sediment.
Under high energy conditions in a stream, erosion of the sediment bed can occur,
and individual sediment particles can be carried downstream either by sliding along
the surface of the sediment or by being suspended in the stream. In a sandy sediment,
the process normally results in the formation of dunelike structures that progress
downstream by the process of erosion on the upstream face and deposition on the
downstream face. The erosion process is depicted in Figure 2.2.
During this overturning and migration process, sediment particles are exposed
and either scoured and suspended in the stream or reburied by other sediment
particles. During exposure to the stream water, contaminants sorbed onto the sedi-
ment particles can be desorbed, and contaminants in the adjacent pore water can be
mixed into the overlying water. It should be noted that this process occurs in the
sand and gravel bed where, typically, organic carbon and contaminant levels are
relatively low. Once particles are set in motion, the following three types of move-
ment can be recognized:
• Rolling and sliding along the bed surface (i.e., bed load transport)
• Suspension in the free stream (i.e., suspended load)
•A transitional motion characterized by saltation and particle jumps

L1667_book.fm Page 73 Tuesday, October 21, 2003 8:33 AM
©2004 CRC Press LLC

The ability to predict the onset of resuspension in a sediment remains largely
limited to cohesionless, coarse-grained particles. Site-speciÞc measurements of bed
and/or suspended load sediment transport are needed to characterize cohesive, Þne-
grained sediment. The onset of particle motion in a cohesionless sediment is deter-
mined by the balance between the submerged weight of a particle and the lift
associated with the water ßow over the particle. Raudkivi (1967) discusses the
physics of the onset of the particle motion and shows that the onset of particle motion
occurs when a critical threshold friction velocity is exceeded.
(2.2)
Here

u

*c

is the critical threshold friction velocity;

r

p

and

d

p

are the particle density
and diameter, respectively;

r

w

is the density of water; and

b

is a coefÞcient incor-
porating the angle of repose of the particle (i.e., the slope of the upstream face of
the sediment dune) and the partial coverage by other sediment particles. The friction
velocity is related to the surface shear stress,

t

b

,

by u

*c

= (

t

b

/

u

w

)

1/2

. This relationship
simply emphasizes that sediment resuspension occurs when the lift caused by the
overlying ßow overcomes the weight of the particle. The friction velocity is a
parameter related to the surface friction that can be determined by velocity proÞle
measurements. Raudkivi (1967) suggests that

b



is approximately 0.2.
For cohesive sediments, the friction velocity required to produce particle motion
is signiÞcantly larger for a particular particle size than would be suggested by the
above relationship. The property of cohesiveness is a complicated function of particle
size, bulk density, mineralogy, organic content, and salinity. These properties vary
signiÞcantly with position and time. Often as a result of lack of sufÞcient data on
the deposit properties with position and time, these variations are not fully incorpo-
rated in sediment transport models. The rate of erosion,

E

, is related to the local bed
density,

r

s

, and the probability of a particle becoming resuspended, which for a
cohesive sediment is related to the bottom shear stress,

t

b

.

FIGURE 2.2

Sediment movement due to erosion processes.
=1
I
STAGNATION
POINT
BED LOAD
PA RTICLE
SEPARATION
POINT
SUSPENDED
PA RTICLE
ANGLE OF

REPOSE
DRAG
ugd
c
pw
w
p
*
=
-
b
rr
r
L1667_book.fm Page 74 Tuesday, October 21, 2003 8:33 AM
©2004 CRC Press LLC
(2.3)
The exponent on the bottom shear stress depends on the bed properties but is
typically between two and three for cohesive sediments. This implies that the erosion
rate depends on the fourth to sixth power of stream velocity because bottom shear
stress typically depends on the square of velocity. The strong dependence on stream
velocity emphasizes that a critical component of any effort to model sediment
dynamics is knowing the stream hydrodynamics. Although it is not yet possible to
predict the relationship between erosion rate and shear stress for cohesive sediments,
it is possible to make measurements from which the values of A, n, and m can be
determined (McNeill et al., 1996).
Net sediment transport is the difference between the erosion rate, deÞned above,
and the deposition rate. In general, deposition can be modeled with a relationship
of the form
(2.4)
where P

