EPA/600/R-16/210 | September 2016 jwww.epa.gov/research
United States
Environmental Protection
Agency
A Manual to Identify
Sources of Fluvial Sediment
Office of Research and Development
National Risk Management Research Laboratory
Land Remediation and Pollution Control Division
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
A Manual to Identify Sources of Fluvial
Sediment
by
Allen Gellis, U.S. Geological Survey, Maryland-Delaware-D.C. Water Science
Center
Faith Fitzpatrick, U.S. Geological Survey, Wisconsin Water Science Center
Joseph Schubauer-Berigan, USEPA ORD, NRMRL-Cincinnati
Interagency Agreement/Grant/Contract Number
DW-14-95825801
Prepared in cooperation with the U.S. Geological Survey
Project Officer
Joseph P. Schubauer-Berigan, Ph.D.
Land Remediation and Pollution Control Division
National Risk Management Research Laboratory
Cincinnati, Ohio, 45268
i
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
I - in I i
The U.S. Environmental Protection Agency, through its Office of Research and Development, funded and
conducted the field and laboratory research activities described herein under an approved Quality Assurance
Project Plan. Mention of trade names or commercial products does not constitute endorsement or
recommendation for use. This report has been reviewed by the U.S. EPA's Office of Research and Development
National Risk Management Research Laboratory and approved for publication. Approval does not signify that
the contents necessarily reflect the views and policies of the Agency. This report covers a period from July 30,
2012 to August 15, 2016 and work was completed as of August 15, 2016.
ii
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
[r , -|
The U.S. Environmental Protection Agency (USEPA) is charged by Congress with protecting the
Nation's land, air, and water resources. Under a mandate of national environmental laws, the Agency
strives to formulate and implement actions leading to a compatible balance between human activities
and the ability of natural systems to support and nurture life. To meet this mandate, USEPA's research
program is providing data and technical support for solving environmental problems today and building
a science knowledge base necessary to manage our ecological resources wisely, understand how
pollutants affect our health, and prevent or reduce environmental risks in the future.
The National Risk Management Research Laboratory (NRMRL) within the Office of Research and
Development (ORD) is the Agency's center for investigation of technological and management
approaches for preventing and reducing risks from pollution that threaten human health and the
environment. The focus of the Laboratory's research program is on methods and their cost-effectiveness
for prevention and control of pollution to air, land, water, and subsurface resources; protection of water
quality in public water systems; remediation of contaminated sites, sediments, and ground water;
prevention and control of indoor air pollution; and restoration of ecosystems. NRMRL collaborates with
both public and private sector partners to foster technologies that reduce the cost of compliance and to
anticipate emerging problems. NRMRL's research provides solutions to environmental problems by:
developing and promoting technologies that protect and improve the environment; advancing scientific
and engineering information to support regulatory and policy decisions; and providing the technical
support and information transfer to ensure implementation of environmental regulations and strategies at
the national, state, and community levels.
Sediment is one of the most common causes for the loss of stream-biologic integrity. Identifying
sediment sources is an important step in the USEPA's sediment TMDL process. The objective of this
study was to develop a guidance document for sediment source analysis. The guidance document
developed synthesized studies that incorporate sediment fingerprinting and sediment budget approaches
in agricultural and urban watersheds.
Cynthia Sonich-Mullin, Director
National Risk Management Research Laboratory
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
Dwledgements
We would like to acknowledge Ron Landy, ORD Regional Science Liaison to USEPA Region 3, for
administrative support for this project, Cindi White USEPA/National Analytical Radiation
Environmental Laboratory (NAREL), in Montgomery, Alabama for radionuclides analysis, and Cynthia
Caporale, for elemental analysis at the USEPA Region 3 Environmental Science Center in Fort Meade,
Maryland for sediment samples from Smith Creek, Virginia. We would also like to acknowledge
Amanda Wroble and George Schupp, USEPA Region 5 Chicago Regional Laboratory, for analyses of
trace elements and phosphorus in sediment samples from Otter Creek, Wisconsin in 2014-15. Timothy
Auer, USGS West Trenton Publishing Service Center, Baltimore, Maryland, and William Gibbs, USGS
Tacoma Publishing Service Center, Tacoma, Washington are thanked for illustration preparation.
Richard Brenner, USEPA, and Leah Kammel, USGS, provided technical reviews, and Marilyn Dapper,
NCOA/USEPA provided editorial assistance.
iv
-------
A Manual to Identify Sources of Fluvial Sediment
Table of Contents
EPA/600/R-16/210
Notice/Disclaimer ii
Foreword iii
Acknowledgements iv
Table of Contents v
List of Figures vii
List of Tables viii
List of Boxes ix
Acronyms and Abbreviations x
Executive Summary xi
1.0 Introduction and Background 1
2.0 Background on Sediment Budgets 4
2.1 Definition of a Sediment Budget 4
2.2 Major Sediment Sources and Sinks in a Watershed 8
2.2.1 Upland Sediment Sources and Sinks 10
2.2.2 Stream Corridor Sediment Sources and Sinks 10
3.0 Design of a Sediment Budget 15
3.1 Deciding On Sediment Budget Time Frames and Spatial Scales 17
3.1.1 Time Frame 17
3.1.2 Spatial Scale 18
3.2 Getting To Know Your Watershed 18
3.2.1 Existing Studies and Expert Knowledge 19
3.2.2 Watershed and Stream Network Delineation 19
3.2.3 Land Cover and Physiographic Setting 20
3.2.4 Hydrologic Alterations 20
3.3. Geomorphic Setting 20
3.3.1 Planform View 22
3.3.2 Longitudinal Profile 22
3.3.3 Lateral View 24
3.3.4 Watershed Reconnaissance 24
3.3.5 Stream Corridor Sediment Sources and Sinks 26
3.3.6 Upland and Hillslope Soil Erosion 30
3.4 Sediment Budget Design and Considerations 30
3.4.1 Sample size 35
v
-------
A Manual to Identify Sources of Fluvial Sediment EPA/600/R-16/210
3.4.2 Use of Sediment Inventories to Construct a Sediment Budget 36
3.4.3 Field Measurements in the Construction of a Sediment Budget 40
3.5.4 Upland Measurements 50
3.5.5 Age Determinations of Sediment Collected in the Field 54
3.5.6 Measuring Sediment Transport and Export 54
3.5.7 Budget Calculations 55
4.0 Sediment Fingerprinting 61
4.1 Tracer Selection 63
4.2 Target Sample Collection 63
4.2.1 Target Sample Preparation 66
4.3 Source Sample Collection 67
4.3.1 Source Sample Preparation 67
4.3.2 Grain Size - Laboratory Analysis 67
4.3.3 Organic Content 68
4.3.4. Field and Laboratory Quality Assurance 68
4.4.1 OutlierTest 70
4.4.2 Correcting Source Tracers for Sediment Size and Organic Content Variability 70
4.4.3 Bracket Test 70
4.4.4. Stepwise Discriminant Function Analysis 74
4.4.5 Computation of Source Percentages 74
4.4.6 Limitations and Uncertainty in Sediment Fingerprinting 75
4.5 Analyzing Results 76
4.5.1 Weighting Sediment Apportionment 76
5.0 Summary 81
References 89
vi
-------
A Manual to Identify Sources of Fluvial Sediment EPA/600/R-16/210
List of Figures
Figure 1. Causes of impairments in rivers and streams of the United States shown by percentage of impairments
(USEPA, 2016) 1
gure 2. Components in a sediment TMDL process 2
gure 3. Drainage Area vs. SDR (Walling, 1983) 5
gure 4. Watershed sources and sinks diagram 8
gure 5. Fish Creek historical sediment budget 11
gure 6. Pleasant Valley sediment budget 11
gure 7. Diagram of a streambank 12
gure 8. Historical overbank sedimentation rates example 13
gure 9. Example of variability in P concentrations in eroding streambanks in an agricultural lowland stream 14
gure 10. Steps in developing a sediment budget 16
gure 11. Dams and upstream sedimentation 20
gure 12. The fluvial system three zones: sediment source, transport, and deposition 21
gure 13. Longitudinal profiles and sediment 23
gure 14. Example recon form 25
gure 15. River walk-through example field form 28
gure 16. Bank erosion/lateral recession rates Wl NRCS technical guide 29
gure 17. Relation of bed material flux and bar area (O'Connor et al., 2014) 30
gure 18. Sediment size distribution in suspended sediment 33
gure 19. Sediment budget by inventory example 38
gure 20. Elements of a reach 41
gure 21. Diagram of meandering channel and terrace line 43
gure 22. Example of selecting upland areas 51
gure 23. Examples of field approaches to monitor upland erosion 52
gure 24. Illustration of cesium-137 sampling technique 53
gure 25. Outline of the sediment fingerprinting approach 61
gure 26. Steps in sediment fingerprinting 63
gure 27. Photos of auto samplers 65
gure 28. Summary of statistical operations used in sediment fingerprinting to apportion sediment sources. 70
gure 29. Example of how the size-correction factor is applied to a source group 71
Cover Photos: Linganore Creek, Maryland, Photos by Allen Gellis, USGS.
vii
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
II.
Table 1. Literature review on sediment budget studies completed over a range of spatial and temporal scales
with errors reported for each study (Modified from Gellis et al., 2012) 7
Table 2. Methods used in sediment budget analysis 9
Table 3. Geomorphic features identified on a river walk-through 26
Table 4. Standard particle size classes and size ranges used by the U.S. Geological Survey (Lane, 1947; Colby
1963) 34
Table 5. Ranges in dry bulk density for submerged and aerated sediment 36
Table 6. Rules of how uncertainties are propagated 60
Table 7A. Example of tracers that have been used in sediment fingerprinting studies. (See Miller et al2015, for
a more extensive list of tracers used.) 62
Table 7B. Example of 38 elemental metals used in elemental analysis for fingerprinting sediment (Gellis et al.,
2015) 62
Table 8. Equations used to correct for bias in untransforming the tracers (Gellis et al., 2015) 73
viii
-------
A Manual to Identify Sources of Fluvial Sediment EPA/600/R-16/210
List of Boxes
Box 1. Example of determining the number of sites to monitor in a study 36
Box 2. Example how reaches were selected for streambank monitoring in Linganore Creek, MD (Gellis et ai,
2015) 42
Box 3. Example of rasterizing the stream network to choose reaches for monitoring 42
Box 4. Example of how erosion and deposition rates are determined for a bank face 44
Box 5. Example of determining streambank change where pins are placed at unequal distances on the bank
face 45
Box 6. Example of how to convert streambank measurements at the reach scale (cm) to a mass (Mg) for first
order streams 46
Box 7. Example of how bar lengths are estimated for a given stream order 48
Box 8. Computation of the sediment budget 58 - 60
Box 9. Example of weighting sediment fingerprinting results by the load computed for each sample 78
Box 10. Example of weighing sediment fingerprinting results by the sediment load for each sampled storm
event 79
Box 11. Combining sediment budget and Sediment Fingerprinting results 85-88
ix
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
\ iHri- iiis .>tt- < .'.Lf-it', roans
3-D
Three-dimensional
AC
Acre
ADS
ADS Environmental Services
ASTL
Acoustic Sensing Technology, Ltd. (United Kingdom)
ATV
All-Terrain Vehicle
BC
Brown and Caldwell
CCTV
Closed-Circuit Television
CIP
Cast Iron Pipe
DE
Dissipation Energy
DEM
Digital Elevation Model
DFA
Discriminant Function Analysis
GIS
Geographic Information System
GPS
Global Positioning System
LiDAR
Light Detection and Ranging
QAPP
Quality Assurance Project Plan
RPD
Relative Percent Difference
SFM
Sediment Fingerprinting Model
TMDL
Total Maximum Daily Load
TOC
Total Organic Carbon
USACE
U.S. Army Corps of Engineers
USD A
U.S. Department of Agriculture
USEPA
U.S. Environmental Protection Agency
USGS
U.S. Geological Survey
X
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
I icitlh-' 'f Hr I IF I iir t i 4''
Sediment is an important pollutant of concern that can degrade and alter aquatic habitat. A sediment
budget is an accounting of the sources, storage, and export of sediment over a defined spatial and
temporal scale. This manual focuses on field approaches to estimate a sediment budget. We also
highlight the sediment fingerprinting approach to attribute sediment to different watershed sources.
Determining the sources and sinks of sediment is important in developing strategies to reduce sediment
loads to water bodies impaired by sediment. Therefore, this manual can be used when developing a
sediment TMDL requiring identification of sediment sources.
The manual takes the user through the seven necessary steps to construct a sediment budget:
1. Decision-making for watershed scale and time period of interest
2. Familiarization with the watershed by conducting a literature review, compiling background information
and maps relevant to study questions, conducting a reconnaissance of the watershed
3. Developing partnerships with landowners and jurisdictions
4. Characterization of watershed geomorphic setting
5. Development of a sediment budget design
6. Data collection
7. Interpretation and construction of the sediment budget
8. Generating products (maps, reports, and presentations) to communicate findings.
Sediment budget construction begins with examining the question(s) being asked and whether a
sediment budget is necessary to answer these question(s). If undertaking a sediment budget analysis is a
viable option, the next step is to define the spatial scale of the watershed and the time scale needed to
answer the question(s). Of course, we understand that monetary constraints play a big role in any
decision.
Early in the sediment budget development process, we suggest getting to know your watershed by
conducting a reconnaissance and meeting with local stakeholders. The reconnaissance aids in
understanding the geomorphic setting of the watershed and potential sources of sediment. Identifying
the potential sediment sources early in the design of the sediment budget will help later in deciding
which tools are necessary to monitor erosion and/or deposition at these sources. Tools can range from
rapid inventories to estimate the sediment budget or quantifying sediment erosion, deposition, and
export through more rigorous field monitoring. In either approach, data are gathered and erosion and
deposition calculations are determined and compared to the sediment export with a description of the
error uncertainty. Findings are presented to local stakeholders and management officials.
Sediment fingerprinting is a technique that apportions the sources of fine-grained sediment in a
watershed using tracers or fingerprints. Due to different geologic and anthropogenic histories, the
chemical and physical properties of sediment in a watershed may vary and often represent a unique
signature (or fingerprint) for each source within the watershed. Fluvial sediment samples (the target
sediment) are also collected and exhibit a composite of the source properties that can be apportioned
through various statistical techniques. Using an unmixing-model and error analysis, the final
apportioned sediment is determined.
xi
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
1.0 Introduction and Background
Sediment is an important pollutant that can lead to loss of stream-biologic integrity, whether in
suspension in the water column or as deposition on a stream or lake bottom (Waters, 1995). In a
summary of stream impairments for the United States compiled from state reports from 2006 to 2014,
sediment and turbidity was listed as the major source of stream impairment (USEPA, 2016) (Fig. 1). Of
particular concern are fine-grained silts and clays, which can degrade habitat, clog water supply intakes
and fill reservoirs, and often carry phosphorus and/or contaminants harmful to humans and aquatic life
(Waters, 1995; Larsen et al., 2010). Sediment impaired water bodies, usually identified by fair to poor
macroinvertebrate index scores, are placed on the 303D list where a sediment Total Maximum Daily
Load (TMDL) is implemented under the Clean Water Act (USEPA, 1999). A sediment TMDL is the
maximum amount of sediment a water body can contain and still meet its water quality standards and
beneficial uses. When a stream is identified as impaired by sediment, it is required in the TMDL
framework to identify sediment sources (USEPA, 1999) (Fig. 2).
0 50000 100000 150000 200000 250000
River and Stream miles
Figure 1. Causes of impairments in rivers and streams of the United States shown by percentage of
impairments (USEPA, 2016).
1
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
Allocate
loads
Develop
monitoring plan
Develop numeric
targets
Link targets and
sources
Identify problem
Source
Develop
implementation plan
Figure 2. Components in a sediment TMDL process.
The TMDL Technical Advisory Group (TAG), a group composed of scientists from universities, Federal
and state agencies, and non-governmental organizations, made the following recommendations in a 2002
review of sediment TMDLs in Georgia (USEPA Region 4) :
Develop a carefully crafted inventory of the potential sediment sources and pathways by which
sediment enters the water body.
Use currently available information, including water quality monitoring data, watershed
analyses, information from the public, and any existing watershed studies.
Conduct thorough onsite watershed surveys that help determine the relative contribution of
sediment from various sources.
Conduct follow up monitoring with special emphasis for Phase I TMDLs.
The objectives of the sediment-source assessment should be to characterize the types, magnitudes, and
locations of the source(s) of sediment loading by compiling an inventory of all potential sources through
identification on maps, existing data, and field surveys (USEPA, 1999). Monitoring, statistical analyses,
and modeling are recommended in order to determine the relative magnitude of sediment-source
loadings and watershed-delivery processes (USEPA, 1999). Understanding the role of stream-related
fluvial processes in transporting sediments from watershed sources, delivery, and storage is a key focal
point of this manual. Understanding fluvial processes is especially useful for determining the relative
magnitude of sources from upland soil erosion compared to fluvial erosion and river-related mass
wasting. It may also add insights as to the relative sources of particulate-phase phosphorus sources,
transport, and storage within channels and near channel areas. Human activity, such as construction and
urbanization, can alter hydrology and runoff, which can lead to increased rates of fluvial erosion and
river-related mass wasting (Wilkinson and McElroy, 2007; Fitzpatrick et al., 1999). In the process of
identifying specific sources of sediment and the magnitude of the problem, the results can be used to
develop an implementation plan based on the proximity of active sediment sources to important areas
within a river system, such as spawning beds, water intakes, and drinking water reservoirs (USEPA,
1999). Calculations of sediment loading from specific sources can help determine if those loadings
differ from natural or background rates.
2
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
It is of utmost importance to determine, measure, and monitor sediment loads and identify the sources
causing a sediment or nutrient problem early on in the TMDL process (Fig. 1) (USEPA, 1999). The
U.S. Geological Survey (USGS), in coordination with many other Federal, state, and local agencies,
monitors suspended-sediment loads and concentrations at watershed outlets to evaluate the success of
land use conservation practices on reducing sediment and nutrients in impaired watersheds (Shipp and
Cordy, 2002; Jastram, 2014). Monitoring sediment loads is an important step in the TMDL process but
for management purposes it is important to identify the sources for the loading. Currently, jurisdictions
across the United States use a variety of approaches to identify sediment sources (Williamson, et al.,
2014). These approaches focus on identifying watershed sources usually related to soil erosion and
monitoring suspended sediment concentration and loads at watershed outlets.
Soil erosion and sedimentation are priority problems being addressed through additional programs
besides the TMDL process by other Federal agencies including the U.S. Department of Agriculture
(NRCS, 2007), the U.S. Army Corps of Engineers and International Joint Commission (Reidel el al,
2010; Hayter et al., 2014), and the Bureau of Reclamation (Bureau of Reclamation, 2007). Several large
regional initiatives are in place that include sediment reduction goals, including the Chesapeake Bay
Restoration Initiative, the Great Lakes Sediment and Nutrient Reduction Program
(http://keepingitontheland.net/projects-glri/), the Upper Mississippi River Healthy Watersheds Initiative
(http://www.nrcs.usda.gov/wps/portal/nrcs/detailfull/national/programs/initiatives/?cid=nrcsdevll_0238
96), and the Gulf of Mexico Hypoxia Task Force (http://www2.epa.gov/ms-htf/hypoxia-task-force-
nutrient-reduction-strategies). Agricultural watersheds are of special interest because of the strong
relation between sediment and particulate-bound phosphorus. In addition to TMDL development, the
approaches presented in this manual may be useful to these other Federal programs for sediment
management.
Examination of several sediment TMDL reports produced by jurisdictions throughout the United States
indicated a reliance on models, Geographic Information System (GIS) analysis, and best judgment to
identify sediment sources in the TMDL framework. These reports did not include a field-based
approach to identify and target sediment sources. The practitioner charged with reducing riverine
sediment loads through identifying sediment sources should have a variety of tools at his or her disposal.
The objective of this manual is to provide practitioners with approaches to identify the important sources
of sediment in a watershed and budget the erosion, storage, and delivery of this sediment. The manual
emphasizes the sediment budget and sediment fingerprinting approaches and discusses the benefit of
combining both approaches. This manual does not provide in-depth descriptions of the many models
that are used in agricultural upland sediment-source assessment. Information on models that are used in
sediment-source analysis can be found in Chapter 8 of a USEPA (2008) report on non-point source
pollution. This manual discusses field- based approaches that may rely on or benefit from
photogrammetric methods, GIS, and models, to identify sediment sources and budget sediment- with
special emphasis on techniques for measuring sources and sinks within the stream corridor. The
techniques are divided into major sections; first, describing how to construct a sediment budget and
second, how to design and sample for a sediment fingerprinting study. It is the a goal of this manual to
educate practitioners on field-based approaches using sediment budget and sediment fingerprinting
approaches as tools to identify sediment sources in the sediment TMDL process. This manual expands
on the methods introduced in Chapter 5 of the sediment TMDL protocol (USEPA, 1999) by further
describing how an integrated sediment budget and sediment fingerprinting approach complement
existing techniques. The approaches presented in this manual are largely field based and are presented
not to eliminate current approaches but to be used in conjunction with existing approaches. The
sediment fingerprinting approach is one of the tools that we highlight in this manual.
3
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
dgets
Sediment is composed of inorganic and organic particulate material that is transported by a stream in
suspension and as bedload. Both suspended sediment and bedload can lead to the impairment of streams
and receiving water bodies. This manual primarily focuses on identification of the sources and flux of
suspended sediment specifically with respect to the fine-grained fraction (<0.063 mm). However, many
of the approaches suggested in this manual for measuring erosion and deposition in stream corridors can
be applied to bedload and total sediment loads.
The sediment fingerprinting approach apportions the relative contribution of the potential sediment
sources in a watershed delivered to a point in the watershed. The sediment budget approach provides
information on the magnitude of the fluxes and the links between sources, storage, and sediment output.
Combining the two approaches can provide resource managers with information on where to target
measures to reduce erosion, sediment delivery, and the net transport of sediment.
i IE11" i fit i f11 >n ',>1 ' iis Pudget
A sediment budget is an accounting framework that can be used to understand processes of sediment
erosion, transport, storage, delivery, and linkages among these elements that occur in a watershed
(Leopold et al., 1966; Swanson et al., 1982; Gellis et al., 2012). The most basic form of the sediment
budget equation is:
I±AS = 0 (1)
where: / = the sediment input
AS = the amount in sediment storage, and
O = the sediment output
In general, the units in equation (I, S, O) correspond to sediment mass overtime (i.e., kg/yr), although
volumes can also be used in a sediment budget (m3/yr). Because sediment is transported episodically
during large floods, the time scale of reference for each source and sink component is extremely
important. Thus, whether sediment is budgeted over a single storm, to years, to decades is important.
Furthermore, the measurements used to quantify erosion and deposition (/, S) can be linear (m), cross
sectional (m2), or volumetric (m3). For source apportionment, a volumetric or mass rate is needed for all
components.
This manual will provide a review of measurements available to quantify O, I, and S and how to use
these measurements to construct a sediment budget. Often measurements are made in small areas (i.e., a
point, cross section, or reach), and extrapolation to the entire watershed is needed to construct the
sediment budget. For example, to budget channel changes, bank measurements can be made with pins,
i.e., linear (cm) measurements that are averaged over the bank face to get a cross sectional area change
(m2). The cross sectional area change is extrapolated to a selected stream length (m) to calculate a
volume (m3). The volume of change in streambanks is converted to a mass by multiplying the
volumetric change (m3) by the density of bank material (g/cm3) to determine a mass (kg).
4
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
In the construction of a sediment budget, the terms 'gross' erosion and 'net' delivery are used to
describe sediment eroded from an area of interest, which can range from the plot scale to the entire
watershed, that is delivered to a site further downstream (de Vente et al., 2007). The sediment delivery
ratio (SDR) is the ratio of delivered sediment, expressed as a yield per unit area, divided by the gross
erosion, usually expressed as a percent:
In most cases, gross erosion is greater than the sediment yield, the difference being due to sediment that
goes into storage (S in Eq. 1) as it is transported through the watershed. Although gross erosion in the
literature usually refers to upland erosion, it can and should include fluvial erosion (de Vente et al.,
2007). SDRs estimated from sediment budgets can range from zero to 100 % (Smith and Dragovich,
2008; Walling and Collins, 2008). Walling (1983) plotted contributing area versus the SDR and noted
that as drainage area increases, the SDR decreases, reflecting the increase in storage areas with
increasing area (Fig. 3). Roehl (1962) depicted the SDR as an area power law relation:
Sediment Yield (sediment delivered to the point of interest) (v-^/yr
Gross erosion
SDR = aAp
(3)
where:
A is the basin area (km2), a is a constant, and (3 is a scaling function (0.01 to -0.25).
100
E
1
0,01 0.1 1 10 100 1000
Drainage area, km2
Figure 3. Drainage area plotted against the sediment delivery ratio (after Walling,
1983).
5
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
The differences in gross erosion and delivered sediment are important to the land manager. If
knowledge of watershed soil loss or erosion directly at a site is needed, then measurements of gross
erosion may suffice. If knowledge of sediment delivery to a point in the watershed or to lakes, estuaries
and the ocean is needed, then constructing a sediment budget and determining sediment delivery is
necessary. Sediment delivery is also needed for identifying possible lag times between upland best
management practices and downstream stream water quality (Meals et al., 2010; Sharpley el al, 2013).
For example, in large watersheds, efforts to reduce sediment may take less than 10 to more than 50 years
to produce measurable differences at the watershed outlet (Meals et al., 2010).
Sediment budgets can be used for a variety of purposes:
• Sediment source identification for pollutant purposes - TMDLs
• Assess the effects of land use practices - i.e., agriculture
• Monitor the effectiveness of management actions to reduce sediment - i.e., stream restoration
• Determine sediment contributions from natural factors, e.g., erosion following a wildfire or the
contributions from landslides after a major storm
• Determine the long-term effect of stressors - i.e., dams or climate change
• Put current sediment budget results in the context of historical sediment budget rates
• Determine the input from tributaries
• Determine how different geologic areas contribute sediment
Sediment budgets have been performed on all continents, at varying spatial and temporal scales, and
using a variety of techniques (Table 1).
6
-------
A Manual to Identify Sources of Fluvial Sediment EPA/600/R-16/210
Table 1. Literature review on sediment budget studies completed over a range of spatial and temporal scales with errors reported for each study (Modified
from Gellis etai, 2012).