dep
is the probability of capture of the depositing particle, w
s
is the vertical
settling velocity of the particles, and C
s
is the suspended sediment concentration.
The probability of deposition tends to decrease as the bed shear stress increases.
The local particle concentration can be modeled as a decreasing exponential with
height above the bed (Jones and Lick, 2000). There are also signiÞcant differences
between cohesive (ßocculating) sediment and noncohesive (i.e., sandy) sediment.
Sandy sediment deposition can be modeled employing the formulation of Cheng
(1997). In cohesive sediments, deposition is affected by aggregation or disaggrega-
tion processes that are complex functions of sediment and stream conditions (Lick
and Lick, 1988).
The most easily measured quantity with which to calibrate sediment transport
models is suspended sediment load in the overlying water column. Because the
source of suspended sediment concentrations can be runoff from surrounding soil
or transport from points upstream in addition to direct sediment bed erosion, such
measurements are of limited usefulness in assessing or calibrating sediment bed
erosion. Changes in suspended sediment load with location or with time in response
to variations in ßow are a more sensitive indicator of local erosion and deposition.
Changes in bathymetry also provide a more direct indication of sediment bed move-
ment. Changes in water depth can be small or highly variable in space, suggesting
that the uncertainty in measured changes can limit usefulness as a calibration tool.
2.2.2.2 Sediment Transport Model Minimum Requirements
A hydrodynamic model must accurately describe the space and time variations in
stream velocity because of the critical role velocity plays in sediment transport. A
ßexible sediment transport model cannot rely on measured velocity Þelds because
simulated conditions are likely to extend beyond the measurement database. Data

requirements for such a hydrodynamic model include streambed geometry and
EP A
ero
b
n
s
b
n
s
m
==()tr tr
DPwC
dep
ss
=
L1667_book.fm Page 75 Tuesday, October 21, 2003 8:33 AM
©2004 CRC Press LLC
bathymetry, ßow rates, downstream water surface elevation (e.g., boundary condi-
tions), and water and atmospheric properties (e.g., salinity, temperature, wind). Water
surface elevation and current velocity measurements and water property documen-
tation can be used to calibrate or validate the model. Data requirements for a sediment
transport model include measured sediment loads (i.e., magnitude and composition)
and bed properties including erosion potential (e.g., as measured via “sedßume”;
McNeill et al., 1996). Measured suspended solid concentrations and sediment bed
elevation changes provide calibration and validation data. For consistency, both
hydrodynamic and sediment transport models should use the same numerical grid
of sufÞcient resolution to resolve bathymetry and bed variability.
The models employed for both hydrodynamics and sediment transport should
be coupled prognostic, material balance–based models. The dimensionality required
depends on the system as does the degree of sophistication required in the wind-