Continent
Spatial scales
Time scales
Methods
North America (Leopold et al.,
1966; Costa, 1975; Dietrich and
Dunne, 1978; Trimble, 1983;
Knox, 1985; James, 1989;
Phillips, 1991; Sutherland, 1991;
Beach, 1994; Faulkner and
Mclntyre, 1996; Knox, 2002;
Gaugush, 2004; Allmendinger et
al., 2007; Renwick, et al. 2005;
Fitzpatrick et al., 1999; 2009;
2015)
South America (Meade et al.,
1990; Trauth et al., 2003)
Europe (Macaire et al., 2002;
Gruszowski et al., 2003; Belyaev
et al., 2005; Evans and
Warburton., 2005; Houben et al.,
2006; Rommens et al., 2006; Van
der Perk and Jetten, 2006)
Asia (Schick and Lekach, 1993;
Oguchi, 1997)
Africa (Dunne, 1979; Sutherland
and Bryan; 1991; Wijdenes, and
Bryan, 2001; Walling et al., 2003;
Garcin et al., 2005),
Australia (Loughran et al., 1992;
Brizga and Finlayson, 1994; Page
et al., 1994; Wasson et al., 1998;
Wallbrink et al., 2002)
Antarctica (Pollard and DeConto,
2003)
m2 (Brunton and Bryan,
2000)
ha (Sutherland, 1991;
Wijdenes and Bryan,
2001; Wallbrink et al.,
2002; Polyakov et al.,
2004; Evans and
Warburton., 2005; Hart
and Schurger, 2005)
10 - 100's km2 (Trimble,
1983; Knox, 1985; James,
1989; Beach, 1994;
Faulkner and Mclntyre,
1996; Oguchi, 1997; Knox,
2002; Slaymaker et al.,
2003; Walling et al., 2003;
Garcin et al., 2005
Fitzpatrick et al., 1999;
2009; 2015)
>1,000 km2 (Costa, 1975;
Meade et al., 1990;
Phillips, 1991; Brizga and
Finlayson, 1994; Gaugush,
2004; Renwick et al.,
2005; Houben et al., 2006)
days (Page et al., 1994;
Springer et al. 2001; Van
der Perk and Jetten, 2006)
months (Sutherland and
Bryan, 1991; Polyakov et
al., 2004; Belyaev et al.,
2005)
years (Leopold et al.,
1966; Schick and
Lekach,1993; Phillips,
1991; Gruszowski et al.,
2003; Gaugush, 2004;
Renwick, et al. 2005;
Rovira et al., 2005;
Fitzpatrick et al., 2009;
2015)
centuries (Costa, 1975;
Trimble, 1983; Knox,
1985; James, 1989;
Beach, 1994; Faulkerand
Mclntyre, 1996; Wasson et
al., 1998; Knox, 2002;
Fitzpatrick et al., 1999;
2009; 2015)
millennia (Knox, 1985;
Oguchi, 1997; Macaire et
al., 2002; Slaymaker et al.,
2003; Houben et al., 2006;
Rommens et al., 2006)
field measurements (Leopold et al., 1966;
Costa, 1975; Knox, 1985; James, 1989;
Phillips, 1991; Sutherland and Bryan, 1991;
Beach, 1994; Faulkner and Mclntyre, 1996;
Knox, 2002;Gaugush, 2004; Evans and
Warburton, 2005; Rovira et al., 2005;
Fitzpatrick et al., 1999; 2009; 2015)
radionuclides (Ritchie et al., 1974;
Sutherland, 1991; Wallbrink et al., 2002;
Walling et al., 2003; Fitzpatrick et al., 2009)
multiple geochemical fingerprints (Walling
and Woodward, 1992; Nimz, 1998; Wasson et
al., 2002; Gruszowski et al., 2003; Walling,
2005)
pond and lake sedimentation (Foster et al.,
1988; Erskine et al., 2002; Phippen and Wohl,
2003; Renwick, et al. 2005)
sediment cores (Costa, 1975; Knox, 1985;
2002; James, 1989; Beach, 1994; Faulkner
and Mclntyre, 1996; Slaymaker et al., 2003;
Belyaev et al., 2005; Houben et al., 2006;
Rommens et al., 2006; Fitzpatrick et al., 1999;
2009; 2015),
models (Phillips, 1991; Belyaev et al., 2005;
Renwick, et al. 2005)
maps and photogrammetry (James, 1989;
Brizga and Finlayson, 1994; Faulkner and
Mclntyre, 1996; Wasson et al., 1998; Garcin et
al., 2005; Renwick, et al. 2005; Fitzpatrick et
al., 1999; 2009)
7
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
A substantial number of tools and approaches used to measure or estimate /, S, and O in Eq. 1 are
available (Table 2) and selection will depend upon many factors, such as financial resources and
temporal and spatial aspects of the study. We highlight a few approaches in this manual that are
favorites among the authors and that have worked well in USEPA Regions 3 and 5. References
provided or searches on the Internet can be used to identify other approaches. It is also anticipated that
many of these approaches will change in the future due to technological advances.
2.2 Major Sediment Sources and Sinks in a Watershed
In general, watershed sediment sources can be separated into two broad categories based on their origin:
1) uplands and hillslopes, and 2) stream corridors (Gellis and Walling, 2011). Upland sediment sources
most often include soil erosion from various land use and land cover types, such as forest, cropland,
pasture, construction sites, and roads (Fig. 4). Stream corridor sources include streambanks and channel
beds. Also included in stream corridor sources is sediment derived from mass wasting where channels
intersect valley sides and terrace walls. Gullies span the two sources but are usually included as channel
sources. Hillslope erosion is usually included in upland erosion. Floodplains and alluvial fans are
usually sediment sinks, but can become sources during large floods. Differentiating between these two
broad categories (upland and channel sources) is important because sediment-reduction management
strategies differ by source and require very different approaches — reducing agricultural sources may
involve soil conservation and tilling practices, whereas reducing channel sources of sediment may
involve stream restoration, bank stabilization, and grade control to arrest downcutting.
Colluvial elusion and
soil creep
^ Terrace scarp
X- " mmh
Gaily— •
stream* \
Ro A«"-'rai
field
, Point)
bar
Mid-
dnoM)
Figure 4. Watershed sources and sinks diagram.
It is important to note that sediment sources and sinks can vary by location in the watershed and by
season, and, therefore, the rates and the time scale of interest for the sediment budget become important.
For example, in the agricultural Midwestern United States, soil erosion is most prevalent during the
spring months when large areas of soils are bare, compared to late summer when crops thickly cover any
bare soil. Streambank erosion can often be greater in winter months, if the location undergoes freeze
8
-------
A Manual to Identify Sources of Fluvial Sediment EPA/600/R-16/210
thaw cycles (Wolman, 1959).
Table 2. Methods used in sediment budget analysis. [NA = not applicable]
Sediment
budget
element (I
= input, S
= storage,
0 =
output)
Method
Can be used to quantify
Dimensions
measured
Time scale of
measurements
Channels (1)
Aerial
photography &
LiDAR
Channel changes in width, depth,
sinuosity, bar formation, and channel
pattern
m, m2,m3
Years, decades
Channels (1)
Bank pins
Bank erosion and deposition
m, m2,m3
Days (individual
storms) to years
Channels (1,
S)
Rapid
geomorphic
assessments
Qualitative condition of the
erosional/depositional characteristic of
the streambanks
NA
NA
Channels (l,S)
Scour chains
Quantify change in the channel bed
m, m2
Days to years
Channels (l,S)
Surveys
Changes in channel width, depth, and
slope
m, m2, m3
Per storm (days)
to years
Channels (S)
Stratigraphy
Identification of time horizons
(anthropogenic, geologic, Radionuclides),
to estimate changes in deposition rates
NA
Years, decades,
millennia
Deposition
(S)
Floodplain pads
Floodplain deposition rates
m, m2, m3
Per storm (days)
to years
Sediment
transport (0)
Collection of
suspended
sediment and
bedload
Sediment transport and loads
kg
Per storm (days),
years to decades
Uplands (1)
Aerial
photographs
Qualitative description of areas that may
contribute sediment (agriculture, mining,
landslides, roads, etc.)
NA
NA
Uplands (1)
Aeolian dust
traps
Quantify eolian deposition
g
Years
Uplands (1)
Sediment traps
and nets
Sediment yield in contributing area
Kg
Per storm (days)
to years
Uplands (1)
and channels
(1)
Sediment
fingerprinting
Quantify the contribution of sediment
from source areas
%
Days to years
Uplands (l,S)
137Cs
Upland erosion and deposition rates
tons/hectare
Decades (50
years)
Uplands (l,S)
Dendrochronol
ogy
Coring trees and counting rings to
determine deposition and erosion rates
cm
Decades
Uplands (l,S)
Pins, erosion
bridges
land surface erosion, unpaved roads
mm
Per storm (days)
to years
Uplands
(l,S,0)
Lake/pond
bathymetric
surveys
Sediment loads and sediment yield in
contributing area, changes in
sedimentation overtime
kg
Years, decades
9
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
2.2.1 Upland Sediment Sources and Sinks
Upland sediment sources are those that occur outside the stream corridor and involve mainly soil erosion
on varying land use and land covers, as well as more infrequent events such as mass wasting
(landslides), and erosion from areas affected by fire. Erosion on upland surfaces occurs through
sheetwash, overland flow, rilling, and gullying (USDA, 2007). Upland areas have sediment sinks in flat
areas at the base of slopes that are not bisected by channels. This may happen at the grassy fenced edge
of a field or at the base of a hillslope.
The time of year is important with respect to sediment sources. Cropland may become an important
sediment source during tilling in the spring and harvesting in the fall when large areas are cleared of
vegetation. Tillage operations and the type of farming practices used may also affect sediment sources.
No-till operations, for example, which can have greater percentages of vegetative or residue cover
throughout the year can also reduce soil erosion by varying amounts (Huggins and Reganold, 2008).
2.2.2 Stream Corridor Sediment Sources and Sinks
Stream corridor sources include channel sediment that is directly eroded and transported by flowing
water. Most commonly this is thought of as streambank erosion but may also be derived from incising
channel beds or mass wasting where a channel intersects a valley side and terrace wall (Fig. 4). In many
studies, it is assumed that the channel bed is not a source or a sink along a main stem because any
deposition of sediment is thought to be temporary and originating from upstream sources and is,
therefore, not treated as a separate source (Gellis et al., 2009). However, in agricultural lowland
streams, many of which are impaired by sediment, there is significant storage of sediment in channels
and adjacent floodplains; thus, storage in the channel bed cannot be assumed to be negligible (Figs. 5,
6). For example, in Pleasant Valley, a small 19 km2 agricultural lowland stream on the Wisconsin
impaired waters list, fine-grained soft sediment stored along the channel bed is estimated to be
equivalent to 8 years' worth of annual loading exported from the watershed (Fig. 6). In actively eroding
streams, such as arroyos or incised channels (Gellis, 2012; Fitzpatrick et al., 1999), the bed of the stream
may be an important sediment source. In a steep forested watershed in the northern Great Lakes region,
channel incision was about equal to floodplain deposition on average over a multi-decadal time scale
(Fig. 5).
10
-------
A Manual to Identify Sources of Fluvial Sediment EPA/600/R-16/210
Streambank
uoo Gully
Upland \ 910
1,100 x / Channel
\ / /4.100
Bluff
15,000
Chequamegon Bay
15,900
Floodplain
4,940
Channel
1,570
Figure 5. Example of a historical sediment budget; in tonnes per year;
for a steep forested watershed in the upper Great Lakes (Fitzpatrick et
at., 1999; Fitzpatrick and Knox, 2000). Bluff erosion was identified as a
major source of sediment that was causing downstream sedimentation
problems in fish spawning riffles.
RUSLE2 SOIL LOSS
1.1
TONSAC/YR
WATSKSHE
EXPORT
0.15
TONS/ACYR
BANK EROSION
0.06
TONyAcnrR
FINE SEDIMENT !L
SAVINGS AND LOAN
1.2
TONS/AC
Figure 6. Example of a
sediment budget for
2007-10 from Pleasant
Valley, an agricultural
lowland stream on
Wisconsin's impaired
streams list. The sources,
export, and sinks are
from different methods -
upland soil loss from
RUSLE2, watershed
export from a USGS
monitoring station, bank
erosion from a 2009
11
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
Streambanks can be an important sediment source and erode by three mechanisms: 1) freeze-thaw
processes (Wolman 1959; Wynn 2006); 2) fluvial erosion (Julian and Torres 2006; Wynn 2006); and 3)
mass wasting (Darby etal, 2007; Wynn, 2006). Freeze-thaw action of the bank surfaces causes the soil
to expand and loosen. The material that is loosened is readily available for transport by a range of flows
that inundate the bank surface (Lawler, 1986; Wolman 1959). Fluvial erosion is the detachment,
entrainment, and removal of particles or aggregates from the streambank by the hydraulic forces of
water. Flydraulic forces are related to the shear stress that the flow exerts on the bank. Sediment grain
size, cohesiveness of grains, and vegetation are also important in whether streambanks are erodible
(Wynn, 2006).
The composition of streambanks may be highly variable depending on whether the stream is cutting
through floodplain deposits, older terrace fills, or valley sides of potentially glacial deposits, colluvium,
or bedrock. Floodplain deposits can also range from coarse to fine-grained, depending on whether the
stream is cutting through fine-grained over-bank deposits or coarse-grained older channel bar and bed
deposits. Streams construct banks through the two main processes of 1) overbank deposition and 2)
point bar formation (Fig. 7); both can be important in constructing floodplains (Wolman and Leopold,
1957; Moody, etal., 1999). Actively eroding terrace cuts and valley sides can be major sources of
sediment (Fig. 4), especially in middle main stems where the meander wavelength of a stream
approaches the valley width. Because their height can be quite substantial compared to floodplain
elevations, they can be major sources of sediment to downstream reaches (Fitzpatrick and Knox, 2000).
(A)
Overbank
sediment
Terrace
currrent£2l2
Main
Point bar
Bankful
Cross
section
Overbank
sediment
Terrace
Cutbank
Figure 7. Diagram of (A) channel
meander bend looking from
above. (B) Cross section (X-
X')though a meander bend
showing cut bank, point bar and
overbank sediment deposits, and
terrace.
Active channel
streambank
Terrace (cutbank)
_ Channel
~ migration
Bankful
Overbank
deposits
Terrace
scarp
X'
Point bar deposits
Channel deposits
(B)
Valley fi
12
-------
A Manual to Identify Sources of Fluvial Sediment EPA/600/R-16/210
The age of sediment in streambanks may vary considerably in the vertical direction depending on the
origin of the deposits. A good example of differing ages of the active streambanks is in the Mid-
Atlantic region of the United States where the lower portion of the banks is typically composed of older
geologic-aged material (in thousands of years) underlying a younger, historical sediment deposit (100's
of years) (Jacobson and Coleman, 1986; Donovan etal., 2015). The historical sediment deposit is often
related to European colonization of the Mid-Atlantic region where trees were cleared for agriculture and
soil erosion and gullying of the cleared areas delivered sediment to the channel where it was deposited
on the floodplain on top of older deposits. This historic sediment is often referred to as 'legacy sediment'
(Jacobson and Coleman, 1986) and has resulted in thick accumulations of sediment in river bottomlands
and floodplains across the United States (Lowdermilk, 1934; Happ et aL, 1940; Leopold, 1956; Knox,
1972, 1987, 2006; Trimble, 1974; Beach, 1994; Faulkner, 1998; Montgomery, 2007) (Fig. 8). In many
parts of the United States, the historical sediment was deposited in mill ponds that were later breached
and incised by the channel to form the modern streambanks (Walter and Merritts, 2008). Much of the
sediment eroded during the 18th, 19th, and 20th centuries still remains stored in floodplains, along channel
margins, or in former mill ponds (Costa, 1975; Knox, 1972, 1977, 1986; Magilligan, 1985; Trimble,
1974; Fitzpatrick et al., 2009; Walling and Owens, 2002; Allrnendinger et al.. 2007). Some of the
historical sediment actively eroding in streambanks has higher phosphorus concentrations compared to
prehistoric counterparts, especially if a farmstead with livestock was located nearby (Fig. 9).
2S.OOO
20.000
11.000
10.900
S.OOO
0
Mississippi Rrver Locfc end Dam No. 7 closed 193?
Com filUla
Wfieflr mmm—m—mm
Agriculture
18W
1900 1927
1W6 1176 I8» I9Q2 W3 1925 '929
J » Ui I Ml
Soybeans „
Doc umenied floods
1993
_afiI iJL
Rb Iroads corsmicted
Chiraqoan* Bu-lnqtonKiiittBrr
Nontawswrn Aiiiroaf Santa P© Fsairc>»i
Eufo-Arnefican
sctHi-nunl
Soil Ccnsevstion
JlfcPl
Practices
Hohien Dam
Halfway Crwk Ovgrbarik
Sedimentation Hates
*3| iTormes/yri
IW 1850 I KG 1870 IBM 1890 1JCG 9IC 1920 1333 19*0 1950 1360 1370 1960 1990 2000 2010
Vaar
Figure 8. Historical overbank sedimentation rates for the mouth of Halfway Creek, a tributary to the
upper Mississippi River. Sedimentation rates peaked in the 1920s and 1930s and have decreased
since the adoption of soil conservation practices (Fitzpatrick et al., 2007).
13
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
Phosphorus in Eroding Bank Sediment
¦ Grass a Pasture • Woods
!& Grazed/feeding area
2.0
^ 1-5
3
=• 1.0 A ¦ ¦
& %
0.5 A
0.0
0 5 10 15 20
Drainage Area (miA2)
Figure 9. Concentrations of phosphorus in eroding streambank sediment in a
lowland agricultural stream in the upper Mississippi River basin, Pleasant Valley,
2009. Most sediment had concentrations between 0.5 and 1.0 mg/g, except for a
heavily grazed pasture with a feeding station near the stream.
Gullies fit in a category between upland erosion and channel erosion and can be important sources of
sediment that propagate on upland surfaces in urban, agricultural, and forest settings (Poesen et al,
2003). Gullies grow and extend through several processes related to a variety of mechanisms:
groundwater sapping, base-level lowering, downstream channel incision, and increased runoff (NRCS,
2007). In some cases, separating gullies from channel sources can be difficult where the difference may
be related to the observation that gullies can evolve quickly from incisional to depositional states
(Starkel, 2011).
Mass wasting on streambanks is the failure of all or part of the bank as a result of geotechnical
instabilities. Mass failures can occur from fluvial erosion undercutting the toe of the streambanks and
creating unstable conditions leading to bank failure (Simon et al'., 2000). Bank failures and mass
wasting are common during the recessional period of stormflow when seepage forces overcome the
resistance of a grain's cohesion (Fox et al., 2007; Simon et at., 2000).
It is important to differenti ate between streambanks of the acti ve stream channel and fluvial terraces
(Fig. 7). When streams incise, the floodplain no longer receives flows at annual intervals and forms a
terrace. Many streams in the western United States, for example, have had several periods of incision or
down-cutting through the Quaternary, and a flight of terraces has formed (Haynes, 1968). Terraces may
only be inundated during extreme flow events. Streams may actively erode into terraces, especially in
meandering systems.
Channel incision may occur in almost any setting by a number of causes such as: change in downstream
base level, climate change, tectonics, channelization, or where flow or sediment have been altered,
causing the channel to incise. For example, in urban areas, increased flow from impervious areas has
14
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
caused many channels to incise. In the southern United States, many steams were channelized which
increased channel slope, leading to channel incision (Simon, 1989). Channelized streams are often deep
and the highest level, the former floodplain, is now a terrace, and a new floodplain is being constructed
at some distance below the terrace.
V* I -r *h;sr s'l INI 11-1 Isf I -.PI'! -r
When faced with examining sediment problems at the watershed scale, the practitioner has to decide
whether utilizing a sediment budget is a viable option for addressing the question at hand. We suggest a
seven-tiered framework to assist in the design and construction of a sediment budget (Fig. 10).
1. Decision-making for watershed scale and time period of interest
2. Familiarization with the watershed by conducting a literature review, compiling background
information and maps relevant to study questions, conducting a reconnaissance of the watershed, and
developing partnerships with landowners and jurisdictions
3. Characterization of watershed geomorphic setting
4. Development of a sediment budget design
5. Data collection
6. Interpretation and construction of the sediment budget
7. Generate products (maps, reports, and presentations) to communicate findings.
This tiered approach proceeds from the early stages of the sediment budget design where determining
the important questions and the time and spatial scales that are needed to address the sediment problem
are important (Step 1). Compiling existing information on erosion, transport, and deposition of
sediment in the watershed and getting to know the constituents of the watershed are important in the
early stages of sediment budget design (Step 2).
15
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
Is a sediment budget framework appropriate to addressing the problem?
Considerations
Get to know your watershed
Existing studies, reconnaissance, stakeholders
Watershed Scale
Time frame
Monetary budgets
Geomorphic Setting
- Identification of sediment sources and sinks
- Landscape connectiveness
Sediment Budget Design
Approach - Is a field, remote sensing, or modeling approach
used to answer the question
\
Field
\
Data Collection
Field (stream and upland)
Number of sites and reaches
Measurement type
Frequency of measurements
Synthesis of data
Remote Sensing
Modeling (see
Chapter 8 of USEPA,
(2008) report)
-INTEPRETING DATA
Determining geomorphic change
- Conversion to mass
- Extrapolation to other portions of the watershed
-Computation of the sediment budget
- Error analysis
REMOTE SENSING
-Time period of imagery
- Selection of sites for measurements
- Processing to quantify change
Products
Maps, reports, presentations
Figure 10. Steps in developing a sediment budget
16
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
' j II- cicfiifv;
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
3.1.2 Spatial Scale
The spatial scale and resolution of the sediment budget are also important in decisions related to time
scales and design. Important questions related to spatial scale include:
• What is the size of the watershed where the sediment budget should be performed?
• How many sediment transport monitoring stations can be established?
• A sediment monitoring station should be located at the watershed outlet, but are there other
scales in the watershed where monitoring would be of interest?
• Strategically placed sediment-transport monitoring station(s) can help identify sub-watersheds
or specific stream reaches that have high sediment loadings compared to the overall loading at
the watershed outlet. The spatial scale also affects the design of the measurements (placement
and type) and monitoring, where at larger watershed scales, remote sensing and aerial
photographic analyses play an important role in selecting reaches for field measurements and
monitoring.
" p ttiri. 1 *. fM i v 'p if tershed
Usually in the development of a TMDL, partnerships form among Federal, state, and local government
agencies as they come together to perform the steps outlined in Figure 1. From this partnership comes a
wealth of knowledge about the watershed and the sediment problems encountered as well as the
stakeholders involved. Especially knowledgeable for agricultural watersheds are USD A county
conservationists usually co-located with the local USDA-NRCS office. The USEPA (2008) handbook
for developing watershed plans to improve water quality is an excellent resource to learn how to develop
partnerships. This information can be especially useful in the early stages of designing your sediment
budget.
For both large and small watersheds, GIS is essential for displaying and analyzing relevant maps and
data. This step is also similar to the Reconnaissance Level Assessment (RLA) in the Watershed
Assessment of River Stability and Sediment Supply (WARSSS) (Rosgen, 2006). The RLA involves 15
important steps that fit under constructing sediment budgets:
1) Compile existing data
2) Review landscape history
3) Summarize activities that potentially affect sediment supply
4) Identify relations between sediment and geomorphic processes
5) Review the landscape and map the watershed
6) Identify hillslope processes
Document surface erosion
Document mass erosion
Assess hydrological processes
10) Identify streamflow changes
11) Analyze channel processes
12) Detect direct impacts to streambanks and channels
13) Summarize problem verification process - recognition of places, processes, and sources
14) Eliminate sub-watersheds or river reaches that are not sediment problems
15) Select sub-watersheds and reaches for further assessment.
18
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
The methods outlined below in more detail are those that we have found particularly helpful in the early
stages of designing sediment budgets.
3.2.1 Existing Studies and Exp< owledge
Many watersheds with sediment problems have been part of state priority watershed programs since the
1980s and 1990s, and many have had soil conservation practices implemented, some going back to the
1930s and 1940s with the Soil Conservation Service efforts. These programs had similar goals of
reducing and controlling sediment and nutrients. Other studies were geared toward fish habitat.
Sediment deposition, substrate particle size, and bank erosion are typically collected as part of state and
Federal habitat assessments and, thus, may be helpful in the early design of the sediment budget
(Simonson etal., 1993; Fitzpatrick et al., 1998; Kaufman et al., 1999).
3.2.2 Watershed and Stream Netw ilineation
In the sediment budget, an outline of the watershed boundary and rivers draining to the outlet need to be
delineated. Most often a GIS or web-based tools will be used for these delineations. A watershed is
defined as the area of land that drains water, sediment, and dissolved materials toward a common outlet.
The USEPA's Watershed Assessment, Tracking, and Environmental Results System (WATERS)
(http://www.epa.gov/waterdata/waters-watershed-assessment-tracking-environmental-results-system)
builds a stream network and its watershed based on the National Hydrography Dataset (NHD), the
National Elevation Dataset (NED), and the Watershed Boundary Dataset (WBD). The integration of
these three national datasets can provide the base layers of watershed boundaries and stream networks
needed to design a sediment budget study. For smaller watersheds, additional higher resolution datasets
may need to be found for delineating sub-watersheds and topographic relief. The basis for the NHD
stream network are streamlines on USGS 1:24,000-scale maps. These streamlines form the framework
for much of the watershed geomorphological analyses that have been done in the United States over the
past 50 years (Leopold, 2006; Fitzpatrick, 2016). The network of channels is likely more complex than
shown by the NHD, especially for ephemeral channels in headwaters. A representative stream network
is needed to be able to represent the stream lengths to which sediment sources and sinks are applied.
Some exceptions to be aware of when building a stream network and related watershed boundaries are
the following:
• In urban areas, storm sewers often transcend topographic boundaries and additional data on
storm-sewer networks are needed.
• In arid areas, water is often artificially routed through pipes or canals across topographic divides
into adjacent watersheds for storage or supply for agricultural and industrial uses and drinking
water supplies.
• Some watersheds contain depressions or closed basins with no surface water outlet, such as
kettle ponds in glaciated terrain, karst landscapes, and playas in the arid western United States.
These areas are referred to as noncontributing areas by the USGS, meaning they do not directly
contribute to surface water drainage.
19
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
3.2.3 Land Cover and Physiographic Setting
Identifying and mapping patterns of land cover and physiographic setting are important for identifying
possible areas prone to soil erosion. The patterns can also be used to stratify a sampling design for
sampling soils for sediment fingerprinting that will be discussed later. For example, areas of bare
ground such as unpaved roads and construction sites are important to identify and map. This
information can be found in existing GIS coverages, or examination of aerial photographs.
3.2.4 Hydrologic Alterations
When getting to know your watershed, it is important to be familiar with past and current hydrologic
alterations that may affect the erosion, transport, and deposition of sediment directly or indirectly by
altering runoff or low flow characteristics. These include inter-basin transfers, stormwater and
wastewater discharge, tiling, withdrawals, and dams. In addition, it is important to note any downstream
alterations that may affect the vertical base level or lateral constriction of the stream. This can include
dams and other impoundments. Downstream features can affect upstream sources and sinks of
sediment, and many times they are not included in analyses because they fall outside of the upstream
watershed area. An example of this is the upstream aggradation effects of a dam (Fig. 11). In an
example of the Balsam Row Dam, an impoundment on the Wolf River in northern Wisconsin (Fig. 11),
the impounded section with fine-sediment deposition extends for about 3 km upstream of the dam
(Fitzpatrick, 2005). However, sedimentation of the coarser portion of the sediment load is along the
next 3 km upstream of the dam. A longitudinal profile of water surface and channel bed help to
delineate the extent of sedimentation effects upstream of a dam.
Keshena Falls
Ice jams coincident
with flattening of
water surface slope
Balsam Row Dam
T9
T3
T4.5
T7
T4
j.999 thalweg
Post-dam sand and gravel accumulation
Post-dam fine-grained sediment accumulation
Figure 11. Sedimentation upstream of the Balsam Row Dam on the Wolf River,
Wisconsin (from Fitzpatrick, 2005).
3.3. Geomorphic Setting
The next step after becoming familiar with your watershed is developing an understanding of the geomorphic
setting of your watershed. This involves how upstream and downstream reaches differ with respect to sediment
sources, transport, and delivery.
20
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
The fluvial system can be generalized into 3 zones fSchumm, 1977):
• Zone 1 - high up in the watershed, typically an area of erosion;
• Zone 2 - zone of transport; and
• Zone 3 - zone of deposition (Fig. 12).
For the watershed as a whole and in each of the three zones, the geom orphic setting of the stream system
can be examined in a planform, longitudinal, and lateral view of geomorphic features and processes.
These views of the fluvial system provide important information that is necessary in the construction of
a sediment budget.
ZONE 3
Sediment Deposition
Figure 12. The fluvial system
three zones: sediment source,
transport, and deposition
(Schumm, 1977.
ZONE 2
Sediment Transfer
At even lower elevation a river
wanders and meanders slowly
across a broad, nearly flat valley.
At its mouth it may divide into
many separate channels as it flows
across a delta built up of
riverborne sediment and into the
sea.
ZONE 1
Sediment Production
Mountain headwater streams
flow swiftly down steep
slopes and cut a deep
V-shaped valley.
Rapids and
waterfalls are
common.
Low elevation streams
merge and flow down
gentler slopes. The
valley broadens and
the river begins
to meander.
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
3.3.1 Plan form View
The overall setting of the stream planform (looking down from above) relative to the valley is important
for describing sediment sources and sinks and how the landscape is connected to the stream (Brierley
and Fryirs, 2005). In headwaters (zone 1; Fig. 12), channels are generally in confined valleys where any
sediment produced by hillslope erosion or mass wasting is quickly delivered to the stream channels.
Headwater channels tend to be steep, allowing for efficient transport and delivery of almost all sediment
to downstream areas.
In middle areas of the stream network (zone 2; Fig. 12), stream valleys have widened from zone 1, and it
is here that you may see meandering channels with point bars and cutbanks. Channels may still have
slopes that efficiently transport sediment from upstream areas. If the meander pattern intersects valley
sides (called entrenched valleys), these areas can be major sporadic sources of sediment. Valley
confinement or the degree to which the stream impinges on hillslopes and terraces is an important
feature of zone 2 that has implications for ecosystem management (Nagel et al., 2014). Floodplain areas
in zone 2 are irregular but provide information on storage.
Downstream in lowland settings (zone 3; Fig. 12), where the meander wavelength is always less than the
valley width, streams tend to be more depositional and favor storage of sediment in channels and
floodplains. These reaches are usually not sediment sources.
Along the planform view, there may be broad changes in sinuosity and stream planform that reflect
sediment dynamics (Schumm, 1977). These are reflective of the relative proportion, particle size,
amount of suspended load and bedload, as well as the stream's capacity to transport its load. For
example, a stream may switch from a single thread meandering riffle/pool to a straight braided reach,
possibly indicating a local source of high sediment supply.
3.3.2 Longitudinal Profile
For a vertical view of channel slope changes that indicate whether the channel is a potential sediment
source or sink, it is useful to construct a longitudinal profile along the main stem using the MID
streamlines and topographic contour lines (Fitzpatrick, 2014; Fitzpatrick, 2016). In Fig. 12, most
watersheds are described by having steep reaches in the headwaters with slope decreasing downstream.
In previously glaciated landscapes with relatively young drainages (less than 14,000 years), the
headwaters may be in wetlands with gentle slopes (Fitzpatrick et al., 2015). In either situation, steep
reaches may be prone to bank erosion, incision, and direct inputs of sediment from hillslopes and valley
sides, whereas gentle reaches may be mostly depositional. In general, a concave-up longitudinal profile
reflects steep headwaters and more gentle-sloped mainstem at the watershed outlet. Streams that have
concave-up profiles are expected to have consistent increases in discharge and channel size with
decreases in slope and bed material size (Gilbert, 1877; Fryirs and Brierley, 2013). In young landscapes,
there may be inflections or reaches along the longitudinal profile that are convex. These streams
intersect geologically variable terrains and features that may not show up on a surficial geology map.
Some common points of inflection are outcrops of erosion resistant bedrock or perhaps an end moraine.
Along with the stream planform and valley type, the longitudinal profile helps to form a framework to
help guide field inventories and monitoring (Fitzpatrick and Knox, 2000).
22
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
Original channel
New channel
Aggradation/new alluvium
Drainage divide
Tip of new
channel
Aggradation
Tip of original channel
Stage I Premodified Stage II Channelized Stage III Degradation
Stage IV Degradation and Widening Stage V Aggradation and Widening stage v| Quasj-Equilibrium
LI *
Aggradation/new alluvium
Degradation/erosion
Stage I,
Stage
Stage V
Stage IV
Stage VI
Aggraded
Aggradation
Figure 13. Spatial and temporal changes in stream networks and effects on
longitudinal profiles and erosion and deposition: (A) headword expansion of a stream
network from land clearing and erosion (modified from Strahler, 1958), (B) Channel
Evolution Model, temporal changes due to headword knickpoint migration along
channelized streams (from Simon and Hupp, 1986).
23
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
The longitudinal profiles can be used to explain the longitudinal continuum of sediment erosion,
transport, and deposition over time, especially in situations where there have been changes to the
drainage network or widespread channelization (Fig. 13). Channel and floodplain aggradation are
common along mainstems that have had drainage network extension (James, 2013) (Fig. 13 A). This is
typical in agricultural settings with steep uplands where historical gullying has extended the channel
network into hillslopes (Happ, 1940; Faulkner, 1988). James (2013) specifically describes how legacy
sediment accumulation in river bottoms from anthropogenic-related erosion provides a window into past
deposition rates along floodplains and channel margins.
3.3.3 Lateral View
A lateral view of the stream network can provide useful information on the degree of channel incision
and channel aggradation. For example, in channelized streams, the channel evolution models (CEM) are
helpful for explaining the evolution or trajectory of the vertical and lateral connectivity of sediment-
related channel and floodplain processes (Simon and Hupp, 1986) (Fig. 13B). In the CEM, upstream-
migrating incision is caused by channelization of downstream reaches. Eventually the incised channels
widen and aggrade as incision continues in upstream reaches. Channel and bank erosion are common in
Stages III and IV. It is helpful to know the status of stream channels in your watershed along this
evolutionary pathway.
^ Watershed Reconnaissance
After deciding on space and time scales, getting to know your watershed through available data, and
identifying broad patterns in geomorphic settings, the next step is to plan a reconnaissance survey of the
watershed, stream corridor, and its valley, and further refine knowledge about sediment conditions
(Fitzpatrick, 2014). We cannot emphasize enough how one or several reconnaissance surveys of the
watershed are needed to better understand sediment conditions and start planning for sediment
inventories and monitoring site selection. A car, airplane, or helicopter ride provides a great view of the
watershed. A windshield survey can help identify upland features (farms, construction sites, gullies,
landslides, etc.) and some channel features (eroding streambanks and incised channels), but car rides
often limit you to bridge crossings that may not be indicative of the stream. River walk-throughs are
especially valuable in areas of the stream that are harder to get to by roads and, therefore, build upon the
watershed reconnaissance.
An example of a form that can be used in a watershed reconnaissance is shown in Fig. 14. Areas of
sediment deposition (sinks) may also be noted and included in the reconnaissance (large wide, vegetated
floodplains, fans, impoundments) (Fig. 14). Because roads only provide a limited view of the
watershed, hiking, boating, or canoeing can provide other ways of viewing the watershed. Examining
aerial imagery should be included in the reconnaissance. Speaking to local landowners, county
conservationists, and town historians is extremely helpful. You will find that landowners are very
knowledgeable of the history of the channel and, if you are fortunate, they may have historical
photographs of the channel.
24
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
Watershed Name Stream Name,
Area Name Latitude Longitude Pate . Time
Cropland
Pasture
Timber
Harvesting
Mining
Unpaved
Roads
Active
construction
Ditches
Street and
parking lot
residue
Landslides,
mass
wasting
Gullies
Burn
areas
Lakes and
other
impoundments
Other
CHANNEL FEATURES
Stream Name Reach Name
Latitude Longitude .Date _Time
STREAMBANKS
REACH
FLOODPLAIN
FANS
Bank information
Vegetation
falling or
leaning into
channel
Exposed
roots
% of reach
with Bank
slumping,
bank
erosion
Describe
water level
(low,
medium,
high)
Is
channel
incised?
Dominant
sediment on
Channel Bed
(fines, sand,
gravel,
cobble,
boulders,
bedrock
%
Prr sence
of soft
sediment
in reach
% of
reach
with
bars
Average
width (left
+ right)
Type
vegetation
Note
presence
and type
of fan
Bankfull
Height, m
BankfuH
Width, m
Terrace
Height, m
Bank Angle,
degrees
Are. headcuts
present in reach?
% Occurrence of woody
debris in reach
Describe human structures present in reach
Is reach transport limited or supply limited?
NOTES
Figure 14. Example of form used in the reconnaissance of a watershed to assess upland and channel
conditions.
25
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
3,3,5 Stream Corridor Sediment Sources and Sinks
After the watershed reconnaissance, the next step is to delineate (inventory) sediment sources and sinks
along the stream corridor. These can be identified through the approaches suggested above, aerial
photographs, maps, reports, and watershed surveys. Stream reconnaissance tools have been developed
by many agencies that are very helpful for a rapid evaluation of the stream corridor (for example
Thorne, 1993; Kline etal., 2004; and Rosgen, 2006). Inventorying sediment sources and sinks along an
entire corridor can be done by walking all the drainages in small watersheds (generally less than 10-20
km2) with overview and supplemental data gathered through aerial photograph and historical map
analyses. A short list that helps to remind one who is conducting a river walk of the indicators of
geomorphic instability and possible sediment relations is shown in Table 3. An introduction to some of
the more common methods used by the authors are described below.
Table 3. Geomorphic Features Identified on a River Walk-Through.
Bank erosion width and height
Bars - length and width
Thickness of soft sediment on the channel
bed (length and width)
Recent overbank sedimentation (on
surfaces and in bank cuts)
Major changes in substrate texture
Sand deposition on channel bed and
floodplain
Bankfull channel width and depth
Channel incision
Overall bank heights
Slope estimate
Bedrock outcrops
Indicators of geomorphic stability
Photographs
Road crossing conditions and slope of road
approach
Logjams
Vertical grade controls
Gullies
Mass wasting/valley side failures
3.3.5.1 Aerial Photograph and Historical Map Analyses
Current and historical photogrammetric information (aerial photographs, aerial LIDAR (Light Detection
and Ranging) is a useful method to examine your watershed and identify sediment sources and sinks.
Geomorphic features and sediment sources such as construction sites, landslides, gullies, etc., can be
seen on imagery. These areas can be listed as sites to visit for the watershed reconnaissance.
Depending on the resolution of the aerial photographs and the presence of trees in the river corridor,
spatial variations in channel width, along with accounting for valley side and bluff failures, can be
assessed. If the river is large enough, the areal extent of non-vegetated bars can also be measured. Field
verification is needed for the depth of channel incision.
Quantifying channel location changes over time can be an important component of a sediment budget.
Although information on channel changes is not essential in the early stages of reconnaissance,
determining where the greatest changes in channels occur can be used to target monitoring sites. This
information can also be used in constructing the final sediment budget. Aerial photographs and maps,
when overlaid, give a sense of changes in width, length, and location with time. From these changes,
estimates of rates of lateral erosion and deposition can be made. Lateral migration rates in meanders can
26
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
be determined using a comparison of multiple years of aerial photographs. Lateral migration rates are a
good indication of the maximum rate of bank erosion. Accompanying field-based ground truthing for
heights of erosion or thicknesses of deposition can be added for estimating the total volume of sediment
eroded or deposited.
Digital aerial photographs and maps need to be geo-referenced and entered into a GIS using a second-
order polynomial transformation, with as many control points as possible. A minimum of 8-12 points
are needed to minimize error in geo-referencing (Hughes et al., 2006) and keeping track of resolution
and accuracy. Combining photogrammetry with maps can also provide a useful method to assess
channel change (Donovan et al., 2015; Fitzpatrick et al., 1999; Fitzpatrick et al., 2009).
Airborne LiDAR is increasingly becoming an important tool to quantify channel morphology and
channel change over time (Faux et al., 2009; Dietterick et al., 2012; Roering et al., 2013). LiDAR data
can be converted to DEMs, and using a GIS or appropriate software, such as the Forest Service River
Bathymetry Toolkit
(http://www.fs.fed.us/rm/boise/AWAE/projects/RBT/RBT_lidar_hydro_downloads.shtml), channel
morphology can be derived. One powerful application of LiDAR data for erosion prediction is
calculating the stream power index for segments along the stream network, using slope and flow
accumulation generated from a LiDAR-derived DEM (Nelson, 2010; Danielson, 2013). This index has
a major assumption that flow accumulation is proportional to the watershed area (Wilson and Lorang,
1991).
3.3.5.2 River Walk-Throughs
Using the form similar to what is shown in Fig. 14, or a tally in a field note book, the idea is to walk
most, if not all, of the stream corridor, accounting for characteristics of the banks, bed, and floodplain
(Young et al., 2015). A pace of about 10 km a day might be covered, depending on terrain and
vegetation. Observations of channel features will assist in identifying reaches of sediment erosion and
storage and, again, target sites for monitoring. Photographs and a hand-held GPS are also important.
Many of the measurements made in the river walk-through will help in the final construction of the
sediment budget.
The general features of eroding banks that are described in a reconnaissance are shown in Figure 15.
Figure 16 shows some general guidelines of bank retreat rates. One of the authors has found these rates
to be in the ballpark compared to geomorphic monitoring. For example, a small riparian grazed stream
in southwest Wisconsin had a bank retreat rate of 4.1 cm/yr, determined through monitoring, which
corresponds well with the descriptions of streams in the NRCS category of moderate lateral recession.
27
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
River
Maine:
Date: *5 h{v>lb •units
t " 7 " I 1 B
s:?r.l fi
Relet ID
10
teners
,tlportions, msfr.r
^ in nl-n'-lltir
Le.iFo>'v)
Star+
naci^.
jdwy Joy
US B-p
jar*,
S«A
CWbSlDM
11 xr
4
touec,
icwtt
Sofi
5-j,
ie^s
pool
A'etip.J vvidt'i
rj-rttire1 w did
IhcTiVPfi w.iffr
d ?otti
CHrinwl f/t pi|-
Dir. ni' (ji^d 0F0
5
5.6
O.2.
¦3
ho
ho <"0(g3'
-t? ft
#«»g¥
sa fi
bo w {jv~
53 fi
Boot gv
5»©
1®o co gy~
isa 'i
bo cc©
:sj fi
i3\j .0 h',
jsa fi
lo-iSgv "1bo«6v
saB
}$» fl
togs '
1*2*
0<3-
r-n ma lit
mid trans
|'Jj co e*
15-8 fi
Snfi rvht
I v. VV a "}
4
sicox?
KC-3
*
it 'X fd? oil ipo (t-rf «*lt po rH rlt CI Tfr «i t £0 tTH rflt
Urfrv -tiid 'jarfi 'irid 1*^ n ti tr-iiv jrn'ri Unn^ r^in tJ?r"'
v>/3 ! • f
X t ; ; ' I i
borrgv '.Hi rn fjy bnr-.*;-;; hnrngy [ob CO gv 'buioiiv
| gg11 |sai- xt fi 'si fS Isnf «i fi
J.vt*p«
Bar (iimrnvaH
(UW * T|
* DSO
Banfe'erdSofTCeFor
liignt ;l < m;
'wrt, erosion
'.vlfcj'.-JtL' '11.0
pniidJlt in Ti-i rlt
r, rt tt»o< mid rraii
20X
s k ,r
bu c c p/ bo 11' ^ ]! m i o
f! SJ fl is.' fi
Fw. i .tothI retres!
15*3 f
L-
t>0 CC CV ibo CO gv bu cc^v
sj li w fi
: bL' u L.
IIU LU gV
s*r*
5OCX
m F-
'¦ucogv Bo cop
m>
T
511 fl
Wociiff'''
K,- n
W»l»
Silft
Sq1S|«^
8
Ih^v b it i,; 1 y>
Rsnnt i;r) (i * w > Tf
Ljv!^' DCD
n^r: ro gv
iir* p-'/
ill
ibn r,Ggv
'sg fi
¦ i a co rv
sfl r
i- 160*1 ft 100<
tSt.i htflc b®
ho i"o gv t"o co pv ^D rr>pv *bo co jv
»)') Qii >.'f'
H-jiMlpId ii uvLtb->i^\
I i|Lj i-" F, !;lii
;r: :wkt!
Flaat'piu rt t ^.ifii<
Ki OW
gv ,»8 fl
gv sa fl
gv si Si
|V
L.
Hbt
O.I
gv sa fi
g¥ sa fl
gv sa fi
1 S3 fi g» S3
W.tJp 1 n|i
JtCfrl mijutiuii
nt,i
'..TCt.lTll
Arii.trfJAiiuti
LVl.lt'>5111
irtmiir'uiiu
til' Bir
ri%cin^ DU'ciDf
-ota
"FT
W
2:
Figure 15. River walk-through example field form [D50 abbreviations are: bo =
boulders, co=cobbles, gv = gravels, sa = sand, fi = fines; see Table 4 for size
breakdown of these sediment categories.
28
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
Lateral Recession
Rate ft/yr
(cm/yr)
Category
Description
0.01-0.05
(0.3-1.5)
Slight
Some bare bank but active erosion not readily apparent. Some rills
but no vegetative overhang. No exposed tree roots.
0.06-0.2 Moderate Bank is predominantly bare with some rills and vegetative overhang.
(1 8 6 0) Some exposed tree roots but no slumps or slips.
0.3-0.5
(7.0-15)
Severe Bank is bare with rills and severe vegetative overhang. Many exposed
tree roots and some fallen trees and slumps or slips. Some changes in
cultural features such as fence comers missing and realignment of
roads or trails. Channel cross section becomes U-shaped as opposed to
V-shaped.
0.5+ Very Bank is bare with gullies and severe vegetative overhang. Many
(>15) severe fallen trees, drains and culverts eroding out and changes in cultural
features as above. Massive slips or washouts common. Channel
cross section is U-shaped and stream course may be meandering.
Figure 16. Estimates of lateral recession rates for four degrees of bank erosion (from
NRCS, 2003).
Depositional settings that are noted on the reconnaissance form include the floodplain, bars, and in-
channel soft sediment accumulation. Soft sediment deposits are defined as sediment that you would sink
into and have trouble pulling your boot out from. Fine channel sediment deposits indicate that fine-
grained sediment may be important in the system, as it is often these types of fine sediment deposits that
cause a river to be listed for impairment. We describe later in the manual how calculating the volume of
fine deposits is important. But at this step in the assessment, it is important to identify reaches where
fine sediment is present.
Exposed bars of sand or coarser gravel (those with surfaces that are above the low flow water level) can
be an important component of sediment transport and aquatic habitat (Pitlick and Wilcock, 2001). Many
types of channel bars are recognized in streams (Hooke and Yorke, 2011). Bar area has also been shown
to be correlated to sediment flux (O'Connor et al., 2014; Fig. 17).
The presence of fine sediment and bars may allow you to make a decision on whether the reach is
'supply limited' or 'transport limited.' A stream is supply limited when it is able to transport all the
sediment that is supplied to it; hence, sediment transport is limited by the supply. Transport limited
occurs when the sediment supply to the stream is in excess of the ability of the stream to transport it. A
channel scoured to bedrock would indicate that it is supply limited, such as might occur below a dam. A
channel with abundant bars and soft sediment deposits might indicate that the channel is transport
limited.
29
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
10
1
1 1
CD
0)
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
• Is the sediment budget going to be inventory or monitoring based, or both, i.e., is there time to collect
data on erosion, transport, and deposition rates?
• What resolution of individual budget components is appropriate, and is it feasible?
• How will the sediment budget results be displayed?
• Is there a need to incorporate management techniques for future projections?
Methods used in sediment budgets are tailored to questions being asked, the time and spatial scales of
interest, and available funds and personnel. The steps outlined in Fig. 10 were designed with
agricultural watersheds in mind; however, with a little modification and familiarity with the watershed
setting, the steps would also work for forested or urban watersheds. Methods for constructing fluvial
sediment budgets span disciplines and specialties from soils, geology, hydrology, and engineering. This
manual draws from several references and tailors the techniques towards sediment TMDLs. The
sediment budget approach is also a watershed approach. Planning and using a watershed approach are
summarized in the USEPA (2008) handbook for developing watershed plans. In addition, the following
agencies have a history of working with specific aspects of sediment budgets and an inquiry into current
references by these agencies on methods may be helpful:
• Upland sources and soil erosion - NRCS (e.g. RUSLE2)
• Landslides - USGS
• Gullies-NRCS
• Bank erosion - NRCS, USDA-ARS, U.S. Forest Service
• Channel bed erosion and deposition, impoundments - USGS, U.S. Forest Service, USACE, Bureau of
Land Management
• Floodplain sedimentation - USGS
• Sediment transport (loads, yields) and export - USGS; Forest Service
There are several broad questions that are worth considering at the start of the sediment budget design
process. For example, the question How detailed should a sediment budget be? relates back to the
original objective(s) as well as the time frame needed to derive an answer, financial resources, and the
scale of the watershed. For example, a sediment budget performed for a large watershed (-2500 km2)
may focus on sediment export measurements from tributaries (Meade, 1994), or involve aerial
photograph interpretation of channel erosion and sedimentation. Smaller scale sub-watersheds (<500
km2) may be identified by geologic setting or land cover (Collins et al., 1998) and may involve field
measurements and inventories. At any scale, an identification of potential sediment sources is needed,
and these may be at different spatial scales along the stream network, in tributaries, and along main
stems. Other questions that arise may include: Are there particular reaches or a sub-watershed that
have high sediment loads? and Is there a section of stream with intensive grazing? These questions
are part of the TMDL process.
When working on sediment budgets in Puerto Rico (Gellis et al., 2006), the research group noticed
termite mounds everywhere. The termites would excavate sediment and place the loose soil on mounds.
The research group remarked, "Maybe this is an important sediment source that we overlooked?" When
we started our sediment budget analysis, we did a reconnaissance of the study watershed by driving the
watershed, conducting an aerial flight, and hiking different sub-watersheds. We were able to
31
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
qualitatively rank the types of sediment sources, which we decided to be in order of importance from
construction, to agriculture (cropland and pasture), to forest. We thus made a decision that termite
mounds were probably not an important sediment source.
The Puerto Rico study is a good example of identifying the relative magnitude of importance of various
sources and sinks before delving into a detailed monitoring program that only covers a small source or
area. We could have spent an enormous amount of time and resources determining how much sediment
was supplied from termite mounds, and, in the end, they may have supplied a small percentage to the
overall watershed sediment budget. Qualitatively we assumed that because of the small size of a termite
mound relative to other sediment sources in the basin (e.g., construction sites), the contributions of
termites would be small; a conclusion we could not substantiate. However, if the focus of the study was
to budget sediment at the hillslope scale, then quantifying the impact of termites might have been
important. Incidentally, termite mounds as a source of sediment has been explored (Dietrich et al.,
1982). Furthermore, if the termite mounds were a local source of sediment to a sediment transport-
limited reach of a channel with important sensitive aquatic habitat, even though contribution to the
overall sediment loading is small, the impact to a specific reach may raise its importance in terms of
management.
Is there any way to know precisely how important sediment sources are at the start of your
assessment? No, not precisely, but qualitatively by getting to know your watershed before you select
sampling locations. Know the land cover and get familiar with the local issues, usually concerning
upland soil erosion and streambank erosion. Spend as much time as necessary to become familiar with
the geomorphic setting of the stream corridor. After conducting a watershed reconnaissance and
spending some equally important time looking at the stream corridor, using aerial photograph
interpretation, and conducting a field-based river walk-through, you should know if there are key
locations of erosion and deposition along the river corridor.
Although no areal or linear coverage of a sediment source is absolute for it to be characterized as an
important source, you might choose a threshold of 5 or 10% of the study area or a length of the stream
corridor. For upland sources in watersheds with a uniform geologic setting, you might conclude that
pastureland covering 10% of the entire watershed area is not worth the effort of monitoring, whereas a
construction site or a forest fire that covers 10% of the watershed area may be important. Also, do not
only select upland land use sources that are expected to be major contributors of sediment. Land cover
types that are not known to be large contributors of sediment may still be of interest, especially if they
make up a large percentage of the watershed. For example, it may be important to note that forested
areas were measured and erosion rates were low. In Linganore Creek, Maryland, forests covered 27%
of the watershed and contributed 3% of the total sediment (Gellis et al., 2015).
Early on in the design of the sediment budget, the sediment size of interest is important. The focus of
most TMDLs is fine-grained sediment, which includes clay, silt, and some fine sand (Table 4).These
particle sizes are transported as washload in suspension throughout the water column during runoff
events (Colby, 1963; Edwards and Glysson, 1999; Rasmussen et al., 2009) (Fig. 18). They usually
travel at speeds similar to flowing water. The suspended sediment load likely contains mostly washload,
usually described as the component that is transported through the stream network rather quickly. In
contrast, larger particle sizes usually travel at slower speeds as they bounce along the bottom as bedload
(Edwards and Glysson, 1999). Coarse sediment is assumed to be derived mainly from stream bed and
bank erosion and mass wasting of valley sides and terrace cuts. However, the stream corridor likely
contains fine-grained sediment as well that is eroded from banks, valley sides, and terrace cuts.
32
-------
A Manual to Identify Sources of Fluvial Sediment
E PA/600/R-l 6/210
£
r
o
o
o
F;
fc
o
o
6
fc
H
e
r-
L
€
• ->
fv
\IfHT as ?C*a
405 « V
? 3 « 5
v't'.OCII", tV FtfT
°FR SfCONC
Figuie 18. Distribution of the vertical concentration of sediment mz
-------
A Manual to Identify Sources of Fluvial Sediment EPA/600/R-16/210
Table 4. Standard particle size classes and size ranges used by the U.S. Geological Survey (Lane, 1947; Colby 1963).
Class and subclass
Size range (mm)
Boulder
Very large boulders
2,048-4,096
Large boulders
1,024-2,048
Medium boulders
512-1,024
Small boulders
256-512
Cobbles
Large cobble
256-128
Small cobble
128-64
Gravel
Very coarse gravel
32-64
Coarse gravel
16-32
Medium gravel
8-16
Fine gravel
4-8
Very fine gravel
2-4
Sand
Very coarse sand
1-2
Coarse sand
0.5-1.0
Medium sand
0.25-0.5
Fine sand
0.125-0.25
Very fine sand
0.062-0.125
Silt
Coarse silt
0.031-0.062
Medium silt
0.016-0.031
Fine silt
0.008-0.016
Very fine silt
0.004-0.008
Clay
Coarse clay size
0.002-0.004
Medium clay size
0.001-0.002
Fine clay size
0.0005-0.001
Very fine clay size
0.00024-0.0005
Most TMDL projects concentrate on reducing the suspended sediment load, thus it is important to track
the particle size distribution of each component considered in the sediment budget. The focus on the
fine-grained component of the sediment load stems from the historical emphasis on soil conservation
practices associated with soil erosion from agricultural lands with predominantly fine-grained soils.
Tools and approaches used in a sediment budget can be separated into three categories: 1) sediment
inventories, 2) remote sensing, and 3) field measurements (Fig. 10; Table 2). Each of these approaches
can be used separately or in conjunction with the other approaches. For this manual, we focus on
inventories and field approaches that can be separated into 1) channel and 2) upland measurements
(Table 2). Throughout the manual, we provide references related to studies that have used
photogrammetry (LiDAR and aerial imagery). Sediment inventories reveal the spatial extents and
34
-------
A Manual to Identify Sources of Fluvial Sediment EPA/600/R-16/210
temporal variability (Fitzpatrick etal, 2009), whereas repeat measurements, such as returning to the
channel and upland monitoring sites over a known period, allows for collection of location-specific rates
of erosion and sedimentation (Gellis et al., 2012,2015). It is helpful to use both repeat field
measurements and one time inventorying for a modern sediment budget. Reid and Dunne (1996)
provide a text on how to construct a sediment budget quickly using inventorying techniques.
In any approach selected, development of a data-collection plan is one of the most important aspects of a
sediment budget. Taylor (1990) defines three types of sampling plans: 1) intuitive based on judgment of
the observer, 2) statistical sampling plan of random design to avoid measurement bias, and 3) regulatory
that requires data of a given type and frequency. Sediment budgets typically incorporate types land 2.
For example, intuition may tell you that a large construction site present in a watershed may be an
important source of sediment or, as we previously pointed out, that termite mounds may not be
important. However, it is understood that without data, whether your intuition is correct cannot be
determined. Statistical designs are also important in sediment budgets to assure unbiased measurements
over the watershed of interest and that the number of samples collected is adequate. Site selection
should be carefully designed to provide reliable assessments of erosion and deposition.
'ample size
An important aspect of any effort to quantify the erosion and deposition of geomorphic features is
determining how many samples to collect. Formulas can be used to estimate the sample size needed to
produce a confidence interval estimate with a specified margin of error (Schreuder and Ramirez -
Maldonado,2004; Singh and Masuku, 2014). For more information on sampling design and sampling
size, the reader is encouraged to consult USEPA, 2002, and Artiola et al., 2004.
The confidence interval for a given mean (|i) is as follows in a two-sided test:
Za
ME = <4)
Vn
a
where ME is the margin of error; a is the standard deviation; Z — is the confidence coefficient and o is the
confidence level (CL). In most cases the CL is 95 or 90%. The confidence coefficient is found by taking (1-
CL/100)/2 and finding the appropriate value in a Z Table; found in most standard statistics books. CL values of
95 and 90 have Z values of 1.96 and 1.645, respectively.
By reordering Eq. 1, we can solve for n, the number of samples.
'Za ^
n = I i (5)
1 ME '
35
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
An assumption made in determining the sample size is that the data are normally distributed. Knowledge of the
standard deviation can be obtained from other studies, and the margin of error is decided by the study design.