forcing model. As emphasized above, the sediment transport model should be fun-
damental and mechanistic to aid extrapolation beyond the calibration validation
database. At a minimum, the sediment transport model should separately track
ßocculating clay and silt, and noncohesive sandy components. Models that simulate
only a single size class of particles and do not capture the physics of resuspension
and settling are not able to correlate sediment and contaminants adequately. Predict-
ing contaminant dynamics requires unique identiÞcation of the sources and sinks of
particular sediment deposits. Seemingly accurate reproduction of total suspended
sediment loads does not necessarily predict contaminant dynamics. To describe
sediment and contaminant dynamics, deposition should account for the effect of
shear stress and, for cohesive sediments, ßocculation effects. Resuspension should
account for site-speciÞc bed properties and, if necessary, the effects of bed armoring
by consolidation or size gradation.
Finally, it should be emphasized that sediment transport is not dependent upon
contaminant fate. Sediment transport model calibration using contaminant fate and
transport measurements is inappropriate. Instead, a sediment transport model pro-
vides an additional constraint upon the contaminant transport.
2.2.3 CONTAMINANT FATE AND TRANSPORT MODEL COMPONENTS
2.2.3.1 Contaminant Fate and Transport Processes
in Unstable Sediments
The conventional paradigm for deÞning the fate and behavior of hydrophobic and
other contaminants that are strongly associated with the solid phase is that con-
taminant dynamics are largely deÞned by sediment dynamics. To the extent that
contaminants remain associated with the solid phase, the description of sediment
erosion and deposition also deÞne contaminant transport. As indicated previously,
however, the extent to which contaminants are associated with the solid phase
depends on the effective sediment–water partition coefÞcient and the local solid-
phase density. Suspended sediment loads in rivers rarely exceed 1 g/l even down-
stream of dredging operations. If the effective sediment–water partition coefÞcient
is of the order of 1,000 l/kg, 1 g/l of suspended sediment suggests that the water

L1667_book.fm Page 76 Tuesday, October 21, 2003 8:33 AM
©2004 CRC Press LLC
column contaminant is 50% dissolved and 50% associated with the particulate
fraction. At lower suspended sediment concentrations or lower sediment–water
partition coefÞcients, the contaminant is found predominantly in the dissolved
state in the water. Under such conditions, contaminant fate is no longer tied to the
fate of the sediment, and other processes, including vaporization and in-stream
dispersion, dominate contaminant transport.
The surface layer of the sediment bed provides the crucial link between sedi-
ments, water, and biota. Invertebrates that live in or on the sediment bed accumulate
contaminants from the surface layer. The contaminant concentration in the surface
layer is controlled by both sediment dynamics and chemical hydrophobicity. Dep-
osition and resuspension rates control the rate at which contaminated surface sedi-
ments are buried, that is, are moved out of the active surface layer. In addition, the
depth and extent of surface layer mixing controls the degree to which buried con-
taminants are kept available to the surface layer. Chemicals that are relatively more
hydrophobic are more strongly associated with the particulate matter in the bed.
This results in lower concentrations dissolved in the pore water and, therefore, lower
concentrations available for diffusion and advection in the overlying water column.
2.2.3.2 Contaminant Fate and Transport Processes
in Stable Sediments
Under some conditions, contaminant dynamics can be decoupled from hydro-
dynamics and sediment erosion and deposition. A variety of processes, including
diffusion, advection, and bioturbation, can be responsible for contaminant
exchange under conditions where bulk sediment transport is unimportant. Under
these conditions, contaminant exchange at the sediment–water interface serves as
a boundary condition for a fate and transport model within the water column.
Determining water column contaminant concentrations and the fate and transport
of water-borne contaminants requires the deÞnition of the sediment boundary as
a source or sink of contaminants.

In general, the controlling resistance to contaminant mass transfer to or from
stable (i.e., noneroding) sediments can be either the movement within the benthic
boundary layer in the overlying water or the movement within the sediment bed.
Because of the traditional focus of contaminated sediment models in accurately
describing the sediment transport, simple lumped exchange coefÞcients are typi-
cally used to characterize contaminant release from the sediment. Lumping the
individual transport processes in the overlying water and in the sediment bed with
effective mass transfer coefÞcients k
s
and k
w
, with units of length and time, the
ßux between the bulk sediment and the sediment–water interface (at concentration
W
si
) can be written as follows:
(2.5)
Similarly, the ßux between the sediment–water interface (at concentration C
wi
)
and the overlying bulk water can be written as follows:
Flux k W W
ss s si
=-r ()
L1667_book.fm Page 77 Tuesday, October 21, 2003 8:33 AM
©2004 CRC Press LLC
(2.6)
DeÞned in this manner, the mass transfer coefÞcients, k
s
and k