Box 1 provides an example of selecting the number of streambanks to monitor for erosion and deposition.
Box 1. Example of determining the number of sites to monitor in a study.
In this example, the number of banks to monitor for a sediment budget is determined. Based on
previous work in Linganore Creek, MD (Gellis etal., 2015), the standard deviation of bank change is
6.1 cm. The margin of error selected is +- 2 cm with a 90% confidence level (CL). Using Eq. 3, the
number of banks needed is 26. If a margin of error of 1 cm is desired, the number of banks needed is
102, which would demand much greater resources and time.
3.4.2 Use of Sediment Inventories ¦ ».* nstruc liment Budget
This section describes the basics for conducting sediment inventories to construct a sediment budget.
Sediment inventories usually encompass a one-time visit that covers as much of the stream corridor as
possible, whereas monitoring implies repeated measurements at selected representative sites.
Inventories involve area or volume measurements of erosion and deposition. Volumes can be converted
to mass if the volume-weight conversion or bulk density is measured or referenced (i.e., Chow, 1964)
(Table 5). Inventory estimates of erosion and deposition rates are based on literature values or repeat
measurements from nearby watersheds with similar physiography and climate. Usually, estimates of
erosion and deposition rates are at time scales on the order of decades or centuries where emphasis
should be placed on inventorying at spatial scales over several orders of magnitude (Fitzpatrick, 2014).
Table 5. Ranges in dry bulk density for submerged and aerated sediment
[To convert lb/ft3 to g/cm3, multiply by 0.016.]
Texture class
Chow (1954)
Permanently
submerged
(lb/ft3)
Chow
(1964)
Aerated
(lb/ft3)
NRCS
(2002)
Soils
Clay
40-60
60-80
65
Silt
55-75
75-85
80
Clay-silt mixtures (equal parts)
40-64
65-85
Sand-silt mixtures (equal parts)
75-95
95-110
Clay-silt-sand mixtures (equal
parts)
50-80
80-100
Sand
85-100
85-100
105
Gravel
85-125
85-125
110
Poorly sorted sand and gravel
95-130
95-130
Fine sandy loam
100
Loamy sand
100
Sandy loam
100
Loam
90
Sandy clay loam
90
Clay loam, silt loam, silty clay, silty clay loam,
85
Organic
22
36
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
In typical sediment budget studies, sediment sources and sinks along the stream corridor are often
overlooked because management agencies have programs in place that address potentially large upland
sources from soil erosion associated with agriculture and urban construction through established
conservation and enforcement programs. In many cases, the models that management agencies use to
determine sediment sources only include agricultural sources and do not reflect bank erosion processes,
such as in the Chesapeake Bay Watershed (Meng et al., 2010, Brakebill et al., 2010; Shenk and Linker,
2013). Measurements of sediment sources and sinks along the stream corridor are described in detail in
previous Section 3.3.5, and Figure 15 shows a field form with a tally that can be kept of the most
common sources of 1) banks and bluff erosion and channel incision, and 2) sinks of floodplain and
channel bed.
In large watersheds where it is not possible to evaluate the entire stream network, rates estimated in the
inventory are applied to processes along specific reaches based on their geomorphic setting and relation
to watershed-wide zones of sediment sources, transfer, and accumulation. For stream corridor
measurements, the length of stream channel that the rate is applied to is based on the stream inventory.
In many instances, the hierarchical order of the stream network, called stream order (Strahler, 1956), is
applied (Knox et al., 1974; Gellis et al., 2015). However, riparian land use, such as grazing, and
historical alterations to the channel, such as channelization, can have overriding impacts on channel
erosion and deposition that are separate from watershed size and watershed-wide land use. For example,
mapping channels in regard to their channel evolution stage (Fig. 13) can be helpful for delineating
appropriate stream lengths (Simon et al., 2004.)
Whittlesey Creek, a small Lake Superior watershed in Wisconsin with sedimentation problems and
important habitat for Coaster Brook Trout rehabilitation, is shown as an example where an inventory
style sediment budget was useful for screening potential stream restoration alternatives for a U.S. Fish
and Wildlife Service Refuge (Fig. 19) (U.S. Army Corps of Engineers, Great Lakes Hydraulics and
Hydrology Office, 2010). The study used the U.S. Army Corps of Engineers (USACE) Sediment
Impact Analyses Methods (SIAM) model coupled with a HEC-GeoRAS steady state model (Gibson and
Little, 2006; Little and Jonas, 2010; U.S. Army Corps of Engineers, Great Lakes Hydraulics and
Hydrology Office, 2010) (Fig. 19). The SIAM model tracks sediment transport potential through a
stream network by grain size and accounts for the spatial variations in wash load and bedload employing
user-defined threshold particle size diameters. The model routes washload continuously through the
stream network, but computes differences between bed-material supply and sediment transport capacity.
The model is flexible where local sediment sources and sinks can be added on a reach-by-reach basis in
addition to loadings from the next upstream reach. For Whittlesey Creek, reach-specified estimates of
sediment sources and sinks along the stream corridor were provided by the USGS to USACE by
conducting a river-walk-through style inventory. Upland washload from soil erosion was estimated
from a drainage-area weighted annual loading from an adjacent stream, North Fish Creek (Fitzpatrick,
1998).
From the inventory, it is immediately apparent that valley side/bluff erosion in upper Whittlesey Creek
and the middle section of the North Fork Creek are major sediment contributors, with an order of
magnitude higher contribution than upland soil erosion and channel incision. Overbank deposition in
lower Whittlesey Creek is an order of magnitude higher than channel deposition, but overall, most of the
sediment provided to the stream corridor downstream of the eroding bluffs is transported out of
Whittlesey Creek and to Lake Superior. A SIAM model was run for baseline conditions, by which
differences in relative sediment transport capacity can be compared on a reach-by-reach basis (Fig. 19)
37
-------
A Manual to Identify Sources of Fluvial Sediment EPA/600/R-16/210
or for different management alternatives. For example, reducing peak flows in the upper reaches would
reduce erosion and ultimately reduce sediment loads to Lake Superior, but not change sedimentation in
lower Whittlesey Creek, where the Refuge and Coaster Brook Trout habitat are located.
• Eroding valley side or terrace —
Entrenched or confined valley
o Eroding bank
o Stabilized bluff or bank
Erosion (tons/yr)
Deposition (tons/yr)
Inventory
SIAM
feeder
net erosion
Potential
SIAM
upland
trib bank/
bank/
minus
Local
model
soil
channel
bluff/gully
channel
overbank
channel
deposition
Balance
SIAM Alternative Reduce
River
reach
erosion
erosion
erosion
incision
sedimentation
deposition
(tons/yr)
(tons/vr)
peak flows in upper reaches
North Fork
Creek
NF1
33
22
132
48
0
13
222
-6700
Decrease erosion
NF2
27
27
717
4
0
3
773
-18000
Decrease erosion
NF3
11
11
30
0
18
3
32
20000
decrease aqqradation
Upper
Whittlesey
Creek
UW1
52
39
492
58
0
12
629
-3500
decrease erosion
UW2
11
10
679
29
38
44
647
-400
increase erosion
UW3
25
6
912
85
112
120
797
-280
No chanqe
UW4
42
8
586
42
41
45
592
-29000
No change
UW5
17
12
394
29
81
48
323
23000
No chanqe
UW6
38
19
196
57
106
54
150
2600
No change
Lower
Whittlesey
Creek
LW1
47
43
247
0
169
27
141
-52000
No chanqe
LW2
8
3
14
24
166
15
-132
-2300
No chanqe
LW3
7
0
6
0
376
34
-397
-2900
No chanqe
LW4
3
0
21
0
105
21
-102
6000
No change
LW5
8
0
152
64
230
64
-70
54000
No change
LW6
4
0
0
0
188
35
-220
6400
No chanqe
Watershed total
333
200
4580
441
1634
537
3383
SIAM model loads are provided as order of magnitude/relative difference, not actual values
Blue values indicate potential bed aggradation (transport limited)
Red values indicate potential bed degradation (supply limited)
Figure 19. Sediment budget by inventory.
3.4.2.11nventorying Upland Sediment Sources for Soil and Gully Erosion
Erosion on upland surfaces can occur through sheetwash, rilling, gullying, and mass movements. Many
of these features can be observed in a reconnaissance of the watershed and aerial imagery. Agricultural
agencies use a variety of models to estimate soil erosion. The most commonly used model is RUSLE2
(http://fargo.nserl.purdue.edu/rusle2_dataweb/RUSLE2_Index.htm), although there are many others,
including tools to help estimate sediment delivery, such as
(http://crpub.epa.gov/si/si_public_record_report.cfm?dirEntryId=76041). The RUSLE2 model contains
Galligan's Ford,
(Plugged Flow)
Cherryville Road
USGS Gage
Figure 20
Whittlesey Creek
SIAM Reaches
U.S. Army Corp* of Engineers
Detroit District
38
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
an algorithm to account for some sediment storage, but because the model is applied at a field-by-field
scale, it usually overestimates the amount of sediment that is delivered to streams.
Besides modeling, other features to note in agricultural areas that might not be apparent on GIS land
cover or soil maps are dams and irrigation diversions, poor grazing practices, feedlots, and direct
drainages to channels. For roads, important features to note are the density, location, number and type
of stream crossings, road surfaces, type of drainage along the roads, and any evidence of excessive
deposition downstream of a road crossing (Rosgen, 2006).
3.4.2.2 Inwentopfing Channel Corridor Sources
The goals of inventorying channel sediment are to estimate erosion and deposition along the stream
network. These sources and sinks are described in the earlier section for river walk-throughs and can be
tallied in the field using a form or field book (Fig. 15).
Depositional settings include within channel soft sediment accumulation and bar formation as well as
overbanks. For soft sediment and sand deposition that covers the entire channel bed in some reaches,
the volume can be calculated by multiplying the thickness by the channel width and length of channel
measured during the river walk-through (Fig. 15).
The length and width of eroding banks are measured and the locations are recorded with a GPS. Some
general guidelines of bank retreat rates are shown in Fig. 16. One of the authors has found this to be 'in
the ballpark numbers' compared to geomorphic monitoring. For example, a small riparian grazed
stream in southwest Wisconsin had a bank retreat rate of 4.1 cm/yr, which corresponds well with the
NRCS category of moderate bank erosion. In contrast, measurements of 13 eroding bluffs along a
northern Wisconsin Lake Superior tributary, North Fish Creek, from 1938-1990 had an average retreat
rate of 65 cm/yr (Fitzpatrick, 1998). A quick look at bank cuts or measuring the thickness of un-
vegetated deposits of sand on a vegetated floodplain surface give some indication of very recent
overbank sedimentation following the last flood (Fitzpatrick, 2014). In areas where trees lose their
leaves in the fall, new deposition is readily apparent.
3.4.2.3 Valley Cross Sections and Coring
Geological field methods can be employed to construct cross valley and channel diagrams of overbank
sedimentation and modern channel elevation compared with historical elevations (Fitzpatrick et al.,
1999; 2009; Fitzpatrick, 2014). This helps to estimate the volume of overbank sedimentation. If a
valley cross section bisects a relict channel and the time of the cutoff is determined, much can be learned
about rates of lateral migration and potential bank erosion, as well as rates of incision and aggradation.
Cores are collected along the valley transect to determine the thickness and texture of deposition. The
valley cross sections are tedious to measure; thus, it is important to know something about the stream
network conditions before they are located. It is helpful to locate the valley cross sections throughout
the river valley and in different stream orders. These methods are useful for distinguishing modern from
historical and natural rates. This technique has been used successfully in Wisconsin for a variety of
watersheds where it was determined that post-1830 bank retreat rates were approximately 17 cm/yr
compared to 1.4 cm/yr prior to Euro-American settlement (Knox, 1972; Fitzpatrick et al., 1999;
Fitzpatrick et al., 2006).
39
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
Helpful guides for collecting and describing terrestrial cores and cuts are the "Soil Survey Field and
Laboratory Methods Manual" (Burt, 2009) and the "Field Book for Describing and Sampling Soils"
(Schoeneberger et al., 2012). The USDA textural triangle and color chart (U.S. Department of
Agriculture, Soil Conservation Service, Soil Survey Staff, 1951; Munsell Color, 1975) are used to
classify sediment cores for texture and color. Particle size texturing in the field is done using the soil
texture classes arranged on a texture triangle (Soil Survey Division Staff, 1993) and grading of texture
by rubbing soil between the fingers (Milfred et al., 1967). The field descriptions should be as similar as
possible to methods used to describe cores collected in liners and brought back to the lab. Of particular
importance for sediment budgets is to determine the density of the sediment in order to convert
estimated volumes to mass.
Buried soils (paleosols) are commonly found in floodplain deposits and are indicators of past stability in
floodplain surfaces (Happ etal., 1940; Birkeland, 1984; Retallick, 1985; Fenwick, 1985; Bettis, 1992).
Floodplain surfaces are typically subject to local erosion and widespread deposition. Buried floodplain
surfaces may be observed as a thin, dark organic-rich zone caused by an accumulation of decomposing
organic matter on the floodplain surface. In humid temperate climates, a stable floodplain surface with a
relatively slow or negligible sedimentation rate will have a dark organic-rich deposit at the surface,
similar to what is found on pre-settlement floodplain surfaces (Knox, 1987). Episodic sedimentation
events are noted by deposits of lighter colored sand or sandy loam deposits over dark fine deposits,
indicating a change in the rate of fluvial deposition (James, 2013). Besides being lighter in color and
sandier, post-settlement alluvium tends to be less compacted compared to its pre-settlement counterpart.
Recent fluvial deposits tend to have prominent stratified bedding representative of vertical accretion
(settling of suspended sediment) and lack soil development on the floodplain surface compared to
presettlement deposits (Knox, 1987). Other landforms that are depositional and represent sediment
storage areas are levees, alluvial fans, and crevasse splays (Vanoni, 2006).
Buried channels are notable in cores by the presence of very coarse sand, gravel, cobbles, or boulders
that show up in bank exposures as lenses of coarse grained material similar to the width of the modern
channel (Bridge, 2003).
3,4,3 Field Measurements in the Constructi ment Budget
This section describes field measurements used in a sediment budget to quantify the input, storage, and
export of sediment (Eq. 1; Table 2).
3.4.3.1 Stream Corridor Measurements
Elements in the stream corridor that are examined for change over time include: the floodplain, channel
banks, channel bars, and the channel bed. Many approaches exist in the literature to measure or estimate
channel change (Table 1). Most field approaches involve estimating a change at a point (linear) and
interpolating to the next point or to a series of points to obtain a cross-sectional change (Fitzpatrick et
al., 2005). By extrapolating between cross sections, a volume can be obtained. The process can be
automated with a GIS or other computer application tools such as WinXSPRO, a channel cross-section
analyzer developed by the U.S. Forest Service Stream Team
(http://www.stream.fs.fed.us/publications/winxspro.html).
40
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
A watershed is comprised of many kilometers of stream channels of varying stream order (Strahler,
1957). Sites to measure channel changes can be organized around a reach (Fig. 20). Reaches are often
defined from one meander loop to the next meander and incorporate floodplain, streambank, streambed,
and bars. Although a reach can be heterogeneous, containing a riffle, glide, and pool, other aspects such
as riparian land use and land use history, human alteration, and distance from nearest vertical grade
control should be similar.
Figure 20. Elements of a reach.
Selection of the number of reaches to measure channel change may depend upon the time frame of the
study, budgets, and field reconnaissance (e.g. Fitzpatrick et a/.. 2005). To capture spatial variability,
stream channels of varying contributing areas should be monitored. Monitoring by stream order can in
part fulfill this objective. In most watersheds, 1st order streams comprise the majority of stream lengths.
However, bank heights in 1st and 2nd order channels may be low and the propensity for contributions of
streambank sediment limited. Thus, the practitioner has to face a subjective choice of weighting the
number of reaches for monitoring based on stream lengths of each order (Box 2) or selectively choosing
the number of reaches for each stream order. Although the selection of reaches for monitoring should be
as unbiased as possible, it is recognized that other factors may be involved in site selection, such as
owner permission or a site of extreme importance or interest. In the example shown in Box 2 for
Linganore Creek, MD, the final reach selection was based on the field reconnaissance and not by the
length of each stream order. A reconnaissance in Linganore Creek indicated that lower order streams
showed little evidence of erosion whereas higher order streams showed higher and steeper banks and
thus, more potential for erosion. Although 1st and 2nd order streams had the greater total length of
streams, a decision was made to include more streambanks in 3rd to 5th order streams (Box 2).
Once the number of reaches to monitor streambank change is determined, there are two approaches to
choose the locations for monitoring: 1) a random design, and 2) a regularly spaced design. In a random
design, the stream network is rasterized or gridded and individual cells are selected randomly using a
random number generator (Box 3). In a regularly spaced design, reaches are selected on a regular basis,
such as every 5 kilometers. Using either approach, after reaches are selected, a reconnaissance of the
area will be needed to determine whether this is an acceptable site for monitoring as well as to obtain
landowner permission.
Nj
Floodplain markers
Cut bank
41
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
Box 2. Example how reaches were selected for streambank monitoring in Linganore Creek, MD (Gellis
etal., 2015).
Linganore Creek is a 5th order stream. The length of each stream order is divided by the total length of
rivers to get a weighting factor for each stream order. The weighting factor is multiplied by the total
number of reaches that will be used to monitor streambank change (n = 50) to obtain the number of
reaches that will be monitored for that stream order. However, based on the field reconnaissance in
Linganore Creek, it was decided that the potential for erosion was greater in 3rd to 5th order channels,
and the number of reaches selected for the study was increased for these stream orders.
Stream Order
First
Second
Third
Fourth
Fifth
TOTAL
Length (km) Weighting factor
134.4 0.58
47.6 0.20
37.7 0.16
7.3 0.03
6.0 0.03
232.9 1
Number of reaches
selected based on a
total of 50
29
10
8
2
1
50
Final reaches
selected
9
8
17
7
9
50
Box 3. Example of rasterizing the stream network to choose reaches for monitoring.
In this example, a hypothetical portion of the stream network showing two 2nd order streams is shown. The
second order streams are gridded into 10 m by 10 m cells. In the GIS, an ID number is assigned to each cell.
All 2nd order cells and their corresponding ID numbers are exported into a spreadsheet program. Within the
spreadsheet program, a random number generator selects the number of 2nd order channels of interest.
,322 442
443 563
683 803 923
1277 1397 1517 1637 1757 1877
127aM^—1878 1998 2198
2199 2319 2439 25592579 2
1160 1280 2810
¦ 2811 2931
2812
2693 2813
2694
2695
2696
2697
2698 2818
2819
2820
12821
2822
2822
X ¦
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
3.5.2.1 Bank, terrace, and valley side erosion
Bank erosion has traditionally been moni tored through placement of bank erosion pins. Pins of varying
sizes and materials have been used, generally with pins ranging between 50 to 100 cm lengths and 0.5 to
1.0 mm widths on smaller streams (Peppier and Fitzpatrick, 2005; Bartley et a/., 2008; Gellis et al.
2015). The pins are hammered into a bank, and a small portion is left protruding out of the bank (3.0 to
7.0 cm). The number of pins on a streambank is related to the height of the bank. A general rule is to
install one pin for every 40 cm of height; for example, a 100-cm high bank would have three pins. Pins
can be placed at equal intervals, or if a strati graphic unit is of interest, pins may be placed at unequal
intervals on the bank face.
Pins should be located on both sides of the stream opposite to one another. One side of the stream may
contain a steep eroding bank (cutbank), whereas the other side may be a gently sloping bank or contain a
bar (Fig. 21). If only the eroding side of streambanks are selected, the sediment budget would be biased
towards greater rates of erosion.
Figure 21. Diagram of
meandering channel and
terrace line. At cross
section A, the terrace is not
impinging on the channel.
At cross section B, the
channel left bank has
eroded into the terrace.
Cross
section
iet race
section
Modern
enamel
30 4
30 3
30 2
30 1
30 0
29 9
PS
¦
m
§ L_ Modern channel p
<2
A
0 2 4 6 8 10 12
Stationing from left bank, in meters
30.4
Modern channel
v 30 0
0 2 4 6 8 10
Stationing from left bank, in meters
Pins are measured when installed and at selected times thereafter. The time between measurements may
vary depending on the objectives of the study and how the erosion and deposition rates vary over time.
For example, if the objective is to understand how storm events affect deposition and erosion, then pins
would be measured after individual storms. If the objective is to understand how seasonality affects
channel change, such as freeze-thaw acti vity during winter months, then pins would be measured
seasonally. In general, it is common to measure pins annually over the course of the study period.
However, if erosion rates are high (~ 1 m/yr), pins may erode out of the banks and be lost. High
deposition rates may cause a similar problem where pins are buried and cannot be located. In settings of
high erosion and deposition, pins should be read as frequently as every 6 months.
43
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
Box 4. Example of how erosion and deposition rates are determined for a bank face.
In this example of a hypothetical bank, pins were installed on the left bank on March 1, 2010 and measured
three times thereafter. Pins were placed at equal distances apart (35 cm). The average change for the entire
bank face is determined after each measurement. The final change in the streambank is the sum for each
measurement, in cm. The total change (net change) for the streambank is multiplied by the bank height (shown
here as 1.4 m), resulting in a change in cross-sectional area (cm2). Average annual change is computed as the
total change in the streambank (cm or cm2) divided by the number of days of monitoring (shown here as 570
days) and multiplied by 365.25.
Installation reading,
cm
3/1/2010
Measurement on
8/14/2010
Reset on
8/14/2010
Change from
previous
measurement, cm
Measurement on
4/21/2011, cm
Pin 1 at 17 cm
7
21
6
14
21
Pin 1 at 42 cm
6
18
8
12
31
Pin 1 at 79 cm
9
13
no
4
19
Pin 4 at 114 cm
10
6
no
-4
9
Streambank average
6.5
Reset 4/21/2011,
cm
Change from
previous
measurement, cm
Final
measurement
on
9/22/2011,
cm
Change from
previous
measurement, cm
Total change for
study period
summing changes
for each pin, cm
Pin 1 at 17 cm
4
15
18
14
43
Pin 1 at 42 cm
8
23
22
14
49
Pin 1 at 79 cm
7
6
15
8
18
Pin 4 at 114 cm
no
3
12
3
2
Streambank average
11.75
9.75
28
Pin
Number of days of
study
Total change for
study
period for each pin,
cm/day
Average
annual
change for
study period
by averaging
all periods,
cm/yr
Average annual
change, cm2/yr
17 cm
570
0.075
42 cm
570
0.086
79 cm
570
0.032
114 cm
570
0.004
Streambank average 570 0.049 17.9 2512
The amount of erosion and deposition on each pin is the difference from the current measurement to the
previous measurement (Box 4). Changes for a bank face are estimated by averaging all changes in the
pins (Box 4). If the pins are placed at unequal intervals, then weighting the pin readings by the distance
each pin represents may be favorable (Box 5).
44
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
Box 5. Example of determining streambank change where pins are placed at unequal distances on the bank
face.
Example using the same bank in Box 4 but where the pins were placed at unequal divisions on the bank. The
pins are weighted by the length of bank represented by each pin. The total bank length is 140 m. Positive
values for streambank change indicate erosion.
Pin location from
top bank, cm
Length of bank
represented by
each pin, cm
Weight
Total change
overtime,
cm
Weighted change
Average
annual
change, cm/yr
Pin 1 at 10 cm
10
25.5
0.1821
0.075
0.0137
Pin 1 at 41 cm
41
35.0
0.2500
0.086
0.0215
Pin 1 at 80 cm
80
36.5
0.2607
0.032
0.0082
Pin 4 at 114 cm
114
43.0
0.3071
0.004
0.0011
Streambank
average
0.049
16.3
All streambanks in a reach are averaged to get a reach-averaged change. Since dates of streambank
measurements may vary from one reach to another, or even for the same reach, measurements are
normalized by the amount of days between measurements to obtain a change per day (cm/day) (Box 5).
The reach averaged change is multiplied by 365.25 to get an average annual change (cm/yr).
In many studies, streambanks are measured several times during the course of the project. Changes in
streambanks between resurveys are termed gross changes, and for the entire study period are termed net
changes. Typically, it is the net rate of change (cm/yr) that is of interest (Boxes 4 and 5). In some cases,
large storms may occur during the measurement period and interim rates of change (gross change) may
be of interest.
In a sediment budget, channels are typically grouped or classified. Classification can include stream
order, land use, geology, contributing area, or some other factor that may represent the variability in
channels. Box 6 illustrates how changes in streambanks for a given stream order can be converted into a
mass change (Mg/yr). It is important to note that because there are streambanks on either side of the
channel, the final mass change is multiplied by 2.0.
For large valley side failures and terrace cuts, ground-based LiDAR can be used. This technique is
becoming increasingly popular because of the ease of measurement and quantitative results (Collins et
al., 2008).
45
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
Box 6. Example of how to convert streambank measurements at the reach scale (cm) to a mass
(Mg) for first order streams.
Rates of streambank change were obtained from Gellis etal. (2015) for Linganore Creek, a 147 km2
agricultural and forested watershed in Central Maryland.
Stream
order
Reach ID
Date from
Date to
Total days
Number of banks
measured in
reach
1
A
3/27/2009
12/23/2010
636
2
1
B
3/27/2009
3/16/2010
354
2
1
C
3/30/2009
11/8/2010
588
1
1
D
6/1/2009
12/21/2010
568
3
1
F
7/22/2009
11/3/2010
469
4
1
G
5/1/2009
11/3/2010
551
4
1
H
3/24/2010
11/8/2010
229
6
1
1
5/8/2009
12/22/2010
593
2
1
J
10/15/2008
12/29/2009
440
3
Why no reach "E" in either of these tables? Why are numerous columns b
ank in table below?
Reach ID
Average net
erosion of
bank
(cm2/day)
(+ = erosion)
Margin of error
90% confidence
level (cm2/day)
Density of
streambanks
(g/cm3)
Length
of channels
(km)
Mass
(kg/day)
Mass
(Mg/year)
A
1.67
0.83
B
-0.14
1.81
C
-0.32
1.56
D
-0.08
1.19
F
-0.03
0.98
G
0.00
0.96
H
0.74
1.21
1
0.48
1.40
J
-0.22
1.08
AVERAGE
0.23
0.12
1.22
122
6971
2546
46
-------
A Manual to Identify Sources of Fluvial Sediment
3,6,2,2, Incised Channels
EPA/600/R-16/210
When performing sediment budgets in watersheds containing incised channels, a decision on which
streambanks to monitor should be made. If terraces are found to be a distance away from the active
channel (Fig. 21), they may not be monitored. In other cases, the active channel meanders into the older
deposits and the terrace is in proximity to the active channel (Fig. 21B). In these cases, the terrace
deposits may be eroding and supplying sediment to the channel. A decision on whether to monitor
banks where terrace deposits may be contributing sediment to the active channel may be made after a
reconnaissance. For example, if after a reconnaissance on watershed sediment sources, only 10% of the
channels appear to have terraces contributing sediment, terraces may not be monitored. If 50% of the
streambanks appear to have contributions from terraces, then these banks might be part of the
monitoring program. In other settings where the channel is incised, such as in urban channels, the entire
streambanks are usually monitored.
3,6,2,3 Channel Cross Sections
Surveys of the channel cross section can be used to monitor changes in the channel bed, bars, and banks
(Harrelson et al., 1994; Peppier and Fitzpatrick, 2005; Gellis et al., 2012; Moody and Meade, 2013).
Similar to bank pins, at least two channel cross-sectional surveys are established in a reach. Steel pins
are established on either side of the cross section, vertically into the ground, and a tape is stretched
between them. Pins can vary in length from 2 to 4 ft and 1/2 to 1 in. in diameter. The pins also become
permanent monument markers to establish elevation control over time. Instruments used to survey the
cross sections can include survey levels, total stations, or GPS units.
Each reach contains several cross sections where bank pins are installed. The number of cross sections
selected for monitoring may depend on the heterogeneity of the reach. Variability is a factor in channel
erosion and deposition, and the more cross sections that are monitored will provide a better
understanding of erosion and deposition.. It is recommended that at least two cross sections are installed
in a reach.
The distance between cross sections in a reach can be a function of
Spacing = nW
where Wis bankfull width and n is a number usually between 3 and 20 (Harrelson et al., 1994).
The zero point on the cross section is usually on the left side or left bank of the channel. The left bank is
defined looking in the downstream direction on the left side. The channel survey proceeds across the
cross section noting the station and elevation. Survey points occur at a regular spacing and at breaks in
slope. A simple rule for regular spacing of survey points is to divide the total cross section by 10 and
round down. A 42-m channel cross section would have a survey point shot every 4.0 m. The first shot
is half this distance (2.0 m) and then every survey point afterwards is 4.0 m. If a break in slope occurs
outside of this regular spacing, these points are also surveyed. On resurveys, the tape is stretched to the
same width as in the initial survey. The same spacing is used, but the breaks in slope may change. For
explanations on how to survey and calculate elevations, see Harrelson et al. (2004). Channel changes
interpreted from the resurveys include changes in the total cross-sectional area of the channel, changes
in top width, depth, channel bed, bars, and streambanks (Fig. 7A).
47
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
One choice that arises in channel surveys is how wide the cross-section survey should be. In general,
the novice usually ends the cross sections too soon. The cross-section end points are established a
distance away from the streambank to avoid them from being lost to erosion. Ideally, a cross section
stretches from the edge of the floodplain or terrace on the left bank to the edge of the floodplain or
terrace on the right bank. The edge of the floodplain or terrace may be distinguished by a break in slope,
change in vegetation, or a change in soils. At a minimum, a channel cross section should extend into the
floodplain, two channel widths on either side of the channel in order to incorporate depositional features
from overbank flows, such as levees. In larger rivers, the edge of the floodplain on either side of the
river may stretch for a considerable distance (100s' meters). In these cases, the width of the cross
section may be shorter and floodplain tiles are used to quantify deposition installed away from the cross-
section endpoints.
For large rivers that are too deep to wade, bathymetric surveys can be conducted and joined together
with topographic surveys (Fitzpatrick, 2014). Acoustics devices are becoming increasingly popular for
bathymetric surveys; recent methods are outlined in Lee (2013).
3.5.2.4 Channel Bars
Changes in channel bars can be determined from cross-sectional surveys or by using pins. Cross-
sectional surveys measure a cross-sectional change, and pins measure a linear change. Unlike
streambanks, bars are intermittent and may not appear throughout the stream network. Therefore, a
separate inventory of bars should be made (Box 7).
Box 7. Example of how bar lengths are estimated for a given stream order.
In this example for a hypothetical stream, a tape was laid out at each monitoring reach. The length of
bar present within the tape length was noted. The percent of lengths was averaged for each stream
order. This example is from Difficult Run, VA, an urban watershed (Gellis et al., In Press).
Stream
order
Length of
tape
used in
bar
survey, m
Total
bar
length,
m
Average
bar
width
m
Percent of
reach
containing
bar
Standard
deviation
A
1
50
4.0
0.442
8.0
B
1
50
6.3
0.597
12.6
C
1
50
12.5
0.641
25.0
D
1
50
14.4
0.521
28.8
Average 1st order
18.6
9.9
E
2
63
24.8
1.098
39.4
F
2
200
29
1.162
14.5
G
2
50
7.7
0.4572
15.4
Average 2nd order
23.1
14.1
48
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
If pins are used to record changes in bars, the average change in the pins can be multiplied by the
average bar area (m2) to obtain a volume (m3). If a survey is used, the cross-sectional change in the bar
can be multiplied by the average length of bars to obtain a volumetric change (m3). The volumetric
change (m3) is multiplied by the bar density (g/cm3) to obtain a mass (Mg).
3,6,2,5 Floodplains
Floodplains are important areas of sediment storage (Ross et al., 2004; Hupp et al., 2008). Floodplain
deposition can be quantified in a sediment budget using artificial markers (Kleiss, 1993; Hupp et al.,
2008; Gellis et al., 2015), dendrochronology (Hupp, 2000), surveys (Curtis et al., 2013) and
radionuclides (Kleiss, 1993; Amos et al., 2009; Golosov and Walling, 2014).
Artificial marker layers (clay pads, tiles) have been used to monitor floodplain deposition in sediment
budgets (Schenk et al., 2012; Gellis et al., (2015). Markers are laid out on a tape along a cross section
that may or may not be surveyed. Clay pads are powdered white feldspar clay laid on the floodplain
approximately 20 mm in thickness and placed over an area of -0.5 m2 The clay becomes a fixed plastic
marker after absorption of soil moisture that permits accurate measurement of short-term net vertical
accretion above the clay surface (Ross et al., 2004; Gellis et al., 2009). Square tiles (8x8 cm) made of
porcelain or terracotta are used in a similar fashion to clay pads but have the advantage of being a solid
surface. A small hole is dug in the floodplain and filled with cement, and the tile is laid in. The number
of tile laid out on the floodplain depends on the floodplain width. Sedimentation rates are highest near
the channel margin (Simm and Walling, 1998), and markers may not be evenly spaced, where more
floodplain ,markers are placed near the edge of the channel.
During or at the end of the study period, the clay pads are examined for depth of burial. Depth of burial
for clay pads is measured by coring the ground surface through the clay pads and measuring the vertical
depth of sediment above the artificial clay layer. For tiles, a ruler is used to measure the surface of the
tile to the ground surface. Cores of the deposited sediment over the clay pads are used to determine
floodplain density. For tiles, the entire mass over the tile is sampled and weighed to determine the
floodplain density.
Determining a mass of deposition is similar to the method used for streambanks (Boxes 4 and 5). A
deposition rate for each floodplain cross section can either be averaged using all markers or weighted by
the distances between markers (similar to Box 5) and multiplied by the width of the floodplain (m) to
estimate a cross-sectional area of deposition (mm2). To get a reach average rate of deposition, the cross-
sectional area of deposition for all cross sections is averaged and divided by the measurement period
(mm2/d). All reaches in a given classification (e.g., stream order) are averaged and multiplied by the
stream lengths to produce a volume (m3/day). The volume is multiplied by the average floodplain
density to arrive at a floodplain deposition mass (Mg/day) and multiplied by 365.25 to get an annual
deposition rate (Mg/yr).
3,5,2,8 Measurements of Density
Measurements of sediment density (g/cm3) are necessary to convert volumetric change (m3) to a mass
(Mg). This should be done for all components that are measured as part of the sediment budget, whether
above or below the water surface. Density can be obtained by taking cores of the channel feature with a
49
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
coring device.
For sediment sampling above the water surface, such as for banks or bar deposits, a field core sampling
tool is used to drive a cylinder into the ground (Lichter and Costello, 1994; USDA Forest Service,
2005). Cylinder size will vary depending on the material being sampled. Fine-grained to sandy
streambanks can be cored using metal cylinders 8 to 12 cm in length and 4 to 6 cm in diameter (Gellis et
al., 2015). Coarser material, gravel and cobble, require wider cylinders (>10 cm) and varying lengths
(>10cm). The cylinders are pushed into the material, and the core is extruded. The core is dried and
weighed, and the dry mass of the sediment is divided by the volume of the core cylinder to obtain dry
bulk density (mass/volume). Although density is a relatively simple measurement, it is hard to obtain
samples without compacting the sediment. Therefore, care must be taken when driving the cylinder into
the bank. The USDA Forest Service (2005) describes the impact-driven core sampler that is used to
collect a known volume of soil with a minimum of compaction and disturbance. A density measurement
should be made for each geomorphic element at each reach. To account for variability, it is useful to
take several measurements at the same location.
For submerged sediment in the channel bed, such as soft sediment deposition, a core tube with a known
diameter and length is carefully pushed into the deposit. Clear water is poured off the top of the tube
and the vertical length of the core recorded before storing the sample in a plastic container. Back at the
laboratory, the wet sample volume can be measured and checked against the field measurement of core
penetration. The wet and dry samples are weighed to obtain dry bulk density. Density of soft sediment
can be quite low because of the high water content. For example, in the Driftless Area of southwest
Wisconsin, soft organic rich sediment in a small agricultural stream had a bulk density of 0.8 g/cm3 or
50 lb/ft3.
For general inventories and order of magnitude comparisons, general guidelines are provided in Table 5
for volume-to-weight conversions for dry and wet deposits with variable textures (Chow, 1964).
i Ipland Measurements
This section describes field techniques to measure erosion and deposition on upland land elements,
including contributions from roads.
Similar to selecting reaches for stream channels, a random design should be used for the selection of
upland sites to determine upland erosion and deposition. Using a GIS, polygons of a given land use type
can be rasterized and a random generator used to select pixels (Fig. 22). Dirt roads that are linear
features can be treated as stream channels, rasterized, and pixels selected at random. It is understood
that in some cases there are upland areas that warrant monitoring as determined from the field
reconnaissance or from knowledge of the watershed. For example, perhaps a select land use drains
immediately adjacent to an ecologically well-functioning stream reach that needs to be monitored.
50
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
Agriculture
Figure 22. Example of selecting upland areas for monitoring of erosion and deposition. In
this hypothetical watershed, areas in agriculture are shown and rasterized. Each cell is
numbered and either in a GIS or spreadsheet, cells are selected randomly.
3.5.4.1 Unpaved Roads
Unpaved roads can be important sources of sediment (Reid and Dunne, 1984, Ramos-Scharron and
MacDonald, 2005). Similar to channels, monitoring of unpaved roads is accomplished by selecting
reaches throughout the road network to capture the spatial variability. Measurements of unpaved roads
over time at each reach can be accomplished with level surveys if erosion is expected to be greater than
the precision of level surveying (generally +-0.02 ft). If finer measurements of erosion and deposition
are needed, erosion bridges can be used (Ypsilantis, 2011). An erosion bridge is an aluminum or metal
board that is placed level across two rebar. The width of the bridge can vary depending on the width of
the feature being measured. The rebar should be at least 4 ft in length and hammered into the ground so
when the erosion bridge is place across the rebar it is level. The bridge has 10 equally spaced openings
across its length where a measuring pin is inserted. The measuring pin is ruled in millimeters and is a
length of 2 ft. Because freeze/thaw activity can cause movement of the pins, the elevation of each rebar
is surveyed relative to a stationary benchmark monument. The benchmark can be a plate cemented into
the ground. If infrastructure is nearby, such as the corner of a bridge, this can be used as a benchmark.
The average erosion or deposition rate of unpaved roads at a reach (cm) is multiplied by the road width
(m) to calculate a cross-sectional area change (m2). This value is multiplied by the unpaved road length
(m) to determine a volume. The volume is multiplied by the density (g/cm3) of unpaved road material to
get the final mass (kg). Dividing the mass by the number of days of measurement (kg/day) and
multiplying by 365.25 produces an annual mass determination (kg/yr).
3.5.4.2 Upland and Hillslope Measurements
Various approaches ranging from field collection to fallout radionuclides have been used to estimate
hillslope erosion and deposition. A common field approach is to capture eroded material in pits or traps.
Silt fences, 3 to 15 m across the hillslope, can be installed to capture eroded sediment (Robichaud and
51
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
Brown, 2002). The amount of mass deposited behind the silt fences is determined and related to the
contributing area on the billslope to estimate a yield (Robichaud and Brown, 2002).
Traps installed on hillslopes, often called Gerlach Troughs (Gerlach, 1967), can be used to capture
eroded sediment (Gellis et al, 2012; Larsen et al, 2012) (Fig. 23A). Gellis et al, (1999, 2012) installed
plastic rain gutters, ranging from 52 to 85 cm long and 8.5 to 13 cm deep, with each end capped off on
hillslopes to capture runoff and sediment. Holes drilled into the sides of the rain gutter were attached to
collection buckets with plastic hosing. The contributing area to each trap was bounded with metal
edging (Gellis et al., 2012). After rainfall events, the sediment water mixture was taken to the
laboratory and dried to determine the mass of sediment (Gellis et al, 1999; 2012). The traps were
helpful in quantifying erosion for individual storm events but required a great deal of labor.
Sediment
traps
Figure 23. Examples of field approaches
to monitor upland erosion and
deposition (Gellis et al., 2012). (A)
sediment traps, (B) straw dam showing
deposited sediment in the pool upstream
of the dam, and (C) nails.
Small impoundments and ponds have also been used to determine the sediment yield. Gellis etal.,
(2012) placed straw in zero-order channels, i.e., channels that are of too fine a scale to be delineated as
first-order channels, in New Mexico to create a sediment pool (Fig. 23B). The sediment pool was
surveyed periodically to determine the volume of deposited sediment. Taking cores to determine the
density of the deposited sediment allows the volume to be converted to a mass. The contributing areas
to the dams were surveyed as a total station. Quantifying the contributing area to the dams enabled a
sediment yield (kg/m2/yr) to be determined. Repeat surveys in small ponds can also be used to
determine the amount of deposited sediment (Renwick etal, 2005).
Small pins or nails can also be used to quantify hillslope erosion and deposition (Leopold et al, 1966;
Gellis et al., 2012) (Fig. 23C). On grazed areas of New Mexico, Gellis et al (2010) used a 15-cm-long
nail put through a washer and driven into the ground to measure erosion and deposition. Nails were
arrayed in lines, 4 to 163 m long and measured periodically.
52
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
Field measurements of erosion and deposition have their advantages and disadvantages. Advantages
include direct on-the-ground measurements. Disadvantages include investment in time and labor, and
some of the measurements may only quantify erosion and deposition at small spatial scales (10's of
square meters).
3.5.4.3 Upland Erosion and Deposition Using Cesium-137
The cesium-137 (137Cs) technique has been used worldwide to estimate soil loss and gain (Sutherland,
1989; Ritchie and McITenry, 1990; Gellis t7
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
3.5.5 Age Determinations of Sediment Collected in the Field
Sediment budgets that have the objectives of determining sediment budgets for different time periods
rely on age determinations of sediment. Sediment budget rates often are examined over different time
periods to understand natural versus human-influenced rates or changing land use conditions (Trimble.
1983; Kesel etal., 1992).
Several approaches can be used to determine the age of sediment deposits ranging from historical maps,
artifacts, anthropogenic markers (i.e., old road surfaces and dating cores). A brief review of some of the
most common methods for dating sediment are in Fitzpatrick (2014). Berglund (1986) is a standard
reference for many of the terrestrial methods for Quaternary studies. For aquatic environments, methods
employed by paleolimnology and lake sedimentation studies are helpful (Hakanson and Jansson, 2002).
Radiometric techniques based on the uranium decay series, such as lead-210 (Schelske etal., 1986;
Olsson, 1986; Cohen, 2003) or 137Cs derived from atmospheric fallout from above-ground nuclear
bomb testing that occurred between 1953 to 1973 (Van Metre etal., 2004), are best done with profiles of
fine-grained sediment that have had no erosion, chemical or physical remobilization, bioturbation,
pedogenesis, decomposition, or diagenesis. This technique is most often used for aquatic sediment
profiles but can also be used carefully for fine-grained vertical accretion deposits in terrestrial floodplain
settings.
Radiocarbon (carbon-14) dating (Libby el al., 1949) has a resolution back to about 50,000 years where
wood and organic matter as well as carbonate precipitates can be dated. Radiocarbon dating has been
especially useful for dating wood and charcoal in buried abandoned channels and can be used in both
terrestrial and aquatic environments where organic matter is present.
For floodplain sediment, optically stimulated luminescence (OSL) dating has been used that has a
resolution back to about 800,000 years (Lundstrom et al., 2008). This technique calculates the time
since the sediment was last exposed to sunlight or intense heat and uses quartz or potassium feldspar that
are commonly found in sand. This technique is helpful for determining the age of coarse-grained fluvial
deposits.
3.5.6 Measuring Sedimer mpori and Export
The inputs and storage of sediment should hypothetically equal the output or export of sediment from
the watershed of interest. The output of sediment is typically expressed as a suspended sediment or total
sediment load (kg). In some studies, the output is measured as the volume of sediment in an
impoundment (m3). The measurement of suspended sediment loads has less error than the input and
storage measurements, making this an important aspect of a sediment budget. Sediment transport may
be monitored at the watershed outlet and potentially at multiple sub-basins of interest. The collection
and computation of fluvial sediment transport (suspended and bedload) can be found in (Edwards and
Glysson, 1999; Rasmussen etal, 2009). The details of this exercise are quite involved, and, thus, it is
best to consult these USGS techniques manuals. A short video illustrates a setup on the Patapsco River
in Maryland (https://vimeo.com/29783454). Suspended sediment concentrations and bedload are
collected along with continuous discharge so that a rating curve can be established between
concentration and discharge for computation of event-specific or annual loads. Automated samplers are
usually best to use for suspended sediment because runoff inconveniently happens in the evening hours
of a Saturday night. Some examples of automated samplers are in Anderson and Rounds (2010). Less
54
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
expensive siphon samplers for suspended sediment and nutrient sampling can be strategically located
upstream (Graczyk et al., 2000).
3,6,6,11mpoundment Surveys
Impoundments along stream corridors, whether on main stems or tributaries, consist of small sediment
detention basins that contain important sedimentation records that can be used as sediment budgets as
the storage or output terms in Eq. 1. Field methods for determining the volume and mass of sediment
stored in impoundments are described in Bureau of Reclamation (2006). The techniques are a
combination of high-level surveying to identify the spatial distribution and thickness of post-
construction sediment accumulation and cores to calculate the volume-to-weight conversion factor.
These techniques can be applied to either inundated or dry sediment accumulations. General guidelines
of volume-weight conversion factors for a variety of sediment sizes are given in Table 4. An important
aspect of using impoundments is estimating the trap efficiency, the ratio of sediment inflow to outflow,
of the impoundment (Renwick et al., 2005). Results of impoundment surveys are documented as a
volume (m3) or mass (kg).
3.5.7 Budget Calculations
A sediment budget is an accounting of the sources (erosion), storage (deposition), and delivery
(transport) of sediment in a watershed (Eq. 1). The field measurements and estimates of areas are used
to determine a sediment budget. The final sediment budget is the summation of all measurements of
streambanks, channel beds, bars, floodplains, upland areas, ponds, etc., computed using the following
equations:
/'. = total sediment export to the watershed outlet, in megagrams per year ('+' = erosion. =
deposition);
St = erosion or deposition from streambanks, in megagrams per year;
Sb = erosion or deposition from streambed, in megagrams per year;
= deposition from floodplain. in megagrams per year;
Ag= erosion or deposition from agricultural areas, in megagrams per year;
/•>' = erosion or deposition from forested areas, in megagrams per year;
Pd = the total mass of sediment deposited in ponds, in kilograms per year, and
Ts - Sk + Sb + Fp +Ag + Fr + Pd
(6)
where:
where:
^=e::
Kr(s) * Km * (4 * 100) * (Mk /100)
1,000
(7)
55
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
Sk = erosion or deposition from streambanks in kilograms per year;
Kr(s> = net change in streambanks for stream order (5), in square centimeters per year (Eq. 13);
A';, = streambank sediment density, in grams per cubic centimeter;
L, = length of streams for stream order (5), in meters; and
Mk= percent silt and clay in streambanks of stream order (5).
and:
Brjs) * Bm * (4 * 100) * (Mi /100)
1,000
(8)
Sb = erosion or deposition from streambed. in kilograms per year;
Br(,> = net change in channel bed for stream order (5), in square centimeters per year (Eq. 10);
Bu = channel bed density, in grams per cubic centimeter; and
Mb= percent of silt and clay in channel bed of stream order (5).
and:
F,=Z
Fr(S) * Fa * (L, * 100) * (Mf ./100)
15000
(9)
where:
FP = deposition from floodplain. in kilograms per year;
Fr(,> = net change in floodplain for stream order (5), in square centimeters per year (Eq. 17);
Fp. = floodplain sediment density, in grams per cubic centimeter; and
M/= percent silt and clay in floodplain of stream order (5).
and:
^=[4i*(Sa/l00>*4]* 1,000
F, = [Fcs *(S0 /100)*4-)]* 1,000 (]|)
where:
Ag= total erosion or deposition from agricultural areas, in megagrams per hectare per year;
Ar,= agricultural erosion estimated using 137Cs, in megagrams per hectare per year;
56
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
Sa= percent silt and clay in agriculture areas;
Ai= total area in agriculture, in hectares;
Fr= total erosion or deposition from forested areas, in megagrams per hectare per year;
Fcs= forest erosion estimated using 137Cs, in megagrams per hectare per year;
So = percent silt and clay in forest; and
A/= total area in forest, in hectares.
All sediment inputs and storage terms should balance to the output (Eq. 1), but this rarely happens
(Kondolf and Matthews, 1991). Kondolf and Matthews (1991) reported imbalances as high as 104% of
the total sediment output. Errors involved in developing sediment budgets include limited spatial and
temporal measurements, inaccurate field techniques, natural data variability, overly simplified models,
and extrapolation of measurements to unmeasured areas. It is recommended that in the final
dissemination of the sediment budget the range of values be shown as confidence intervals (i.e., 10th
and 90th percentiles).
3.5.7.1 Error Analysis
Construction of the sediment budget relied on averaging and summing field measurements. A
confidence interval is a range that encloses the true average of the measurement with a specified
confidence interval (i.e., 10th and 90th percentiles). Any standard statistics package can provide
examples on how to determine confidence intervals. In the computation of the sediment budget,
measurement averages are multiplied and added (Eqs. 4-8). The propagation of uncertainty (i.e., 10th
and 90th confidence intervals) has defined rules for multiplication, addition, and subtraction (Bevington
and Robinson, 2003) (Table 6).
57
-------
A Manual to Identify Sources of Fluvial Sediment EPA/600/R-16/210
An example of how a sediment budget is computed is shown in Box 8.
Box 8. Computation of the Sediment Budget
The example shown here is from a sediment budget study conducted by Gellis et al. (2015) for
Linganore Creek, MD from 2008 through 2010. Land use in 2006 in the 147-km2 watershed was 27%
forest, 62% agriculture (pasture and cropland, 8% developed, and 3% other). The watershed is listed on
Maryland's 303D list for sediment impairments, and the cooperator, Frederick County, was interested in
targeting the sources of sediment.
Channels (streambanks, channel bed, and floodplain), uplands (agriculture and forest), and storage areas
(ponds) were measured in the sediment budget. Streambanks were monitored with pins (n = 50
reaches), the channel bed through level surveys (n = 22 reaches), the floodplain using clay pads (n = 20
reaches), agricultural fields (n = 18) using 137Cs, forested slopes (n =13) using 137Cs , and an estimate of
pond storage (n = 195) using photogrammetric, GIS analysis, and literature values (Gellis et al., 2015).
The output of sediment was computed as suspended sediment loads at the mouth of the watershed where
a USGS station was located (Linganore Creek near Libertytown, MD - Station ID 01642438). The
sediment grain size of interest was <0.063 mm (silts and clays), and samples from all sources were
sieved. Confidence intervals (CIs) (10th and 90th percentiles) were used to assess the uncertainty about
the mean of the field measurement.
The following is a summary of the computation of the sediment budget (negative numbers indicate
erosion and positive numbers indicate deposition).
A) Summary of Channel Measurements
Stream
order
Length (m)
Streambank change
(cm2/yr)
Channel bed
change (cm2/yr)
Floodplain
Deposition (cm2/yr)
Mean(107oCI, 90%CI)
1
12200
85.1 (-23.8, 216)
-25.6 (-135, 80.8)
59 (35.7, 81.8)
2
6428
-360 (-552, -166)
-25.6 (-135, 80.8)
59 (35.7, 81.8)
3
4341
-628 (-772, -483)
195 (-954, 490)
508 (296, 722)
4
1805
-1410 (-2620, -609)
2408 (317, 5274)
1572 (887, 2293)
5
641
-562 (-796, -330)
2408 (317, 5274)
1572 (887, 2293)
B) Density Measurements
Stream
Order
Streambank
density (g/cm3)
Channel bed density
(g/cm3)
Floodplain density (g/cm3)
Mean(10% CI, 90% CI)
1
1.22 (1.01, 1.42)
1.85 (1.34, 2.33)
0.80 (0.57, 1.0)
2
1.11 (1.00, 1,21)
1.85 (1.34, 2.33)
0.70 (0.50, 0.89)
3
1.10 (0.98, 1.22)
1.40 (0.50, 2.4)
0.76 (0.58, 0.95)
4
1.06 (0.94, 1.18)
0.70 (0.52, 0.90)
5
1.10 (1.03, 1.16)
0.72 (0.48, 0.96)
58
-------
A Manual to Identify Sources of Fluvial Sediment EPA/600/R-16/210
Box 8. Continued.
C) Percent Silt and Clay
Stream
order
Streambank
Channel bed
Floodplain
Mean(107oCI, 90%CI)
1
40.9 (35.3,
47.3)
2.3(0.4, 0.5)
44.2(34.5, 55.0)
2
30.4 23.2, 39.1)
13.8(2.5, 28.0)
50.0(39.2, 61.9)
3
38.7 (33.2,
44.6)
1.4(0.5, 2.4)
47.6(40.1, 54.9)
4
40.9 (28.7,
53.2)
6.3(0.3, 12.4)
50.4(39.0, 61.7)
5
44.9 (40.0,50.0)
6.3(0.3, 12.4)
55.6(46.3, 65.1)
D) Final Contribution From Channel Elements.
Stream Order
Net contribution from
streambanks (Mg/yr)
Net contribution
from the channel bed
(Mg/yr)
Net deposition on floodplains (Mg/yr)
Mean(107oCI, 90%CI)
1
1040(-309; 6,720)
-11.8(-63.6,40.4)
508(2.36,787)
2
1560(-2490; -746)
-41.6(-224,137)
262(124,414)
3
-2330(-3000,-1660)
-25.2(-124,66.4)
1600(790,2420)
4
-2200(-4210,-1000)
577(-279,1530)
2000(888,3170)
5
-356(-511,-241)
205(-99.1,545)
803(345,1280)
Total from all channel
elements
-5400(-8090,501)
703(-790.2320)
5180(2380,8070)
E) Upland Erosion
Land use
Average erosion (t/ha)
Percent fines
Erosion (Mg/yr)
Agriculture
-19.0 (-22.3, 16.0)
38.2 (35.4, 41.2)
-54,800 (-65,100; -45,300)
Forest
-1.45 (-2.38, -0.65)
40.2 (32.3, 48.0)
-2030 (-2033 ,-2027)
Total upland
-56,800 (-67,100; - 47,300)
-56,800 (-67,100; -47,300)
F) Pond Storage
Total sediment storage (Mg/yr)
Ponds
932 (901, 956)
G) Final Sediment Budget
FINAL SEDIMENT BUDGET (Mg/yr)
47,000 (-49,800; -41,000)
H) Fine-Grained Suspended Sediment Computed at the Watershed Outlet = 5,450 Mg/yr.
Box 8. Continued.
59
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
Combining all erosion and deposition (storage) measurements equaled 47,000 Mg/yr. This indicated
that the sediment budget did not balance the fine-grained sediment mass leaving the watershed measured
at the streamflow-gaging station (5,450 Mg/yr). The difference in sediment (41,500 Mg/yr) was
attributed to measurement error and to sediment that went into unmeasured storage elements.
Overestimation of erosion may also be related to the period in which the 137Cs method estimates erosion
and deposition, which is a period of over 50 years. The channel and floodplain measurements in the
sediment budget were taken over a 3-year period. Large storms such as hurricanes, which cause
significant amounts of erosion that would have occurred over the 50 years, would be reflected in the
137Cs estimates. Another factor that would lead to higher erosion rates in agricultural areas over the
historical period compared to the period of study would be the recent implementation of agricultural
conservation practices to reduce erosion such as no-till, contour plowing, and vegetated buffers. If these
practices were recently established on agricultural lands, the reduction in erosion may not be reflected in
the 137Cs-derived estimates of long-term erosion.
Table 6. Rules of how uncertainties are propagated.
Addition/Subtraction
z — x ±y
A z = V(A*)2 + (Ay)2
Multiplication
z — xy
<
II
N
<1
(t)2 + ©2
Division
N
II
^ 1 *
X
A z — —
y
J
(t)' + (t)2
Power
z = xn
Az =
|n|xn-1Ax
Multiplication
by a Constant
z — cx
Az - |c|Ax
Function
z = fix,y)
Az -
A
©
2 (df\2
(Ax)2 + ^j (Ay)2
3.5.7.2 Displaying Budget Results
Diagrams of sediment budgets are usually the most useful for managers to help understand the
magnitude and inputs of sediment. Some simple examples were given in Figures 5 and 6. Depending
on the level of detail of the budget, data can be displayed along a longitudinal continuum with the width
of the arrows representative of the amount of loading associated with each source and sink.