w
, represent
normalized ßuxes (i.e., the ßux normalized by the concentration driving force in the
respective phase). The concentrations at the interface, W
si
and C
si
, are generally
assumed to be in quasi-steady equilibrium. Some fraction of the contaminants can
be Þxed to the sediment phase and are unable to be released to the pore water or
overlying water as discussed earlier. Regardless of the cause and ultimate extent of
limited contaminant desorption from sediments, the ratio of the quantity effectively
sorbed onto the solid phase at the interface, W
si
, to that which is in the adjacent
aqueous phase, C
si
(K
sw
= W
si
/C
si
), represents an effective partition coefÞcient.
An overall mass transfer coefÞcient, K, can be deÞned using the difference in
the bulk concentrations in each phase (W
s
and C
w
) as the driving force. Equating

the different expressions for the ßux and assuming that the interfacial concentrations
are at an apparent equilibrium (K
sw
= W
si
/C
wi
), the overall mass transfer coefÞcient
can be related to the individual sediment and water coefÞcients as follows:
(2.7)
The relative importance of the sediment and water-side mass transfer resis-
tances thus depends on the relative magnitude of the individual side coefÞcients.
However, this relationship indicates that the importance of water-side mass transfer
resistances increases as the capacity for sorption onto the solid phase increases.
The water-side coefÞcients are a function of ßow and chemical properties. Stream
velocity and bottom roughness are characterized by the bottom friction velocity.
The Schmidt number (N
sc
= D
w
/n
w
, where D
w
is the chemical diffusivity in water
and n
w
is the kinematic viscosity of water) characterizes the inßuence of the
chemical. Density stratiÞcation and local variations in bottom roughness can
inßuence water-side mass transfer coefÞcients, but the average coefÞcient in neu-

trally stratiÞed water bodies is often described by a relationship of the following
form (Reible, 1999):
(2.8)
where k is the von Karman constant (0.4). For many organic compounds, the Schmidt
number is the order of 1000, whereas the friction velocity is typically of the order
centimeters per second (cm/sec). Thus, the water-side mass transfer coefÞcient is
generally of the order of 100 cm/day. K
sw
is of the order of 1000 l/kg or more for
many hydrophobic contaminants, and sediment bulk density is of the order of 1
g/cm
3
(1 kg/l). Thus, water-side mass transfer resistances are the same order as
sediment-side exchange coefÞcients of the order of 0.1 cm/day. Higher sediment-
side exchange coefÞcients imply control by water-side Quantitative Environmental
Flux k C C
wwi w
=-()
11
Kk
K
k
s
ssw
w
=+
r
kuN
wSc
=

*
-
1
2
23
k
/
L1667_book.fm Page 78 Tuesday, October 21, 2003 8:33 AM
©2004 CRC Press LLC
Analysis mass transfer resistances; lower sediment-side exchange coefÞcients mean
water-side mass transfer resistances are of lesser importance. The various processes
that make up sediment-side mass transfer resistances and their magnitudes are
discussed below.
• Molecular Diffusion
Molecular diffusion is the most basic and ubiquitous chemical transport
process within a sediment bed. Random molecular motion in pore water
results in contaminant molecule movement from regions of high pore-
water concentration to those of low concentration. The magnitude of
the contaminant ßux is quantiÞed by Fick’s Þrst law and couples the
concentration gradient to the diffusion coefÞcient. The diffusion coefÞ-
cient in porous sediments is reduced by the Þnite porosity of the sedi-
ments and the tortuous path through the sediments. Tortuosity, t, the
ratio of the actual diffusion path to the straight line distance, generally
has been found to be a function of bed porosity, suggesting the follow-
ing model for diffusivity:
(2.9)
Here D
w
is the molecular diffusion coefÞcient of the contaminant in
free water which is typically of the order of 0.5 to 1.5 cm