60
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
3,5,7,3 Incorporating Management Practices
Once the initial sediment inventory is completed, what-if scenarios can be developed for various
management techniques spanning uplands to channels, assuming that the hydrology has not been altered.
This can be expanded to particulate-bound phosphorus sources and sinks as well. For example, if bank
erosion or valley side failures were stopped, the potential reduction in loading can be calculated.
Likewise, if a section of stream would lose its floodplain connection, it can be determined how much
sediment would continue to be transported downstream.
?' » rrr. hr f h ^ rprinting
Recent advances have been made in developing field-based approaches to identify sediment sources in
watersheds using the sediment-fingerprinting approach (Williamson et al., 2014). This part of the
manual discusses how to develop a sediment fingerprinting study, including collection, laboratory
analysis, sediment source apportionment, and error analysis.
The sediment-fingerprinting approach provides a direct method for quantifying watershed sources of
fine-grained suspended sediment (Gellis and Walling, 2012) (Fig. 25). This approach entails the
identification of specific sources of sediment through the establishment of a minimal set of physical
and/or chemical properties, i.e., tracers that uniquely define each source in the watershed. Suspended
sediment collected under different flow conditions exhibits a composite, or fingerprint, of properties that
allows them to be traced back to their respective sources. Tracers that have been used in sediment
fingerprinting studies are shown in Tables 7 A and 7B.
Erosion of source A
Catchment
precipftatron and
runoff
Erosion of source 8
Routing to channel and
r-edimer-t mixing
Suspended
sediment load
Erosion of scu.rce C
Comparison of source materia' and suspended
seciiroert samples usirg fingerprint properties
Sed'mept source apportionment:
Figure 25. Outline of the sediment fingerprinting approach (from Walling and Collins,
2000).
bl
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
Table 7A. Example of tracers that have been used in sediment fingerprinting studies. (See Miller et ai, 2015, Table
2.1 for a more extensive list of tracers used.)
Tracers used
Reference
Mineralogy
Motha et al., 2003; Gingele and De Deckker, 2005
Radionuclides
Walling and Woodward 1992; Collins etal., 1997;
Nagle et al., 2007; Evrard et al., 2011
Trace elements
Devereux et al., 2010; Mukundan et al., 2012;
Gellis et al., 2015
Stable isotope ratios
Papanicolaou etal., 2003; Fox and Papanicolaou,
2008; Stewart et al., 2014
Magnetic properties
Foster et al., 1998; Slattery et al., 2000; Hatfield
and Maher, 2009
Color
Krein et al., 2003; Barthod et al., 2015
Table 7B. Example of 38 elemental metals used in elemental analysis for fingerprinting sediment (Gellis et ai, 2015).
Silver (Ag)
Cadmium (Cd)
Potassium (K)
Phosphorus (P)
Selenium (Se)
Vanadium (V)
Aluminum (Al)
Cerium (Ce)
Lanthanum (La)
Lead (Pb)
Strontium (Sr)
Yttrium (Y)
Arsenic (As)
Cobalt (Co)
Lithium (Li)
Rubidium (Rb)
Sodium (Na)
Zinc (Zn)
Barium (Ba)
Chromium (Cr)
Magnesium (Mg)
Antimony (Sb)
Niobium (Nb)
Beryllium (Be)
Cesium (Cs)
Manganese (Mn)
Gallium (Ga)
Titanium (Ti)
Bismuth (Bi)
Copper (Cu)
Molybdenum (Mo)
Thorium (Th)
Thallium (TI)
Calcium (Ca)
Iron (Fe)
Nickel (Ni)
Scandium (Sc)
Uranium (U)
The steps in sediment fingerprinting are shown in Fig. 26. Potential sediment sources in the watershed
are identified in the same procedure as was used in the sediment-budget analysis (Sections 3.2; 3.4).
Sediment sources include but are not limited to agriculture, forest, construction sites, urban sources,
channel banks and beds, drainage ditches, and floodplains. Target sediment includes suspended-
sediment, bed sediment, floodplain sediment, and reservoir or lake sediment (Miller et al., 2015).
62
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
Delineation of sediment-source areas and
number of samples needed
Source material sampling
Drying and sieving to < 63 |im
Drying and sieving to < 63 p.m
Recovery of sediment from
bulk samples
Type of target sampling
program
Fingerprint property analysis
Fingerprint property analysis
Source ascription
Comparison of source and sediment samples using quantitative procedures
Figure 26. Outline of the sediment fingerprinting sampling procedure
iceir Selection
The tracers used in sediment fingerprinting are numerous (Tables 7 A and 7B) (Miller et al., 2015). It is
rare to know a priori which tracers will be able to discriminate between watershed sources. Most
sediment fingerprinting studies use elemental analysis, where for a relatively small cost, 20 or more
elemental results are provided. Several studies have used stable isotope analysis that reflects the organic
content of the sediment. Fallout radionuclides (137Cs, 210Pbex, 7Be) have been shown to discriminate
between channel and upland sources (Matisoff el al., 2005). Other properties that have been used to
distinguish sediment sources include color (Krein et al., 2003), magnetic properties (Slattery eial.,
2000), and mineralogy (Motha et al., 2003). Miller et al. (2015) provide a review of the tracers used in
sediment fingerprinting studies.
Elemental analysis involves inductively coupled plasma combined with mass spectrometry (ICP-MS)
and/or inductively coupled plasma optical emission spectrometry (ICP-OES). These methods are
described online at http://www.epa.gOv/sam/pdfs/EPA-200.7.pdf and
http://www.epa.gOv/sam/pdfs/EPA-200.8.pdf. Stable isotope analysis determines the ratio of r)(l5N/l4N),
abbreviated as r)l5N, and r)(l3C7l2C), abbreviated as r)l3C. Laboratory procedures for stable isotope
analysis can be found at Revesz et al. (2012). Information on the analysis of radionuclides can be found
at http://www2.epa.gov/radiation/marlap-manual-and-supporting-documents.
I rget Sampl flection
The choice of the type of target sediment used to apportion sediment depends on the objectives of the
study. If understanding how sediment sources change through storm events or between events is of
63
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
interest, then storm samples of suspended sediment should be collected (Mukundan etal., 2012; Gellis
et al., 2015. In many studies, the objective is to obtain a target sample that is representative of long-
term conditions, on the order of years. Suspended sediment can be sampled over the course of one or
more years. Bed sediment can be a surrogate for watershed derived fine-grained sediment (Miller and
Miller, 2007) and has been used in several studies to source sediment (Collins and Walling, 2007A, B).
Bed sediment reflects sediment that is eroded and deposited over several events and can be used to
source sediment over time periods of weeks to months (Collins and Walling, 2007A, B). Because flow
and sediment conditions change seasonally, bed sediment should be sampled several times during the
year. Floodplain sediment, which has been used in sediment fingerprinting, is deposited during larger
flow events that may occur at a frequency of years (Owens et al., 1999; Miller et al., 2015). Lake cores
have been used for understanding sediment sources even further back in time (Foster et al., 1998; Pittam
et al., 2009). In the collection of suspended, bed, floodplain, and lake sediment, it is assumed that the
sediment is representative of conditions for the entire watershed.
Prior to planning for the target-sediment sampling design, the tracers that will be analyzed should be
known. Therefore, it is important to contact the laboratory where the analyses will be performed to ask
questions pertaining to sample mass and holding requirements, such as should the samples be
refrigerated? In each of the target-sediment collection schemes, it is necessary to obtain enough mass
for tracer analysis. Different types of tracers have different mass requirements. Elemental analysis and
stable isotopes, for example, require sediment on the order of 1 to 5 grams. Radionuclides may require
up to 30 grams. Therefore, it is important that the sediment sampling technique employed provide the
needed mass. In addition, one should consider the materials used in constructing the samplers. For
example, if the sediment is to be analyzed for metals, the samples should not contain metal. Holding
times of the sediment samples are also important as certain elements, such as short-lived radionuclides
(i.e., 7Be), may lose activity over time.
Commonly, suspended sediment is chosen as the target sediment. Suspended sediment can be collected
using several approaches: manual samples (Edwards and Glysson, 1999), automatic samplers (Gellis et
al., 2015), and passive samplers (Phillips et al., 2000) (Fig. 27). Manual samplers can be used during
storm events to collect suspended sediment for source analysis. Automatic samplers consist of a
peristaltic pump and 24 1-liter (L) clean, plastic storage containers. Intakes for these samplers are
placed mid-stream and at mid-depth. The placement of the intake is based on a site-by-site inspection of
the stream to ensure that the intake is placed in a good transport reach. Backwater areas, or placing the
intake too close to the bed or near obstructions (boulders, bridge piers, trees, etc.), should be avoided.
Automatic samplers are triggered to sample at a preset river stage and will sample at preset times.
Since most sediment is transported on the rising limb of the hydrograph, it may be useful to have the
first samples collected relatively close together. Storm durations vary, and it may be helpful to have the
last samples further out in time. If historical streamflow records exist at the study site or at nearby
watersheds, storm hydrographs should be examined to develop a sample timing scheme.
64
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
Figure 27. (A) Photo of automatic sampler intake at Linganore Creek, Maryland, (B)
Example of passive samplers used in Big Soos Creek, Washington. Four tubes were
installed, each in pairs. [Photos taken by A. Gellis, 2009]
At selected sites, passive samplers, as described in Phillips (2000), can be used to collect suspended
sediment over time. The passive sampler is secured to two posts that are firmly hammered into the
channel bed. To adjust for changing flow conditions, the passive sampler can be moved up and down on
the posts using hose clamps and other fasteners that can be removed. The passive sampler should be
placed in the channel cross section where the sampler will obtain a representative sample. The sampler
can be placed at an elevation between baseflow and bankfull flow or installed submerged. Because
baseflow is subject to seasonal changes, the passive sampler may be moved accordingly. Samples are
collected after storm events, and the contents of the sampler (approximately 8 L of sediment and water)
should be emptied into a clean plastic bottle (-20 L).
Another consideration in the sampling design is the expected suspended sediment concentration of the
stream during high flows. Low suspended sediment concentration streams (-100 mg/L and lower) may
require additional samples to obtain the necessary amount of mass. When using manual and automatic
samplers, it is common to combine samples that were collected over the storm hydrograph, or separate
the samples based on position on the hydrograph (e.g., rising versus falling limb). If passive samplers
are used, several tubes should be placed in the stream to obtain the necessary mass as well as to provide
a backup should some samplers be destroyed during high flows (Fig. 27B). In streams with low
suspended sediment concentrations, the passive samplers may have to be deployed over several events
to obtain the necessary mass.
Bed material for sediment source analysis is typically found in two portions of the channel: 1) fine-
grained channel deposits that may drape coarser bed material, and 2) within the interstitial spaces of
coarse-grained bed material. Fine-grained channel deposits are often found in pools and backwater
areas of the stream and can be sampled with a coring device. Because the fine-grained sediment has
high plasticity, using a coring device that creates a vacuum is necessary to hold the fine material intact.
Sometimes simply capping the coring device will create enough suction to retrieve the sample.
Generally, the top few centimeters of the fine-grained sediment are sampled. Care should be taken not
to sample channel material immediately below an eroding bank.
65
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
Bed material is scoured and deposited during storm events with the depth of scour proportional to the
velocity and water depth. Bed material is mobilized during high flows. During the recessional portion
of the hydrograph, fine-grained sediment becomes deposited in the interstitial spaces between the course
material. The fine-grained sediment found in the coarse-grained material can be sampled by isolating
the bed with a cylinder and sampling the bed material in the cylinder (Fig. 27C). Collins and Walling
(2007b) sampled the fine-grained sediment in the channel bed by isolating the channel bed with a
cylinder, stirring the bed inside the cylinder, and collecting the slurry. Only the bed material on the
surface is stirred. The stirring rod made of PVC can be 1 m in length and 1-2 cm in diameter. Sediment
for sourcing in lakes and other impoundments, as well as floodplains, can be sampled with coring
devices (Pulley etal., 2015).
4,2,1 Target Sample Preparation
Typically, the samples obtained from the automated sampler are processed for suspended-sediment
concentrations to determine suspended sediment loads, although this does not have to be a necessary
objective of sediment source analysis. A decision is made whether to combine bottles or analyze them
separately. This decision is based on the objectives of the study and the mass needed for analysis. If
bottles are combined, this should be performed as soon as they return from the field. Processing of
suspended sediment may include centrifuging or filtering the water-sediment mixture. Centrifuging is
preferable because the lab will return the dried fraction of sediment (< 0.063 mm) from each bottle in a
vial.
If the lab uses filters for suspended sediment concentrations, the sediment is scraped or rinsed off the
filters with de-ionized water to remove the dried sediment that is collected in a glass bowl(s). The glass
bowl(s) is dried at 65°C for 24-48 hours. In some circumstances, the automatic samplers are installed
for the sole purpose of obtaining suspended sediment for sediment source analysis. Bottles from the
automatic sampler are composited, refrigerated, and the water allowed to settle 72 hours. When the
water is clear to the bottom of the container, the clear water is pumped out, and the slurry is transferred
to a non-metal drying bowl using de-ionized water. The slurry is then wet-sieved (see below).
After drying, a sufficient mass needed for analysis is weighed, ground with a ceramic mortar and pestle,
and wet-sieved through a 63-micron polyester sieve using de-ionized water. ASTM D3977-97 (2002)
Method C, wet-sieving filtration, is used to separate sand and coarser material from finer material. The
water-sediment mixture is captured in a glass bowl(s) and dried at 60°C for 24-48 hours. After drying,
the sediment in the bowl is scraped with a plastic blade, weighed, and placed in a plastic vial for
shipping. The coarse material is saved, dried, and weighed to obtain the percentage of fines (<0.063
mm) and coarser material.
The sediment-water mixture from a passive sampler is placed in a cooled area and left for several days
in order to let the sediment settle to the bottom. The samples are ready to be decanted when you can see
the bottom of the container. This may occur when the water is still a little cloudy. The water from the
passive sampler is decanted until only a slurry is present. The slurry is emptied into a glass bowl(s) and
dried at 65°C for 24+ hours until dry. After drying, a sufficient mass needed for analysis is weighed,
ground with a ceramic mortar and pestle, and wet-sieved through a 63-micron polyester sieve using de-
ionized water. The water-sediment mixture is captured in a glass bowl(s) and dried at 65° C for 24+
hours until dry. After drying, the sediment in the bowl is scraped with a plastic blade, weighed, and
placed in a plastic vial for shipping. The coarse material is saved, dried, and weighed to obtain the
66
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
percentage of fines (<0.063 mm) and coarser material.
Bed, floodplain, and lake sediment may have less water in the sampled sediment and can either be stored
in a refrigerated area or immediately wet-sieved. Since these samples have greater mass, depending on
the final mass needed, a subsample is combined with deionized water to create a slurry. Sample
preparation for the slurry follows the same procedure for other slurry samples as described above.
'! "• 45 degrees) with a large portion of the bank
faces (>50%) exposed and not covered with vegetation. Each streambank is sampled from the bottom to
the top of the bank face with a plastic (non-metal) hand shovel. All samples are composited into one
sample. A 10-L plastic container can be used to hold the composited sample. The composite sample is
mixed well with a plastic shovel, and a subsample is loaded into a plastic bag. Roots, leaves, twigs, and
other organic debris can be removed. The total mass sampled from an eroding bank should be a
minimum of 50 to 100 g.
4.3.1 Source Sample Preparation
Upland and bank samples should be refrigerated or put on ice after sampling. The samples are wet-
sieved following the same procedure for slurries of fluvial sediment. Further quantification on sediment
samples and cores involves particle size determinations and organic matter content. These methods are
discussed in detail in Guy (1969) and Burt (2009).
4.3.2 Grain Size - Laboratory Analysis
Either grain size or surface area can be used to correct for grain size. It is common to analyze for grain
sizes less than 63 microns (Collins et al., 2010; Lamba et al, 2015); however several researchers suggest
that analyzing sediment to a narrow range of particle sizes (Hatfield and Maher; 2009) or only to <10
microns (Wilkinson et al, 2013; Laceby et al., 2015) may improve results and reduce error.
After the sample is wet-sieved to the preferred grain size, the surface area or grain size of the sediment is
determined for the portion of sediment that is less than 0.063 mm. Universities and private laboratories
have equipment that can analyze fine-grained sediment for surface area and grain size. Pipette analysis
is considered the standard for determining grain size (ASTM D422-63, 2007), but laser and optical
67
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
methods are acceptable (Konert and Vandenberghe, 1997). The facilities that will conduct the size
analysis should be contacted to determine what type of preparation is needed for the sediment prior to
size analysis. For example, prior to size analysis using a laser diffraction (LISST-100X with mixing
chamber) (Pedocchi and Garcia, 2006), Gellis (2014) prepared the sediment samples for size distribution
analysis by disaggregating the sample in a sodium hexametaphosphate solution that was sonicated for 5
minutes, shaken for 16 hours, and then analyzed on the LISST-100X to determine the median particle
size (D50) (Wolf et al., 2011). Typically, the median surface area and median grain size of the sediment
are used in the grain size correction procedure and can be reported by size-analysis outputs and
laboratory reports.
4.3.3 Organic Content
Enrichment of loss of organic matter during the erosion cycle can also affect the tracer property, and
correcting for differences in organic content between the source and target samples should be
performed. Organic content of the sediment can be determined using loss-on-ignition (ASTM D7348-
13, 2013).
4,3,4, Field a Moratory Quality Assurance
4,3,4,1 Precision and Accuracy
Precision is the degree of agreement among repeated measurements of the same characteristic, or
parameter, and gives information about the consistency of methods. Accuracy is a measure of
confidence that describes how close a measurement is to its "true" value. Duplicate measurements will
be performed for sediment source and fluvial samples.
Field analytical precision will be evaluated by the relative percent differences (RPDs) between field
duplicate samples and/or duplicate readings using the following formula:
RPD = [(Rl - R2)/{(R1 + R2)/2}] x 100 (12)
where: Rl = the larger of the two duplicate values
R2 = the smaller of the two duplicate values
The values of the RPD are intended to provide information on the variability of field samples.
Differences that are greater than 10% are flagged but not discarded. Replicate analyses for the two
samples are averaged for the final sediment sourcing analysis.
4.3.4.2 Data Representativeness
Representativeness is the extent to which measurements actually represent the true environmental
condition. It is the degree to which data from the sampling accurately represent a particular
68
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
characteristic of the watershed that is being tested. Representativeness of samples is ensured by
adherence to standard field sampling, measurement, and laboratory protocols. The design of the
sampling scheme and number of samples for this plan provide representativeness of the part of the
watershed being monitored. The representativeness of the data is dependent on: 1) the sampling
locations, 2) the number of samples collected, and 3) the sampling procedures. Site selection and use of
only approved analytical methods will ensure that the measurement data represents the conditions at the
site. The goal for meeting total representation of the targeted drainage area is tempered by the
availability of time and funding. Representativeness will be measured with the completion of sample
collection in accordance with the approved Quality Assurance Project Plan (QAPP) (USEPA, 2001).
4.3.4.3 Data Comparability
Comparability is the degree to which data can be compared directly to data from similar studies. The
comparability of the data produced is predetermined by the commitment of the staff to use only
approved procedures as described in this manual. Comparability is also guaranteed by reporting data in
standard units, by using accepted rules for rounding figures, and by reporting data in a standard format.
The methods used to determine sediment sources are methods that are established in the literature.
By its very nature, soil is highly heterogeneous and variability among samples is expected (ITRC, 2012).
Comparability will be checked by collecting replicate samples (about 10% of sediment source samples).
Replicate samples are collected by sequentially taking two field samples for each analysis from the same
source. Comparability goals for field replicate samples are to have results within 10% of each other.
Samples that exceed this are not discarded but are flagged. The replicate sample is not used in the
analysis and the two samples, the field and replicate, should not be averaged (ITRC, 2012). Unless there
is clear evidence that the sample has been compromised, the laboratory value should be used.
4.3.4.4 Statistical Methods
Several analytical and statistical steps are used to determine which tracers are most significant in
defining sediment sources (Fig. 28) as follows:
• Determine if there are outliers for each tracer in each source type.
• Test if tracers in each source group need to be corrected for size differences between the source
samples and the fluvial sample.
• Test if tracers in each source group need to be corrected for organic content differences between
the source samples and the fluvial sample.
• Perform a bracket test of size- and organic-corrected fluvial and source samples for each tracer.
• Determine the optimum number of tracers that discriminate among the sources using stepwise
discriminant function analysis (DFA).
• Identify source percentages using a mixing model on the final set of tracers.
• Perform error analysis using Monte Carlo and other designs.
To assist the user in the statistical steps outlined above for sediment fingerprinting, the USGS has
designed the Sediment Source Assessment Tool (Sed SAT) (Gorman-Sanisaca etal., In Preparation).
69
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
Statistical methods to allocate sediment sources
• Imputing non-detects
• Outlier removal
• Size and organic corrections
• Bracket test
• Stepwise Discriminant Function Analysis
• Multivariate unmixing model
• Error Analysis
Figure 28. Summary of statistical operations used in sediment fingerprinting to apportion sediment
sources.
4.4.1 Outlier Test
The presence of outliers can lead to errors in data analysis and statistical conclusions (Helsel and Hirsch,
1992). The first step in the statistical procedure is to remove outliers. In each source group, each tracer
is tested to determine if it has a normal distribution using the Shapiro-Wilk test at a 95% confidence
interval (H0 = samples that are random and come from a normal distribution). All variables that were
not normally distributed are tested again for normality after transformation using a log, power, square
root, cube root, inverse, and inverse square root function (Helsel and Hirsch, 1992). The best
transformation for normality is selected (if necessary aided by visual analysis of histograms), and the
tracers are transformed. The average and standard deviation within each source group for each
transformed tracer are determined. If the tracer value for a given source sample exceeds three times the
standard deviation more or less than the average value, this sample is considered an outlier and the entire
sample is removed from further analysis (Wainer, 1976).
4.4.2 Correcting Source Tracers for Sediment Size and Organic Content Variability
The property of a sediment tracer not only depends on source material but also on grain size and organic
content (Collins etcil., 2010; Horowitz, 1991). As sediment is eroded and transported through the
watershed over time, grain size may change. Generally, sediment delivered out of a watershed has a
finer grain size compared to the source areas (Walling, 2005). The finer grain sizes have potentially
greater surface area to sorb constituents and consequently have a higher tracer concentration.
Conversely, iron oxides that develop on coarser sediment in the silt range may also contain more sites
for constituents to sorb onto resulting in higher concentrations. Organic matter on sediment can also
result in additional sites for sorption of tracers or include tracer elements within organic molecules.
A correction procedure using loss on ignition (LOI) values or total organic carbon (TOC) concentration
of the sediment can be applied to account this effect. It is worth noting that while some researchers
agree that sediment should be corrected for organic content differences, others report that this results in
'overcorrecting' of sediment tracer concentrations (Koiter etcil., 2013; Smith and Blake, 2014).
70
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
To assure that tracer concentrations from sediment sources are comparable to concentrations in target
sediment based on grain size and organic content, the source samples need to be corrected for grain size
and organic content. Gellis et al. (2015) applied a regression approach to correct for grain size
differences between the target and source samples.
For each source group, linear regression was used to determine if the relation of median grain size (D50)
or TOC to a given tracer's concentration was significant. Tracer concentrations are corrected first for
grain size and then for organic content. An example of how tracer concentrations are corrected using
D50 is shown in Figure 29. Corrections of organic content follow the same approach.
100
• Agriculture samples
Regression of agriculture samples
• Fluvial samples
Corrected agriculture samples
Fluvial mean
I I Agriculture mean
80
60
40
20
2
4
6
8
10
12
D50 (mm)
Figure 29. Example of how the size-correction factor is applied to a source group. In
this example, the Dsofrom agriculture samples was regressed against the tracer
lithium (Li). The line of best fit of agriculture D50 and agriculture lithium is negative,
showing the concentration of lithium in the agriculture samples decreases as D50
increases. The average of D50 of fluvial samples is finer than the average D50 of
agriculture samples. To correct for the differences in size, lithium should be
adjusted to be higher, [mg, milligrams; g, grams; D50, median grain size of fine
sediment; jj.m, microns] (Gellis et al., 2015).
Guidelines used to determine if the relation of D50 or TOC to a given tracer's concentration is significant
include determining that the slope of the regression line is significant (p<0.05) and the residuals area
normally distributed. The Shapiro-Wilk test (H0 (null hypothesis)= samples are not random and do not
come from a normal distribution) is used to determine that the residuals follow a normal distribution.
Plots of residuals compared to predicted values as well as histograms and QQplots of the residuals are
also used to determine if the regression model is reasonable. In a plot of residuals compared to
predicted values, a regression model is considered to be reasonable where the residuals show no
curvature or changing variance (Helsel and Hirsch, 1992).
71
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
The steps used to determine the best regression model are: 1) determine if the relation of untransformed
D50 and TOC compared to each source group's tracer concentration is significant; 2) if no significant
relation is found, then D50 and TOC are transformed using the Tukey Ladder of Powers transformations.
The transformed D50 and TOC are then regressed for each source group's tracer concentrations to find
the best regression model; 3) if after (2) is completed, no significant relations are found, then the tracer
concentration values are transformed using the same transformations applied to D50 and TOC . The
transformed tracer concentration values for each source group are regressed with all possible
combinations of transformed D50 and TOC (including untransformed), and the optimum regression
model is selected. If no significant relation is found with D50 or TOC in (1), (2), or (3), a correction
factor is not applied. It should be noted that the tracers <513C, t>i5N, and %N are affected by the relative
proportions of different kinds of organic matter in the sample, not the total organic matter content.
These tracers, <513C, t>i5N, and N should not be corrected by TOC.
If the regression model of a source group's tracer compared to D50 is determined to be
significant, then a correction factor is applied to the tracer as follows:
Cn = {Tiiin) - [(050n - FD50)*m]}A (13)
where:
Cn = untransformed tracer after size correction;
TU(„) = value of tracer (J) (if transformed in source group («);
D50„ = the mean D50 of samples in source (n) (if necessary the values of D50 are transformed and a mean
of the transformed D50 is determined);
FD50 = the mean D50 of target samples (if necessary, the target samples are transformed as the 1)50
samples in source («) and a mean of the transformed variables is determined);
m = slope of regression line of tracers in source group (n) (if necessary, tracer is transformed) versus D50
of source group (n) (if necessary, D50 is transformed); and
A = if the tracer is transformed, the final corrected tracer is untransformed.
If the regression model of a source group's tracer concentration values compared to TOC was
determined to be significant, then a correction factor is applied to the tracer as follows:
C.=K,,-[(CS„-CF)«m]}- (M)
where:
Co = untransformed tracer after organic correction;
TU(„) = original value of tracer (/') (if necessary,) in source group («);
CSn = average TOC of source group (n) (if necessary, the values of TOC are transformed and a mean of
the transformed TOC is determined);
CF = average carbon content of target samples (if necessary, the target samples are transformed by the
same transformation as the TOC samples in source (n) and a mean of the transformed
variables is determined);
m = slope of the regression line of tracers in source group (n) (if necessary, tracers are transformed)
versus TOC of source group (n) (if necessary, TOC is transformed); and
A = if the tracer is normalized by a transform, the final corrected tracer is untransformed.
72
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
Equations 13-14 are used with the mean of a source group's D50 as well as the mean D50 of all target
samples. Corrections can also be made using an individual target sample, where the difference of each
target sample's D50 or TOC and the mean source D50 or TOC (for a given source group) is determined
and a correction is applied to each source sample. If an individual target sample is used the mean in
equations 13-14 are replaced with the individual target sample's D50 and TOC.
When determining the size-correction factor, it is important that the results be unbiased. If a non-linear
model is used, a bias in the estimation can occur (Koch and Smillie, 1986). The bias occurs when
tracers that are transformed are then corrected using Eq's. 13-14 and then untransformed. Bias
correction factors (B) should be applied to the corrected, untransformed tracer concentrations only for
those tracers that are transformed prior to correction. If D50 or TOC are transformed in the
determination of correction factors but the tracer concentration is not transformed, no bias correction is
required. Bias correction factors (B) are determined by using the transformed tracer concentration per
the equations given in Table 8 (Stuart and Ort, 1991). The size- or organic-corrected tracer
concentration is divided by B to calculate the final untransformed tracer value.
Table 8. Equations used to correct for bias in untransforming the tracers (Gellis et ai, 2015).
[B, the bias correction factor; f(y), the transformed value of the tracer; D50, the median grain size of the sediment;
Xs, the mean value of all transformed D50 or total organic carbon (TOC) for a given sediment source; Xf, the mean
<"> 2
value of the transformed D50 for the target samples; b, is the slope of the line of best fit; db. is the standard error
of the regression of D50 of TOC compared to tracer concentration; exp, exponential]
Transformation
B
Squared
i+^t/x.v)F)&r20s-%)2<^
Square root 1 + [/(, ) - (.Vj - x>)ir2(.vs - .yf)-<7?
Cube root l + 3[f(?)-(xf - 2{xs - %)2€Fg
Iii\ eise 1 — [f ^ v") — C xs — xF )b ] — .\j?) 6|
Inverse square
root
Log
0A[C.rs-xF)26-j/2]
73
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
ket Test
A requirement of sediment fingerprinting is that the fluvial tracers must be conservative and not change
during transport from the source to the sampling point. Consequently, the next step in a statistical
analysis is determining that for a given tracer, the fluvial samples are within the range of the equivalent
values obtained for the potential sources (Gellis and Walling, 2011) (Fig. 3). The bracket test is an
important prerequisite before further statistical analyses are performed. Any tracers that do not satisfy
this constraint within the measurement error (10% of each fluvial sample's tracer value) are considered
to be non-conservative and are removed from further consideration. The bracketing test is performed on
tracers after the particle size and organic correction factors are applied.
¦owf criminant Function Analysis
Collins and Walling (2002) and Collins et al. (1997) have suggested that a composite of several tracers
provides greater ability to discriminate between sources than a single tracer. To create the optimal
group of tracers, a stepwise DFA was used to select tracers after size and organic corrections were
applied (Fig. 28). This procedure assumes normality among the variables being analyzed; thus, all
variables used in the DFA were tested for normality using the Shapiro-Wilk test (H0 = samples are
random and come from a normal distribution). All variables that are not normally distributed at a 95%
CI should be tested again for normality after transformation using a log, power, square root, cube root,
inverse, and inverse square root function (Helsel and Hirsch, 1992).
The best transformation for normality is selected (if necessary), and stepwise DFA is performed on the
normalized data. Stepwise DFA incrementally identifies which tracers significantly contribute to
correctly differentiating the sediment sources and rejects variables that do not contribute based on the
minimization of the computed value of the variable Wilks' lambda (Collins et al., 1997). A lambda
close to 1.0 indicates that the means of all tracers selected are equal and cannot be distinguished among
groups. A lambda close to zero occurs when any two groups are well separated (within group variability
is small compared to overall variability). Thus, the model selects a combination of tracers that provide
optimal separation, meaning that no better separation can be achieved using fewer or more tracers. The
statistical program Statistical Analysis System (SAS) was used in stepwise DFA (SAS Institute, 2004).
A probability value of 0.01 was used to determine significance in the stepwise DFA.
iputation of Source Percentages
The final step in the statistical analysis is determining the significant sources of sediment using an
unmixing model (Fig. 3; Eqs. 3, 4, and 5; all modified from Collins et al., 2010). The set of tracer
values that are determined from the stepwise DFA are used in the mixing model but with the particle
size and organic correction factors applied. The mixing model does not use data transformed for
normality, but it does use the values that have been adjusted for D50 and TOC.
74
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
and
/c ] w.
(15)
(16)
where:
Ci= concentration of tracer property (J) in the suspended sediment collected during storm
events;
/', = the optimized percentage contribution from source category (5);
Ssi= mean concentration of tracer property (7) in source category (5) after size and TOC
correction factors are applied in source category (5);
Wi= tracer discriminatory weighting;
n = number of fingerprint properties comprising the optimum composite fingerprint; and
m = number of sediment source categories.
Collins etal. (2010) applied the particle size and organic corrections factors directly in the mixing
model. In this modified version (Eq. 3), the set of tracer values that were determined from the stepwise
DFA were used in the mixing model with the particle size and organic correction factors applied.
The mixing model iteratively tests for the lowest error value using all possible source percentage
combinations. A Ps step of 0.01 is used in the source computations. The tracer discriminatory
weighting value, Wi, is a weighting used to reflect tracer discriminatory power in Eq. 3 (Collins et al.,
2010). This weighting is based on the relative discriminatory power of each individual tracer provided
by the results of the stepwise DFA and ensures that tracers that have a greater discriminatory power are
optimized in the mixing model solutions. The weighting for each tracer that passed the stepwise DFA
test is determined as follows:
ropt (17)
where:
Wi = tracer discriminatory weighting for tracer (/');
Pi = percent of source type samples classified correctly using tracer (/'). The percent of source type samples
classified correctly is a standard output from the DFA statistical results; and
Popt = the tracer that has the lowest percent of sample classified correctly. Thus, a value of 1.0 has a low power of
discriminating samples.
imitations and Uncertainty in Sediment Fingerprinting
A Monte Carlo approach is used to quantify the uncertainty in sediment fingerprinting results produced
by the mixing model (Collins and Walling, 2007). The Monte Carlo simulation randomly removes one
sample from each of the source type groups, and the mixing model is run without these samples. The
Monte Carlo simulation is conducted 1,000 times on each target sample. For each of the 1,000
iterations, the average, minimum, and maximum sediment-source percentage for each source are
determined.
75
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
Another estimate-of-error assessment available uses the source samples and runs them through the
mixing model. Ideally, putting source samples through the model should result in the output indicating
100% of that source. For example, using a bank sample should result in the mixing model showing
100% bank-derived sediment. However, because of variability in source tracer properties and possible
deposition from other sources, the results may not always be 100% accurate. For example, the top
portions of a streambank may be an active floodplain that is still receiving sediment from upstream
sources. If a sample reveals a small percentage of its designated source (i.e., <50%), the history of the
sample, i.e., collection and laboratory results, should be examined and a decision made whether the
sample was contaminated and whether it should be kept in the analysis.
13 -Mtalyziiv, r suits
4,5,1 Weighting Sediment Apportionment
Suspended-sediment samples collected using automatic or manual samplers may be collected at several
times during an event as well as for several events. The practitioner has the option of averaging source
apportionment results for each sample or weighting the results by sediment load. It is reasonable to
weight samples by the sediment load for each sample or for each storm (Box 9). In addition, results for
each storm could be weighted by the total sediment mass transport during the study period (Box 10).
The concept of weighting sediment-fingerprinting results can be demonstrated by using an example of
sediment coring. For example, if a target sample for sediment sourcing was obtained by coring a
reservoir that was constructed 20 years ago; the thickness of the deposited sediment in the reservoir
would vary with the transported load of each event over the 20-year period. The larger sediment loading
events would have thicker sediment deposits. Coring the entire sediment package would weight each
event because the higher loading events would have greater sediment thicknesses compared to other
events. The same logic may apply to weighting samples by the sediment transported for that sample or
that event. The larger events transported more sediment and should be weighted (if sediment records
exist) by loadings. If sediment records do not exist, peak flow or some other hydrologic factor may be
used.
76
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
Box 9. Example of weighting sediment fingerprinting results by the load computed for each sample.
Each sample represents an interval of time. Suspended-sediment load computations for each time interval are computed following Porterfield (1977) using
the USGS program GCLAS (Graphical Constituent Loading Analysis System) (Koltun et al., 2006). Each interval suspended sediment load is
divided by the total load of the event to compute the weighted total load (Col B). Fingerprinting results for each sample are found in columns C to E.
These results can be averaged for each sample in the event. The weight of each sample (Col B) is multiplied by each sample's result (columns F, G, and
H). The sums of each column F, G, and H, is the weighted source percentages.
Col (A)
Col (B)
Col (C)
Col (D)
Col (E)
Col (F)
Col (G)
Col (H)
Sample #
Date
Sample
time
Time interval
sample covers
Suspended
sediment
load for
time interval,
megagrams
Weighted-
total load
Sediment Fingerprinting Results
Samples Weighted by Sediment Load
Agriculture
Banks
Forest
Agriculture
Banks
Forest
1
12/11/2008
16:24
15:00-18:48
25
0.01
13
87
0
0
1
0
2
12/11/2008
19:15
18:48-19:48
55
0.03
22
76
2
1
2
0
3
12/11/2008
20:15
19:48-20:48
67
0.03
27
64
9
1
2
0
4
12/11/2008
21:15
20:48-21:48
146
0.07
52
36
12
4
3
1
5
12/11/2008
22:15
21:48-22:48
454
0.23
40
42
18
9
10
4
6
12/11-12/2008
23:48
22:48-00:48
545
0.27
27
61
12
7
17
3
7
12/12/2008
1:48
00:48-02:48
304
0.15
40
60
0
6
9
0
8
12/12/2008
5:51
02:48-08:54
353
0.18
59
33
8
10
6
1
9
12/12/2008
10:00
08:54-12:04
35
0.02
72
25
3
1
0
0
10
12/12-13/2008
14:17
12:04-04:30
10
0.01
51
49
0
0
0
0
Average sources (%)
Weighted source (%)
Total load
1684
40
53
6
40
50
10
78
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
Box 10. Example of weighing sediment fingerprinting results by the sediment load for each
sampled storm event.
Fingerprinting source results
Date event
Suspended
sediment
load,
megagrams
Weight
Agriculture
Banks
Forest
Agriculture
Banks
Forest
10/31/2009
112
0.02
42
56
2
1
1
0
11/12-13/2009
202
0.04
33
66
1
1
3
0
12/11-13/2008
1684
0.36
13
87
0
5
31
0
1/15/2009
45
0.01
59
36
5
1
0
0
2/3/2009
88
0.02
51
42
7
1
1
0
2/14-16/2009
987
0.21
12
79
9
3
17
2
4/15/2009
54
0.01
40
60
0
0
1
0
5/2-3/2009
455
0.10
33
66
1
3
6
0
5/25-26/2009
376
0.08
16
76
8
1
6
1
6/3/2009
41
0.01
41
55
4
0
0
0
6/5/2009
36
0.01
35
62
3
0
0
0
6/13-14/2009
344
0.07
20
80
0
1
6
0
7/12/2009
69
0.01
9
91
0
0
1
0
7/22/2009
88
0.02
12
88
0
0
2
0
9/12-13/2009
109
0.02
6
93
1
0
2
0
Averaging
Weighted
Sum
4690
28
69
3
19
78
3
To include the importance of samples collected during periods of high sediment loading, the
sediment sources determined for each sample should be weighted by the total amount of
sediment transported for that event (storm weighted) using the following equation:
(18)
where:
Svj= storm-weighted source allocation for sediment source (v) (e.g., streambanks.
agriculture, or forest) (in percent) and event (/');
,V.4v,= sediment source allocation (in percent) for source (v) and storm sample (/);
SLi= sediment load for storm sample (/), in megagrams;
SI.,, = total sediment load for event (/), in megagrams. determined using the program
Graphical Constituent Loading Analysis System (GCLAS) (koltun etal., 2006); and
n = number of samples (/) collected during storm event.
79
-------
A Manual to Identify Sources of Fluvial Sediment EPA/600/R-16/210
As previously discussed, because of mass constraints, a storm sample represents one
instantaneous time or several times on the storm hydrograph. To determine a sediment load for a
given sample (SLi), a time interval is assigned for each sample. Discharge is usually computed at
15-minute intervals but can vary. The time interval for each sample is midway between the
previous and next samples. The first sample is assigned a time to the nearest measured interval
(e.g., 15- minute intervals) before that sample and midway to the next sample time. The last
sample is assigned a time midway to the previous sample time and the nearest measured interval
(e.g., 15-minute intervals) after that sample. Each sample's time interval is input into programs
that compute sediment. For the USGS, this is the program GCLAS (Koltun et a I., 2006;
available at http://water.usgs.gov/software/GCLAS/) and a sediment load (mg) was determined
for that time interval. The sum of sediment loads for each time interval is the total sediment load
for the event (Box 10). If sediment sources are determined for several events, the source
percentages for each event can be weighted by the sediment load of that event compared to the
total sediment load of all events (Box 10).
The sediment source weighting for the entire study period (total load-weighted percentage) is
determined using the equation:
SL, i
^71
V F
(19)
where:
TSvj = total load-weighted sediment-source allocation for sediment source (v);
SVj = sediment source allocation for source (v) for event (/');
SLtj = total sediment load for event (/'), in megagrams (Eq. 6);
m = number of events during the study period = 36; and
SLP = total sediment load for 36 events, in megagrams.
A similar approach for representing sediment sources for an entire study period was used by
Walling et al., (1999) and Gellis et al. (2009).
80
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
5,0 Summary
Sediment is one of the leading pollutants in the United States degrading aquatic habitats. In
order to effectively manage sediment and reduce sediment loads, it is necessary to identify the
sources of sediment. In general, watershed sediment sources can be separated into two broad
categories: 1) upland areas of various land cover, and 2) the stream corridor (channel bed, banks,
and floodplain). Differentiating between upland and channel sources is important because
management strategies to reduce sediment differ by source and require very different
approaches. Reducing agricultural sources may involve soil conservation and tilling practices,
whereas channel sources of sediment may involve bank stabilization and channel restoration to
arrest downcutting. The manual provides information on using a sediment budget approach to
understand the erosion, deposition, and delivery of watershed-derived sediment. We also
highlight the sediment fingerprinting approach in apportioning sediment sources. These
approaches can be used to assist in management strategies concerned with sediment TMDLs,
assessing the contributions from various land uses, and contrasting this to the stream channel, as
well as monitoring the effectiveness of management actions to reduce sediment.
A sediment budget is an accounting of the inputs (erosion) and storage (deposition) of sediment
that in theory match the output or sediment transported out of the watershed. The available tools
and approaches used to construct a sediment budget are considerable, and selection will depend
on many factors, such as financial resources and temporal and spatial aspects of the study.
Determining upland erosion and deposition can involve pins, nets, impoundment surveys, the
Cesium-137 approach, GIS analysis, and modeling. Determining stream-corridor erosion and
deposition involves monumented channel surveys, bank pins, floodplain markers, ground-based
and airborne LiDAR, aerial photographs, and modeling.
The first step in any sediment budget program is to decide on the watershed scale and the time
frame of the study. Before measurements are made, it is important to 'get to know your
watershed' by reading published literature, local reports, local maps, and GIS analyses, all of
which help to understand historical and current sediment problems, land cover in the watershed,
and any hydrologic alterations that may be present. Map and GIS interpretations can create
longitudinal profiles of the channel that are used to classify steep (incisional) and shallow
(depositional) reaches. Aerial photographic surveys can identify upland sediment sources and
channel planform width and depth, which may target areas of erosion and deposition. Meeting
with stakeholders and local management authorities is important in understanding management
problems in the watershed, and lines of communication should be established in the early stages
of sediment budget development.
A car or aerial reconnaissance of the watershed is important in understanding the geomorphic
setting of the watershed. Filling out geomorphic assessment forms during the reconnaissance
can aid in developing qualitative assessments of watershed features, such as: incised reaches,
location of steep banks, depositional reaches with bar occurrence, upland sites of bare ground
(e.g., construction sites), riparian canopy density, etc. The information gleaned from the
reconnaissance can help in site selection for monitoring.
81
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
The approaches and tools used to construct a sediment budget generally fall under three
categories: 1) field approach, 2) remote sensing, and 3) modeling. Site selection can be: 1)
judgement based, such as selectively choosing incised reaches, or 2) a random design, where
sediment sources are rasterized in a GIS and chosen randomly, such as agricultural areas, or 3)
based on regulatory requirements, such as monitoring restoration activities. Site selection should
be carefully designed to provide reliable assessments of erosion and deposition.
Elements in the channel corridor that are examined for change over time include the floodplain,
channel banks, channel bars, and the channel bed. Sites for measuring channel elements can be
organized around a reach that is typically of a length to incorporate channel variability in width,
depth, and sinuosity. This is often achieved with a reach that is one meander length.
Establishing two or more cross sections in a reach can capture the form variability. To capture
spatial variability, reaches of varying contributing areas should be monitored. Monitoring by
stream order can in part fulfill this objective.
Field approaches to quantify channel corridor changes include pins, surveys, and ground-based
LiDAR. Some measurements are linear (cm) (i.e. pins), some are cross sectional (m2) (i.e.
channel resurveys), and others can be volumetric (m3) (i.e. a pond survey). Accounting of
erosion and deposition in a sediment budget is displayed as either a volume (m3) or a mass (kg).
Linear changes are converted to cross-sectional area (m2) by multiplying the linear change (cm)
by the width, length, or height of the channel feature (i.e. height of the channel bank). All cross-
sectional area changes (cm2) are averaged for a reach and then averaged for all reaches in that
stream order. The average change in channel features for each stream order (m2) is multiplied by
the length of that stream order (m) to obtain a volume (m3). Multiplying this volume by the
average density (g/cm3) of the channel feature for the selected stream order provides an estimate
of mass (kg). Dividing the mass by the number of days of the study period (kg/day) and
multiplying by 365.25 provides an annual mass (kg/yr) for that land use type.
Current and historical photogrammetric information (aerial photographs and airborne LiDAR)
can be used to quantify channel widening (width, m). The digital photographs need to be
georeferenced into a GIS and changes over time in the distance between bank streambank edges
can be used to calculate streambanks retreat. Airborne LiDAR can be used to quantify changes
in cross-sectional area (width and depth, m2) and volumetric changes (m3) over time. Digital
elevation models created from the LiDAR data are used with appropriate software to quantify
cross-sectional and volumetric changes in channel morphology.
Floodplain deposition can be monitored using permanent markers installed on the floodplain
(e.g., clay pads and tiles) placed along a cross-sectional transect. All measurements of
deposition (mm) are averaged for the floodplain portion of the cross section to compute a change
in cross-sectional area (m2). The cross-sectional area change for each cross section is averaged
for the reach and then averaged by stream order (m2). The length of each stream order (m) is
multiplied by the average cross-sectional area change for each stream order to obtain a volume of
deposition (m3), which is multiplied by the average floodplain density for that stream order
82
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
(g/cm3) to obtain a mass (kg). Dividing the mass by the number of days of the study period
(kg/day) and multiplying by 365.25 provides an annual mass estimate (kg/yr).
In the early stages of sediment budget development, upland areas for source analysis are mapped
out. Upland areas can be classified by land use or soil type. Unpaved roads also fall under the
classification of upland erosion. Upland erosion can be measured using traps and nets to capture
the mass of eroded sediment (g). The mass is normalized by the contributing area to compute a
yield (kg/m2). Pins can also measure ground surface lowering and deposition (cm) and are
arrayed in grids or along transects. The pin measurements are averaged for the plot area (m2) to
obtain a volume (m3) and then multiplied by the average soil density (g/cm3) in the measurement
area to obtain a mass (kg). The mass is divided by the measurement area to obtain a yield
(kg/m2). Yields (kg/m2) are multiplied by the upland areas they represent (i.e., pasture) to obtain
a total mass of change (kg). Dividing the mass by the number of days in the study period
(kg/day) and multiplying by 365.25 provides an annual mass (kg/yr) for that land use type.
For unpaved roads, at each transect the linear changes in the unpaved road surface (cm) are
multiplied by the width of the road (m) to obtain a cross-sectional area (m2). At each reach, all
measurements on unpaved road surface are averaged and multiplied by the total length of roads
to obtain a volume (m3). This volume is multiplied by the average density of roads (g/cm3) to
obtain a mass estimate (kg).
The cesium-137 technique can be used to estimate soil erosion and deposition. It is based on the
principle that non-eroding areas (i.e. the summit of a stable forested slope) reflect the reference
condition of cesium-137. By comparing the inventory of cesium-137 for several reference sites
to other slopes of varying land use, estimates of erosion and deposition can be made using
appropriate models. Hillslopes of selected land covers are cored for soil samples at several
locations along a transect that extends from the summit to the toe. Samples from each location
are analyzed for cesium-137. Cesium-137 displays a characteristic profile with depth on stable
surfaces. For the reference site, it is important to sample at several cm incremental depths to
establish the site as a reference. To assess variability in reference inventories, it is important to
sample several reference sites.
The output or export of sediment from the watershed of interest is important in closing the
sediment budget. The output of sediment is typically measured as sediment loads or sediment
volumes. Suspended-sediment collection and load computation are standard practices for
agencies such as the USGS, and methods to collect and compute sediment can be downloaded at
their websites. Computation of sediment loads has less error than the measurement of inputs and
storage and, therefore, is an important aspect of the sediment budget. Volumetric measurements
are often based on bathymetric surveys of impoundments (ponds and lakes).
The final step in computing a sediment budget is the summation of all measurements on inputs
and storage terms. All sediment inputs and storage terms should balance to the output, but this
seldom happens. Causes for the discrepancy include measurement error, the natural variability
of channel and upland erosion and deposition, and the number of measurements. It is
recommended that in the final calculation of the sediment budget, error estimates are presented
for the input and storage measurements. This way, the sediment budget can be represented as a
83
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
range. Often the range of values are shown as confidence intervals (i.e. 10th and 90th
percentiles).
Sediment fingerprinting is a technique that apportions the sources of fine-grained sediment in a
watershed using tracers or fingerprints. The sediment fingerprinting procedure establishes a
minimal set of physical and/or chemical properties (tracers) based on samples collected in upland
or channel locations identified as potential sources of sediment. Due to different geologic and
anthropogenic histories, the chemical and physical properties of sediment in a watershed may
vary and often represent a unique signature (or fingerprint) for each source within the watershed.
Fluvial sediment samples (the target sediment) also are collected and exhibit a composite of the
source properties that can be apportioned through various statistical techniques.
There are a number of steps involved in the sediment fingerprinting approach including source
sampling, lab analysis, statistical operations, and results. Potential sediment sources in the
watershed are identified using the same procedure that was used in the sediment budget analysis.
Target sediment can include suspended sediment, bed sediment, floodplain sediment, and
reservoir or lake sediment, and each has its own sampling procedure.
The choice of tracers is broad and can include elemental analysis, stable isotopes, magnetic
properties, color, mineralogy, and radionuclides. Collection of appropriate mass is a
consideration in tracer type analysis. It is important to contact the laboratory where the analyses
will be performed to ask questions pertaining to sample mass and holding requirements.
Soil samples for source analysis collected from upland areas are taken from the top 1 to 2 cm.
To account for variability in the tracer properties, sediment is collected across transects and
composited into one sample. At a channel reach, three to six eroding banks are sampled, each
spaced a minimum of 10 m apart. Streambanks are sampled from the bottom to the top of the
bank face and composited into one sample.
All source and target samples are wet-sieved though a 63-micron polyester sieve. The slurry is
dried at 65°C, and the dried material is collected for laboratory analysis. Fluvial sediment,
because it has been eroded and transported through the watershed, often has a smaller grain size
diameter and greater surface area than the source samples. Grain size differences between the
target and source samples can cause differences in tracer activity. Grain size or surface area
analyses are performed on each sample.
Several statistical steps are used to determine which tracers are most significant in defining
sediment sources. This includes: 1) testing for outliers, 2) testing and correcting tracers for
differences in grain size and organic content between the source and target samples, 3) testing for
the conservativeness of the tracer by confirming that target samples are bracketed by the source
samples where target samples that are not bracketed are not used, 4) determining the optimum
number of tracers that discriminate among the sources using stepwise discriminant function
analysis, 5) identifying source percentages using a mixing model, 6) performing error analysis
using a Monte Carlo and other designs, and 7) analyzing results and applying sediment
weighting factors.
84
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
Combining the sediment budget and sediment fingerprinting results can provide resource
managers with information on where to target measures to reduce erosion, sediment delivery,
and the net transport of sediment (Gellis and Walling, 2011). Areas where sediment is stored can
also be determined by combining the sediment fingerprinting and sediment budget approaches.
Box 11 provides an example from Linganore Creek, MD, where the two approaches were
combined.
Box 11. Combining Sediment Budget and Sediment Fingerprinting Results
Combining sediment budget and sediment fingerprinting results can provide resource
managers with information on where to target measures to reduce erosion, sediment delivery,
and the net transport of sediment (Gellis and Walling, 2011). Areas where sediment is stored
can also be determined by combining the sediment fingerprinting and sediment budget
approaches.
Here the sediment budget results from Linganore Creek are combined with the sediment
fingerprinting results (Gellis et al., 2015). Sediment budget results provided estimates of
gross erosion from upland and erosion from streambanks (Table A). Sediment budget results
show the delivered percentages from each source (Table A). The sediment delivery ratio is
computed as the delivered sediment divided by the gross input of sediment multiplied by 100
(Table A). Subtracting the delivered sediment from the gross erosion from each source type
computes the mass of sediment in storage (Table A).
Table A. Sediment budget and sediment fingerprinting results from Linganore Creek,
MD (Gellis et al., 2015). Why are some cells blank? Should you indicate?
Linganore
Creek Gross erosion
Gross
erosion
Erosion
Watershed
Agriculture
Forest
Streambanks
Export of suspended
sediment
out of the watershed, Mg
5450
Sediment
budget results, Mg
54800
2030
6440
Sediment
fingerprinting source
apportioning results, %
45
3
52
Delivered
sediment, Mg
2453
164
2834
Sediment in storage, Mg
52348
1867
3606
Sediment Delivery Ratio
(SDR). %
4
8
44
85
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
Box 11. Continued.
According to the sediment budget for Linganore Creek, the total gross input of sediment was
63,270 Mg (Table A). Storage elements measured in the sediment budget (Table B) showed a
total storage of 16,240 Mg. The inputs minus the storage measurements yield an output of
sediment of 47,030 Mg. The suspended sediment export at the watershed outlet was 5,450 Mg
(Table A), indicating that the sediment budget did not account for 41,600 Mg. The errors in the
sediment budget could be due to an overestimation of erosion using the Cesium-137 method and/
or not adequately measuring storage areas in the watershed (Gellis et al., 2015).
Table B. Summary of storage measurements from the sediment budget conducted for
Linganore Creek, MD (Gellis et al., 2015)
Storage (Mg)
Ponds
9320
Floodplain
5180
Streambanks
1040
Channel bed
700
The final diagram of the sediment budget and sediment fingerprinting results is shown in Figure.
A.
Agriculture
Forest
Gross Erosion
Streambank inputs
6440 MG
Gross er°s*°n 2030 MG
54,800 .
W_
Export from watershed
5450 MG
Streambank storage 1040 MG
Channel bed storage 700 MG
Floodplain storage 5180 MG
Pond storage 9320 MG
Unaccounted storage
41,500 MG
Figure A. Sediment budget and sediment fingerprinting results for Linganore Creek, MD
shown as a flow diagram.
86
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
Box 11. Continued.
Combining the sediment budget and sediment fingerprinting approaches has implications for
land management agencies interested in reducing sediment. The sediment fingerprinting results
(total load-weighted percentage) indicate that the two main sources of fine-grained sediment
delivered out of the watershed were streambanks (52%) and agriculture (45%). Because
streambanks have a higher sediment delivery ratio than agriculture (44% compared to 4%),
management actions to reduce sediment may be more effective in reducing the net export of fine-
grained sediment if directed toward stabilizing streambank erosion. In addition, streambank
sediment is directly delivered to a stream channel and can have an immediate negative effect on
aquatic habitat. The sediment budget was able to identify and target areas of high bank erosion
(Fig. B).
Ponds and floodplains are important sites of sediment storage. There were numerous ponds in
the Linganore Creek watershed (n = 195) constructed on farms and urban areas that drain 16% of
the Linganore Creek drainage area. The ponds were estimated to store 9,320Mg, which is 15%
of the total eroded sediment. The estimated amount of sediment deposited on floodplains was
5,180 Mg, or 8% of the total eroded sedim ent.