2
/day. The
value of n is typically in the range 1.33 < n < 2 with sandy, granular
sediments exhibiting lower values of n than cohesive sediments.
Thus, D
sw
is typically of the order of 0.1 cm
2
/day. The quasi-steady
diffusion ßux from a bulk sediment concentration at depth, d, can be
written as follows:
(2.10)
This relationship recognizes that diffusion is operative only on the de-
sorbed fraction of contaminants. As a result, diffusion is intrinsically
much slower than water-side mass transport (and therefore a control-
ling resistance to mass transfer) unless the diffusion path length is
very small.
Diffusive transport can be enhanced by processes that enhance con-
taminant release to the mobile water phase. Chief among these is the
presence of colloidal organic carbon in the pore water. The colloidal
organic carbon (often operationally deÞned as dissolved organic
carbon) can migrate through the sediment pore space via Brownian
diffusion, causing migration of associated contaminants. For hydro-
phobic organic compounds, the organic carbon–based partition co-
efÞcient (K
oc
= K
sw
/f
oc

, where f
oc
is the fraction organic carbon) is
DD D
sw w w
n
=ª
e
t
e
Flux
D
K
WW k
D
K
sw s
sw
ssi
s diff
sw s
sw
=- ª
r
d
r
d
()
,
L1667_book.fm Page 79 Tuesday, October 21, 2003 8:33 AM

©2004 CRC Press LLC
often used to approximate the partitioning to colloidal organic car-
bon. The enhancement of pore-water transport processes by colloi-
dal organic carbon present at concentration, D
oc
, can then be
estimated as follows:
(2.11)
• Advective Transport
Advective transport at a superÞcial, or Darcy velocity, V, is also opera-
tive only on pore-water contaminants. Hydraulic exchange processes at
the sediment–water interface occur at a variety of scales. At the sedi-
ment-grain scale, turbulent ßuctuations in the overlying water can en-
hance local transport from sediment to water. On the scale of the uneven
bedforms that typically characterize the bottom surface, local pressure
variations can enhance in-bed migration and contaminant transport (Sa-
vant et al., 1987). Similarly, cross-stream pressure variations at stream
meanders can generate local advective ßows within the underlying sed-
iment. Finally, the regional ground water gradients can drive net ground
water movement into or out of the stream. Local variations in bed prop-
erties also can give rise to signiÞcant variability in the magnitude or
even the direction of this transport. As with diffusion, however, the im-
plementation of these processes in current models is generally limited
to an average exchange coefÞcient, which, at steady state, is deÞned by
the following relation:
(2.12)
Flows driven by local processes and even regional ground water ßows
with signiÞcant spatial variability in bed permeability are not well rep-
resented by an average exchange coefÞcient. Spatial variations in sedi-
mentation contribute to heterogeneous contaminant deposition and

consequent ßow and transport variations. Bed dynamics such as silt-
ation or coarsening over time further complicate the description of ßow
and transport in the hyporheic zone. The local ratio of advective to dif-
fusive processes is given by the following Peclet number:
(2.13)
Note that the Peclet number is not inßuenced by sorption because both
processes are operative on pore-water contaminants. Both processes
also are enhanced equally by colloidal or other mechanisms that in-
D
D
K
k
k
K
sw
coll
sw
oc oc
s
coll
s
oc oc
=+ =+11rr
Flux
V
K
Wk
V
K
sw

s
s adv
sw
=ª
ee
,
N
V
D
Pe
w
n
=
+
d
e
1
L1667_book.fm Page 80 Tuesday, October 21, 2003 8:33 AM
©2004 CRC Press LLC
crease pore-water concentrations. It is important to note that the Peclet
number typically exceeds unity (i.e., advection dominates) at low
ground water seepage velocities. Over a characteristic length scale of 1
cm and with an effective diffusivity of 0.1 cm
2
/day, advection domi-
nates transport if the seepage velocity exceeds 0.1 cm/day.
Measuring ground water ßow velocities and, in particular, stream bed
seepage velocities is difÞcult because these measurements must reßect
the seasonal nature and spatial variability of ground water ßow. The use
of seepage meters (i.e., containers covering a portion of the sediment

bed that collect water that seeps through the sediment) is common. Net
seepage measurements, however, do not fully describe contaminant
transport in that the bulk of any observed seepage can be through sandy
portions of the bed that contain little or no contamination. Chemical
tracers in the overlying water have been used to estimate the net effect
of seepage on contaminant transport (Bencala, 1984). Spatial variabili-
ty is made more complex in that advection also can be driven by local
variations in pressure on the uneven surface of the sediments. These
ßows are not normally incorporated in current contaminated sediment
models except as being a component of the sediment-side mass ex-
change coefÞcient.
Rather than measuring velocity, measuring ground water ßow in the
surrounding aquifer can be useful because this measurement represents
an average inßow or outßow from the waterbody. The general direction
of the ground water ßow can be measured by piezometers placed at dif-
ferent elevations below the bed of the waterbody. If the underlying wa-
ter head is greater than the head in the stream, inßow occurs; outßow
occurs in the reverse situation. In addition to deÞning direction, this in-
formation is used to estimate the ßow rate if the permeability of the me-
dium can be measured. An alternative means of detecting slow vertical
transport by ground water ßow is through tracers, as described by Cor-
nett et al. (1989).
• Bioturbation
The previous discussion largely considered sediment as a collection of
sediment particles separated by water-Þlled pore spaces. In reality, a va-
riety of plants and animals reside in sediments. Root systems and ani-
mal burrows can provide channels for preferential water ßow and
contaminant transport. Even more important, the near-surface sediment
often is mixed continuously by the activities of benthic organisms such
as clams and worms.

Sediment processing by animals residing in the upper layers includes
burrowing, ingestion and defecation, tube building, and biodeposition.
Taken together, these processes are termed bioturbation. A depiction of
the type of animals that can be present and their interaction with sedi-
ments is provided in Figure 2.3. The net result of bioturbation is the ver-
tical and horizontal movement of sediment particles and pore water.
Contaminants on the particles or in the pore spaces likewise are trans-
L1667_book.fm Page 81 Tuesday, October 21, 2003 8:33 AM
©2004 CRC Press LLC
ported in the bioturbation process, which is especially important when
transporting hydrophobic contaminants that are heavily retarded by
pore-water processes. Worm tubes and other macroscopic animal bur-
rows can signiÞcantly enhance contaminant transport by advection
across the sediment–water interface. In addition, direct ingestion of
sediment deposits can lead to rapid transport of sediment and associated
contaminants to the surface.
Bioturbation is not fully incorporated in most system-wide contaminated
sediment models and is often lumped together with other processes as
part of the sediment-side mass exchange coefÞcient or an effective bio-
diffusion coefÞcient. Effective biodiffusion or exchange coefÞcients are
crude approximations at best because they do not separate the various
modes of particle movement exhibited by an organism. In addition, be-
cause they are normally measured with strongly sorbing contaminants,
pore-water pumping and the effect of bioturbation on other diffusive and
advective processes (i.e., creation of secondary porosity, changes in tex-
ture or permeability) are not included. Most Þeld measurements of
FIGURE 2.3 Feeding types of benthic organisms. (After Rhoads, 1974.)
L1667_book.fm Page 82 Tuesday, October 21, 2003 8:33 AM
©2004 CRC Press LLC
bioturbation can be described by an effective diffusion coefÞcient of the

order of 10
–4
to 0.01 cm
2
/day or an effective mass transfer coefÞcient of
10
–5
to 0.001 cm/day based on a 10-cm biologically active zone. As stated
previously, effective molecular diffusion coefÞcients in sediments are of
the order of 0.1 cm
2
/day. For a hydrophobic contaminant (e.g., pyrene)
with an approximate partition coefÞcient, K
sw
~1000 l/kg, bioturbation is
expected to control contaminant migration in the upper layers of a stable
sediment bed. Using 1 cm/year (0.003 cm/day) as an effective bioturba-
tion exchange coefÞcient, the magnitude of the bioturbation and molecu-
lar diffusion coefÞcients are as follows:
(2.14)
Thus, bioturbation-induced transport is of the order of 30 times larger
than diffusive transport in this example. For some elemental species
whose leachable fraction can partition only weakly into the solid
phase, the enhancement by bioturbation may be minimal. However,
pore-water pumping or other soluble transport processes that are not
part of most biodiffusion coefÞcient measurements can be important
for these species.
One area in which very little information is known is hydrophobic con-
taminant transport via gas movement through sediments. The source of
the gas is often biological (e.g., the production of hydrogen sulÞde or

methane by anaerobic metabolic processes in microorganisms). Hydro-
phobic organic compounds tend to sorb preferentially at the gas–solid
interface. The mass of contaminant moving in this manner can be rela-
tively small but still may result in surface recontamination, and the bulk
movement of the gas phase could conceivably alter the physical prop-
erties of the sediment bed.
The dominance of transport by bioturbation and other biological pro-
cesses is greatest near individual organism burrows. In general, biotur-
bation is the primary migration mechanism of sorbing contaminants in
stable surÞcial sediments unless the physical character of the sediment
or its level of contamination precludes signiÞcant colonization by
benthic organisms or ground water seepage is such that advection
dominates. As indicated previously, in most system-wide models of
contaminant transport from sediments, the effect of bioturbation is
lumped into an overall mass exchange coefÞcient with other sediment-
side transport processes.
2.2.3.3 Contaminant Transference via Food Webs
Ultimately, it is the portion of the contaminant that moves via natural processes into
the water or food chain that is the source of exposure and potential risk of contam-
inated sediments to Þsh and higher animals. The exposure and risk depend on the
kk
s diff s bio,,
.
== =
01
1000
0 0001 0 003
cm / day
cm / day cm / day
L1667_book.fm Page 83 Tuesday, October 21, 2003 8:33 AM

©2004 CRC Press LLC
organism’s access to the contaminant, the contaminant’s availability, and the organ-
ism’s assimilative capacity. Contaminant assimilation and risk of effects in higher
animals generally depend on exposure via one of the following pathways:
• Direct exposure to Þsh and higher animals by contaminant release from
resuspended contaminated sediment or by incidental ingestion of contam-
inated bed sediments
• Direct exposure to Þsh and higher animals from contaminant release in
dissolved form from bed sediment to the overlying water
• Indirect exposure to higher animals by predation and plant harvesting and
animals that ingest or otherwise assimilate sediment contaminants
For hydrophobic compounds strongly associated with particulate matter, the third
pathway dominates. The analysis of contaminant transfer and exposure via the food
web is complicated by species diversity and feeding behaviors and movement patterns.
Often, the organism of primary interest is high in the food web with multiple levels
separating the organism from the contamination source. The food webs vary with
time, especially with season, and with space. Bioaccumulation can be analyzed in
two ways. For organisms that exhibit relatively little movement and for which con-
taminant concentrations respond rapidly to changes in exposure concentrations, sim-
ple linear relationships between exposure sources and body burdens are appropriate.
Organisms at the base of the food web, primarily benthic and water column inverte-
brates, are generally in this category. For example, a biota and sediment accumulation
factor (BSAF) can be deÞned as the ratio of contaminant concentrations in the
organism and local sediments. For hydrophobic compounds, the tendency of organ-
isms to accumulate contaminants is often deÞned by the lipid content of the organism.
Similarly, the bioavailability of the contaminant in sediment is controlled to a large
degree by the organic carbon content of the sediment. Thus, the BSAF for hydropho-
bic compounds is generally deÞned as the ratio of the lipid-normalized accumulation
in an organism and the organic carbon–normalized quantity in the sediment.
(2.15)

Here, the quantity W
b
is the concentration of contaminant in the organism
(mg/kg wet wt), f
l
is the fraction of lipids (g lipid/g wet wt), W
s
is the concentration
of contaminant in the sediment (mg/kg dry wt), and f
oc
is the fraction organic
carbon (OC) of the sediment (g OC/g dry wt). Note that the use of the BSAF does
not imply chemical equilibrium, only a steady relationship between contaminant
levels in sediment and organism. Organisms are not in chemical equilibrium with
their environment; the contaminant level reached represents the outcome of uptake,
loss, and dilution processes in the organism. BSAF values are for chemical- and
species-speciÞc insofar as the bioavailability, assimilation, efÞciency, and depu-
ration rates differ among chemicals and species. The appropriate units of the BSAF
are also chemical speciÞc. For example, methyl mercury tends to associate with
BSAF
Wf
Wf
bl
soc
=
/
/
L1667_book.fm Page 84 Tuesday, October 21, 2003 8:33 AM
©2004 CRC Press LLC
protein and, therefore, lipid normalization of the organism body burden is not

appropriate. In this case, the units of the numerator of the BSAF are more appro-
priately mg/kg dry wt.
For organisms far removed from the sediment (e.g., Þsh in a pelagic food web),
BSAFs cannot provide an accurate estimate of contaminant concentrations under
changing conditions in sediments and local waters. In this circumstance, assuming
a constant BSAF results in an inaccurate prediction of contaminant concentrations
in a water column-based food web.
To improve one’s ability of predicting exposure and accumulation at higher
trophic levels, it is necessary to explicitly model contaminant transfer through the
food web. Bioaccumulation models exist that can simulate the transfer of contam-
inants through complex, dynamic food webs over the entire life cycle of an organ-
ism, incorporating complex movement patterns in a dynamic (i.e., time variable)
framework. There are several frameworks available, of which two in particular have
been widely applied to real-world problems. The framework developed by Frank
Gobas (Gobas, 1993; Gobas et al., 1995) has been used at several sites and in the
development of methodologies for computing water quality criteria. The model
framework developed by Thomann and Connolly (1984) has also been applied at
multiple sites with multiple chemicals and has been used by the USEPA in devel-
oping water quality criteria. These models were recently compared and found to
compute similar contaminant concentrations. (Another model whose development
has been supported by the USEPA is Aquatox [USEPA, 2000].) As an illustration,
Farley and Strauss (2000) employed the time-variable, age-dependent bioaccumu-
lation model developed by Thomann and Connolly (1984) to predict PCB transport
through a lower Hudson River food web composed of phytoplankton, zooplankton,
white perch, and striped bass. Time-dependent diffusive uptake, ingestion, egestion
and excretion, growth dilution, and metabolic processes at each of these trophic
levels were modeled.
Bioaccumulation modeling has developed sufÞciently to permit realistic calcu-
lations of the extent of trophic transfer of many chemicals in aquatic food webs,
including hydrophobic organics and metals. Such models generally are limited by

the availability of data to deÞne biological processes, in particular food web structure
and the ultimate source of the contaminant to the food web (i.e., sediment vs. water
column) and movement/migration patterns. Food web models can also be linked
with chemical fate and transport models to give a uniÞed mass balance framework
to help assess potential remediation strategies. Because of the advanced state of food
web models, they can be used effectively to evaluate the sensitivity of model pro-
jections to uncertainty in key biological processes. In this way, the adequacy of
linked chemical fate and food web models can be evaluated in light of the desired
model objectives.
2.2.3.4 Human and Ecological Risk Evaluation
DeÞning exposure and accumulation in receptor organisms does not complete the
analysis. A risk-based approach to managing contaminated sediments ultimately
requires translation of that exposure and uptake into effects. Estimates of relative
L1667_book.fm Page 85 Tuesday, October 21, 2003 8:33 AM
©2004 CRC Press LLC