TOP TEXT IN GRAPHIC IS CUT OFF; CHECK FINAL VERSION ("Sampling Site Numbers")
Lmfinori
Ct—k
watershed
$tudysr«i
a*M irnSAD Wiyj*
I t 2 ItlMlUB
I 1 r-1 h r
0 1 2 3 WMULS
bamptoftg ««numMrs
ynt
EXPLANATION
• oltlmnhii Stemn order
BCMlmhn par 1st
and deobfar ^
E15 -cownl _ M
)Leistf»in{
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
Box 11. Continued.
The final sediment budget product including sediment fingerprinting may be a report with
diagrams. Diagrams of sediment budgets are usually most useful for managers to help
understand the magnitude and inputs of sediment inventories and transport. A useful diagram
is to display the inputs and storage of sediment along a longitudinal continuum with the width
of the arrows representative of the amount of loading associated with each source and sink.
88
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
References
Allmendinger, N.E., Pizzuto, J.E., Moglen, G.E., and Lewicki, M. (2007). A sediment budget for an
urbanizing watershed, 1951-1996, Montgomery County, Maryland, USA. Journal of the American
Water Resources Association 43(6), 1483-1498.
American Society for Testing and Materials (ASTM) (2002). D3977-97, Standard Test Methods for
Determining Sediment Concentration Samples. Retrieved from http://www.astm.org/cgi-
bin/SoftCart.exe/DATABASE.CART/REDLINE PAGES/D3977.htm?E+mvstore.
American Society for Testing and Materials (ASTM) (2013). D7348-13, Standard Test Methods for Loss
on Ignition (LOI) of Solid Combustion Residues, ASTM International, West Conshohocken, PA.
American Society for Testing and Materials (ASTM) (2007). D422 - 63, Standard Test Method for
Particle-Size Analysis of Soils. ASTM International.
Amos, K.A., et al. (2009). The application of caesium-137 measurements to investigate floodplain
deposition in a large semi-arid catchment in Queensland, Australia: low-fallout environment.
Earth Surace. Procesess and Landforms 34, 515-529.
Anderson, C.W., and Rounds, S.A. (2010). Use of continuous monitors and auto-samplers to predict
unmeasured water quality constituents in tributaries of the Tualatin River, Oregon: U.S.
Geological Survey Scientific Investigations Report. 76p.
Artiola, J.F., Pepper, I.L., and Brusseau, M. (Eds.). (2004). Environmental Monitoring and
Characterization. Burlington, MA: Elsevier Academic Press, 404 p.
Bartley, R., Keen, R.J., Hawdon, A.A., Hairsine, P.B., Disher, M.G. and Kinsey-Henderson, A.E. (2008).
Bank erosion and channel width change in a tropical catchment. Earth Surface Processes and
Landforms 33( 14), 2174-2200.
Beach, T. (1994). The fate of eroded soil: sediment sinks and sediment budgets of agrarian landscapes in
southern Minnesota, 1851-1988. Annals of the Association of American Geographers 84, 5-28.
Belyaev, V.R., Wallbrink, P.J., Golosov, V.N., Murray, A.S., Sidorchuk, A.Y. (2005). A comparison of
methods for evaluating soil redistribution in the severely eroded Stavropol region, southern
European Russia. Geomorphology, v.65, p. 173-193.
Berglund B. E. (Ed.) (1986) Handbook ofHolocene Palaeoecology andPalaeohydrology. John Wiley &
Sons, Ltd., London, 869 p.
Bettis, E.A., III (1992). Soil morphologic properties and weathering zone characteristics as age indicators
in Holocene alluvium in the Upper Midwest. Iowa Quaternary Studies Group Contribution 30,
119-144.
Bevington, P.R., and Robinson, D.K. (2003). Data reduction and error analysis for the physical sciences:
New York, McGraw-Hill. 320 p.
Birkeland, P.W. (1984). Soils and geomorphology New York, Oxford University Press. 372 p.
Brakebill J.W., Ator, S.W., and Schwarz, G.E. (2010). Sources of Suspended-Sediment in the Chesapeake
Bay Watershed: A Regional Application of the SPARROW Model. Journal of the American
Water Resources Association 46, 757-776.
Bridge, J.S. (2003). Rivers and Floodplains, Forms, Processes, and Sedimentary Record. Oxford, UK,
Blackwell, 404-416.
89
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
Brierley, G.J. and Fryirs, K.A. (2005). Geomorphology and River Management: Applications of the River
Styles Framework. Blackwell Publications, Oxford, UK. 398 p.
Brizga, S.O, and Finlayson, B.L. (1994). Interactions between upland catchment and lowland rivers: An
applied Australian case study. Geomorphology, v. 9, p. 189-201.
Brunton, D.A., and Bryan, R.B. (2000). Rill network development and sediment budgets. Earth Surface
Processes andLandforms, v, 25, p. 783-800.
Burt, R. (editor/compiler). (2009). Soil survey field and laboratory methods manual. Lincoln, NE, US
Department of Agriculture, Natural Resources Conservation Service National Soil Survey Center,
Soil Survey Investigations Report 1.0(51), 435.
Bureau of Reclamation. (2006). Erosion and sedimentation manual. U.S. Department of Interior, Bureau
of Reclamation Technical Service Center, Denver, CO.
Carter, M.W. and A.A. Moghissi. (Eds.) (1977). Three Decades of Nuclear Testing. — Health Physics.
33, 55-71.
Chow, Ven Te (1964). Handbook of Applied Hydrology: A compendium of water-resources technology.
New York, N.Y., McGraw-Hill Company.
Clune, J.W., Gellis, A.C., and McKee, L.G. (2010). Agricultural soil erosion rates forthe Linganore
Creek watershed in the Piedmont Physiographic Province of the Chesapeake Bay watershed.
Proceedings, Second Joint Federal Interagency Conference, Las Vegas, Nevada, June 27-July 1,
2010, CD-ROM ISBN 978-0-0779007-3-2, 7A3, 8 p.
Cohen, A.S. (2003). Paleolimnology the history and evolution of lake systems. New York: Oxford
University Press. 500 p.
Colby, B.R. (1963). Fluvial sediments—a summary of source, transport, deposition, and measurement of
sediment discharge. U.S. Geological Survey Bulletin 1181-A.
Collins, A.L., and Walling, D.E. (2002). Selecting fingerprint properties for discriminating potential
suspended sediment sources in river basins. Journal of Hydrology, 261, 218-244.
Collins, A.L. and Walling, D.E. (2007a). Sources of fine sediment recovered from the channel bed of
lowland groundwater-fed catchments in the UK. Geomorphology 88, 120-138.
Collins, A.L. and Walling, D.E. (2007b). The storage and provenance of fine sediment on the channel bed
of two contrasting lowland permeable catchments, UK. River Research and Applications 23, 429-
450.
Collins, A. L., Walling, D.E., and Leeks, G. J. L. (1997). Sediment sources in the Upper Severn
catchment: a fingerprinting approach. Hydrology and Earth System Sciences, 7(3), 509-521.
Collins, A.L., Walling, D.E., and Leeks, G.J.L. (1998). Use of composite fingerprints to determine the
provenance of the contemporary suspended sediment load transported by rivers: Earth Surface
Processes and Landforms, 23, (1), 31-52.
Collins, A. L., Walling, D.E., Webb, L., and King, P. (2010). Apportioning catchment scale sediment
sources using a modified composite fingerprinting technique incorporating property weightings
and prior information. Geoderma, 155, 249-261.
Collins, B.D., Brown, K.M., and Frairley, H.C. (2008). Evaluation of terrestrial LIDAR for monitoring
geomorphic change at archaeological sites in Grand Canyon National Park, Arizona: U.S.
Geological Survey Open-File Report 2008-1384.
Costa, J. E. (1975). Effects of agriculture on erosion and sedimentation in the Piedmont Province,
Maryland. Geological Society of America Bulletin 86, 1281-1286.
90
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
Curtis, J.A., Flint, L.E., and Hupp, C.R. (2013). Estimating floodplain sedimentation in the Laguna de
Santa Rosa, Sonoma County, CA. Wetlands 33(1), 29-45.
Danielson, Tyler. (2013). Utilizing a High Resolution Digital Elevation Model (DEM) to Apply Stream
Power Index (SPI) to the Gilmore Creek Watershed in Winona County, Minnesota. Papers in
Resource Analysis, 75(11).
Darby, S.E., Rinaldi, M., and Dapporto, S. (2007). Coupled simulations of fluvial erosion and mass
wasting for cohesive river banks. Journal of Geophysical Research-Earth Surface, 772(F3), 1-15.
de Vente, J., Poesen, J., Arabkhedri, M. and Verstraeten, G. (2007). The sediment delivery problem
revisited. Progress in Physical Geography 31, 155-178. doi: 10.1177/0309133307076485.
Devereux, O.H., Prestegaard, K.L., Needelman, B.A., and Gellis, A.C. (2010). Suspended-sediment
sources in an urban watershed, Northeast Branch Anacostia River, Maryland. Hydrological
Processes, v. 24(11), p. 1391-1403.
Dietrich, W. E., and Dunne, T. (1978). Sediment budget for a small catchment in mountainous terrain.
Zeitschuift fur Geomorphologic supplement, 29, 191-206.
Dietrich, W. E., Dunne, T., Humphrey, N.F. and Reid, L.M. (1982). Construction of sediment budgets for
drainage basins in Sediment Budgets and Routing in Forested Drainage Basins, F.J. Swanson,
R.J. Janda, T. Dunne, and D.N. Swanston (eds.), U.S.D.A. Forest Service General Technical
Report PNW-141, Pacific Northwest Forest and Range Experiment Station, Portland, Oregon, 5-
23.
Dietterick, B. C., White, R., & Hilburn, R. (2012). Comparing LiDAR-Generated to Ground- Surveyed
Channel Cross-Sectional Profiles in a Forested Mountain Stream. General Technical Reports of
the Pacific Southwest Research Station-PSW-GTR-238, 639-648.
Donovan, M., Miller, A., Baker, M., and Gellis, A. (2015) Sediment contributions from floodplains and
legacy sediments to Piedmont streams of Baltimore County, Maryland. Geomorphology, 235, 88-
105.
Dunne, T. (1979). Sediment yield and land use in tropical catchments. J. Hydrol. 42, 281-300.
Edwards, T.K., and Glysson, G.D. (1999). Field methods for measurement of fluvial sediment: U.S.
Geological Survey Techniques of Water-Resources Investigations of the U.S. Geological Survey,
Book 3, Chapter C2, 97, https ://pubs .uses .gov/twri/twri 3 -c2/pdf/TWRI 3 -C2 .ixlf.
Erskine, W.D., Mahmoudzadeh, A., and Myers, C. (2002). Land use effects on sediment yields and soil
loss rates in small basins of Triassic sandstone near Sydney, NSW, Australia. Catena, v. 49, p.
271-287.
Evans, M.G. and Warburton, J. (2005). Sediment budget for an eroding peat moorland catchment in
Northern England Earth Surface Processes and Landforms 30, 557-577.
Evrard, O., Navratil, O., Ayrault, S., Ahmadi, M., Nemery, J., Legout, C,. Lefevre, I., Poirel A, Bonte, P,.
and Esteves, M. (2011). Combining suspended sediment monitoring and fingerprinting to
determine the spatial origin of fine sediment in a mountainous river catchment. Earth Surface
Processes and Landforms v. 36, p. 1072-1089.
Faulkner, D. and Mclntyre, S. (1996). Persisting sediment yields and sediment delivery changes. Water
Resources Bulletin 32, 817 - 829.
Faulkner, D.J. (1998). Spatially variable historical alluviation and channel incision in west-central
Wisconsin. Annals of the Association of American Geographers 88(4), 666-685.
91
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
Faulkner, T.J. (1988). The Shipway Limestone of Gower: sedimentation on a storm-dominated early
Carboniferous ramp. Geology Journal, 23, 85-100.
Faux, R. N., Buffington, J. M., Whitley, M. G., Lanigan, S. H., and Roper, B. B. (2009). Use of Airborne
Near-Infrared LiDAR for Determining Channel Cross-Section Characteristics and Monitoring
Aquatic Habitat in Pacific Northwest Rivers: A Preliminary Analysis. PNAMP Special
Publication: Remote Sensing Applications for Aquatic Resource Monitoring, 43-60.
Fenwick, I. (1985). Paleosols: Problems of recognition and interpretation. In Boardman, J. (Ed.), Soils
and Quaternary landscape evolution. Chichester, Wiley and Sons, 3-21.
Fitzpatrick, F. A. (1998). Geomorphic and Hydrologic Responses to Vegetation, Climate, and Base Level
Changes, North Fish Creek, Wisconsin. Unpublished doctoral dissertation, Department of
Geography, University of Wisconsin-Madison.
Fitzpatrick, F.A. (2005). Trends in streamflow, sedimentation, and sediment chemistry for the Wolf River,
Menominee Indian Reservation, Wisconsin, 1850-1999. US Geological Survey Scientific
Investigations Report 2005-5030, 47 p.
Fitzpatrick, F.A.. (2014). A Geologic Approach to Field Methods in Fluvial Geomorphology, Chapter 5.2,
in Thornbush, M.J., Allen, C.D., and Fitzpatrick, F.A. (Eds.), Geomorphological Fieldwork,
Volume 18, Developments in Earth Surface Processes. Amsterdam, Netherlands, Elsevier, 201-
230.
Fitzpatrick, F.A., 2016, Watershed Geomorphological Characteristics: Chapter 44, in Singh, V., (Ed.),
Applied Handbook of Hydrology, 44.1 - 44.12.
Fitzpatrick, F.A., and Knox, J.C. (2000). Spatial and temporal sensitivity of hydrogeomorphic response and
recover to deforestation, agriculture, and floods. Physical Geography 21(2), 89-108.
Fitzpatrick, F.A., Knox, J.C., and Whitman, H.E. (1999). Effects of historical land-cover changes on
flooding and sedimentation, North Fish Creek, Wisconsin. US Geological Survey Water-Resources
Investigations Report 99-4083, 12.
Fitzpatrick, F.A., Knox, J.C., and Schubauer-Berigan, J.P. (2007). Sedimentation history of Halfway
Creek Marsh, Upper Mississippi Wildlife and Fish Refuge, Wisconsin, 1846-2006: U.S.
Geological Survey Scientific Investigations Report 2007-5209, p. 42.
Fitzpatrick, F.A., Knox, J.C., and Schubauer-Berigan, J.P. (2009). Channel, floodplain, and wetland
responses to floods and overbank sedimentation, 1846-2006, Halfway Creek Marsh, Upper
Mississippi Valley, Wisconsin, in James, L.A., Rathburn, S.L., and Whittecar, G.R., (Eds.),
Management and Restoration of Fluvial Systems with Broad Historical Changes and Human
Impacts. Geological Society of America Special Paper 451, 23-42.
Fitzpatrick, F.A., Peppier, M.C., Schwar, H.E., Hoopes, J.A. and Diebel, M.W. (2005). Monitoring channel
morphology and bluff erosion at two installations of flow-deflecting vanes, North Fish Creek,
Wisconsin, 2000-03. U.S. Geological Survey Scientific Investigations Report 2004-5272, 34.
Fitzpatrick, F.A., Waite, I.R., D'Arconte, P.J., Meador, M.R., Maupin, M.A., and Gurtz, M.E. (1998).
Revised methods for characterization stream habitat in the National Water-Quality Assessment
Program. US Geological Survey Water-Resources Investigations Report 98-4052, 68.
Fitzpatrick. F.A., Peppier, M.C., Saad. D.A., Pratt. D M., and Lenz, B.N. (2015). Geomorphic. flood, and
groundwater-flo\\ characteristics of Bayfield Peninsula streams, Wisconsin, and implications for
brook-trout habitat: US Geological Survey Scientific Investigations Report 2014-5007, 80 p..
http://dx.doi.org/! 0.3133/sir20145007.
92
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
Foster, I.D.L., Dearing, J.A., and Grew, R. (1988). Lake Catchments: an evaluation of their contribution
to studies of sediment yield and delivery process, in Bordas, M.P., and Walling, D.E., eds.,
Sediment Budgets, IAHS Publication No. 174\ International Association of Hydrological
Sciences, Wallingford, UK., p. 413-424.
Foster I., Lees J., Owens P., and Walling, D. (1998). Mineral magnetic characterization of sediment
sources from an analysis of lake and floodplain sediments in the catchments of the Old Mill
reservoir and Slapton Ley, South Devon, UK. Earth Surface Processes andLandforms, 23, 685-
703.
Fox, G.A. et al. (2007). Measuring streambank erosion due to ground water seepage: correlation to bank
pore water pressure, precipitation and stream stage. Earth Surface Processes and Landforms, 32,
1558-1573.
Fox, J.F., and Papanicolaou, A.N. (2008). Application of the spatial distribution of nitrogen stable
isotopes for sediment tracing at the watershed scale. Journal of Hydrology v. 358, p. 46-55.
Fryirs, K.A. and Brierley, G. J, 2013). Geomorphic Analysis of River Systems: An Approach to Reading
the Landscape. Blackwell Publishing, Oxford, UK. 345 p.
Garcin, M., Poisson, B., and Pouget, R. (2005). High rates of geomorphological processes in a tropical
area: the Ramparts River case study (Reunion Island, Indian Ocean). Geomorphology 67(3-4),
335-350.
Gaugush, R. F. (2004). A suspended sediment budget for Pool 13 and La Grange Pool of the Upper
Mississippi River System. Final report to U.S. Army Corps of Engineers. U.S. Geological Survey,
Upper Midwest Environmental Sciences Center, La Crosse, Wisconsin. September 2004. LTRMP
2004-T004. 13 pp. + Appendixes A-D.
Gellis, A.C., and Walling, D.E. (2011). Sediment-source fingerprinting (tracing) and sediment budgets as
tools in targeting river and watershed restoration programs: in Simon, A., Bennett, S., Castro,
J.M., (Eds.), Stream Restoration in Dynamic Fluvial Systems: Scientific Approaches, Analyses,
and Tools, American Geophysical Union Monograph Series 194, 263-291.
Gellis, A.C., Webb, R.M.T., Wolfe, W.J., and Mclntyre, S.C.I. (1999). Effects of land use on upland
erosion, sediment transport, and reservoir sedimentation, Lago Loiza Basin, Puerto Rico. U.S.
Geological Survey Water-Resources Investigations Report 99- 4010, 1-60.
Gellis, A. C., Webb, R.M.T., Wolfe, W.J., and Mclntyre, S.C.I. (2006). Changes in land use and reservoir
sedimentation: a case study in the Lago Loiza basin, Puerto Rico. Physical Geography, 27, 39-69.
Gellis, A.C., Hupp, C.R., Pavich, M.J., Landwehr, J.M., Banks, W.S.L., Hubbard, B.E., Langland, M.J.,
Ritchie, J.C., and Reuter, J.M. (2009). Sources, transport, and storage of sediment at selected sites
in the Chesapeake Bay Watershed: U.S. Geological Survey, Scientific Investigations Report,
2008-5186, 95 p.
Gellis, A.C., Pavich, M.J., Ellwein, A.L., Aby, S., Clark, I., Wieczorek, M.E., and Viger, R. (2012).
Erosion, Storage, and Transport of Sediment in Two Subbasins of the Rio Puerco, New Mexico:
Geological Society of America Bulletin, 124(5/6), 817-841.
Gellis, A.C., Noe, G.B., Clune, J.W., Myers, M.K., Hupp, C.R., Schenk, E.R., and Schwarz, G.E. (2015).
Sources of fine-grained sediment in the Linganore Creek watershed, Frederick and Carroll
Counties, Maryland, 2008-10: U.S. Geological Survey Scientific Investigations Report 2014-
5147, 56. http://dx.doi.org/10.3133/sir20145147.
93
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
Gellis, A.C., Myers, M.K. Noe, G.B., Hupp, C.R. Schenk, E.R., and Myers, L. (In Press). Storms, channel
changes, and a sediment budget for an urban-suburban stream, Difficult Run, Virginia, USA.
Geomorphology, 278, p.128-148.
Gibson, S.A., Little, C.D., Jr. (2006). Implementation of the sediment impact assessment model (SIAM)
in HEC-RAS in: Bernard, J.M. (Ed.), Eighth Federal Interagency Sedimentation Conference.
Federal Interagency Subcommittee on Sedimentation, Reno, Nevada, 65-72.
Gilbert, G.K. (1877). Report on the geology of the Henry Mountains: U.S. Geographical and Geological
Survey of the Rocky Mountain Region, Washington, D.C., 160 p.
Gingele, F.X., and De Deckker, P. (2005). Clay mineral, geochemical and Sr-Nd isotopic fingerprinting
of sediments in the Murray-Darling fluvial system, southeast Australia. Australian Journal of
Earth Sciences 52, 965-974.
Golosov V. and Walling, D. (2014). Using fallout radionuclides to investigate recent overbank
sedimentation rates on river floodplain: an overview in Sediment Dynamics from the Summit to
the Sea. Vol.367, Wallingford, UK: IAHS Press, 228-234.
Gorman-Sanisaca, L., Gellis, A.C., and Lorenz, D.L.., In Preparation, Determining the sources of fine-
grained sediment using the Sediment Source Assessment Tool (Sed_SAT): U.S. Geological Survey
Techniques and Methods.
Gruszowski, K.E., Foster, I.D.I., Lees, J.A., and Charlesworth, S.M. (2003). Sediment sources and
transport pathways in a rural catchment, Herfordshire, UK. Hydrological Processes, v. 17, p.
2665-2681.
Guy, H.P. (1969). Laboratory theory and methods for sediment analysis. Chapter CI, Book 5, Techniques
of Water-Resources Investigations of the United States Geological Survey, 58 p.
Hakanson, L., and Jansson, M. (2002). Principles of Lake Sedimentology. New Jersey: The Blackburn
Press, 315 pp.
Happ, S.C., Rittenhouse, G., and Dobson, G.C. (1940). Some principles of accelerated stream and valley
sedimentation. US Department of Agriculture Technical Bulletin 695.
Harrelson, C. C., C. L. Rawlins, and J. P. Ptoyondy. (1994). Stream channel reference sites: an illustrated
guide to field technique. U.S. Department of Agriculture Forest Service, General Technical
Report RM-245, Fort Collins, Colorado.
Hart, E. A., and Schurger, S. (2005). Sediment Storage and Yield in an Urbanized Karst Watershed.
Geomorphology, v. 70, p. 85-96.
Haynes, C.V. (1968). Geochronology of late-Quaternary alluvium, in Morrison, R.B., and Wright, H.E.,
(Eds.), Means of Correlation of Quaternary Successions. Salt Lake City, UT: University of Utah
Press, 591-631.
Hayter, E., Gustavson, K., Ells, S., Gallani, J., Wolfe, J., Dekker, T., Redder, T. (2014). Technical
guidelines on performing a sediment erosion and deposition assessment (SEDA) at Superfund
sites. U.S. Army Corps of Engineers Engineer Research and Development Center Technical
Report TR-149.
He, Q. and Walling, D.E. (1996). Interpreting particle size effects on the adsorption of 137Cs and
unsupported 210Pb by mineral soils and sediments. Journal of Environmental Radioactivity, 30,
117-137.
Helsel, D.R., and Hirsch, R.M. (1992). Statistical methods in water resources. New York: Elsevier
Science Publishing Co., Inc., 529 p.
94
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
Hooke, J.M. and Yorke, L. (2011). Channel bar dynamics on multi-decadal timescales in an active
meandering river. Earth Surface Processes andLandforms 36, 1910-1928.
Horowitz A.J. (1991). A Primer on sediment trace element chemistry. United States Geological Survey
Open-File Report 91-76, 136 p.
Houben P., Hoffmann, T., Zimmermann, A. and Dikau, R. (2006). Land use and climatic impacts on the
Rhinesystem RheinLUCIFS): Quantifying sediment fluxes and human impact with available
data. Catena, 66, 42-52.
Huggins, D.R., and Reganold, J.P. (2008). No-till, the quiet revolution. Scientific American, 299, 70-77.
Hughes, M.L., McDowell, P.F., and Marcus, W.A. (2006). Accuracy assessment of georectified aerial
photographs: implications for measuring lateral channel movement in a GIS: Geomorphology,
74, 1-16.
Hupp, C. R. (2000). Hydrology, geomorphology, and vegetation of Coastal Plain rivers in the
southeastern United States, Hydrological Processes, 14, 2991 - 3010.
Hupp C. R., Demas, C. R., Kroes, D. E., Day, R. H., and Doyle, T. W. (2008). Recent sedimentation
patterns within the central Atchafalaya Basin, Louisiana. Wetlands 28: 125-140. DOI:
10.1672/06-132.1.
Interstate Technology & Regulatory Council (ITRC). (2012). Incremental Sampling Methodology. ISM-
1. Washington, D.C.: Interstate Technology & Regulatory Council, Incremental Sampling
Methodology Team. \\gs.doi.net\baltimoremd-w\users\baltimore\water\agellis\marvland\EPA
FUNDING\MANUAL\www. itrcweb .org.
Jacobson, R.B. and Coleman, D.J. (1986). Stratigraphy and recent evolution of Maryland Piedmont
floodplains. American Journal of Science, 286, 617-637.
James, L.A. (1989). Sustained storage and transport of hydraulic gold mining sediment in the Bear
River, California. Annals of the Association of American Geographers 79, 570 - 592.
James, A. (2013). Legacy sediment: definitions and process of episodically produced anthropogenic
sediment: Anthropocene 2, 16-26.
Jastram, J.D. (2014). Streamflow, water quality, and aquatic macroinvertebrates of selected streams in
Fairfax County, Virginia, 2007-12. U.S. Geological Survey Scientific Investigations Report
2014-5073, 68 p. http://dx.doi.org/10.3133/sir20145073.
Julian, J. P., and Torres, R. (2006). Hydraulic erosion of cohesive riverbanks. Geomorphology 76, 193-
206.
Kaufman, P.R., Levine, P., Robison, E.G., Seeliger, C., and Peck, D.V. (1999). Quantifying physical
habitat in wadeable streams. U.S. Environmental Protection Agency EPA/620/R-00/003,
Washington, D.C.
Kesel, R. H., Yodis, E. G., and McCrow, D. J. (1992). An approximation of the sediment budget of the
lower Mississippi River prior to\ human modification. Earth Surface Processes and Landforms,
17, 711-722.
Kleiss, B.A. (1993). Cache River, Arkansas: Studying a bottomland hardwood (BLH) wetland
ecosystem. U.S. Army Waterways Experiment Station, Vicksburg, Mississippi, USA. The
Wetlands Research Program, 3,(1).
Kline, M., Alexander, C., Pomeroy, S., Springston, G., Jaquith, S., Cahoon, B., and Becker, L. (2004).
Vermont Stream Geomorphic Assessment Phase 1 Handbook Watershed Assessment Using
Maps, Existing Data, and Windshield Surveys. Vermont Agency of Natural Resources. 82p.
95
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
www.vtwaterquality.org/rivers.htm.
Koch, R. W. and Smillie, G .M. (1986). Bias in hydrologic prediction using log transformed regression
models. American Water Resources Association Water Resources Bulletin, 22(5), 717-723.
Knox, J.C. (1972). Valley alluviation in southwestern Wisconsin. Annals of the Association of American
Geographers 62(3), 401-410.
Knox, J.C. (1977). Human impacts on Wisconsin stream channels. Annals of the Association of
American Geographers 67(3), 323-342.
Knox, J.C. (1985). Responses of floods to Holocene climatic change in the upper Mississippi River
valley. Quaternary Research 23, 287-300.
Knox, J.C. (1987). Historical valley floor sedimentation in the Upper Mississippi River Valley. Annals of
the Association of American Geographers 77(2), 224-244.
Knox, J.C. (2002). Agriculture, erosion, and sediment yields, Ch. 23, in Orme, A.R., (Ed.), The
Physical Geography of North America. Oxford, UK, Oxford University Press, 482-500.
Knox, J.C. (2006). Floodplain sedimentation in the Upper Mississippi Valley: natural versus human
accelerated. Geomorphology 79, 286-310.
Knox, J. C., Bartlein, P. J., and Johnson, W. C. (1974). Environmental assessment of sediment sources
and sedimentation distributions for the Lake LaFarge watershed and impoundment.
Environmental Analysis of the Kickapoo River Impoundment: Institute for Environmental Studies,
University of Wisconsin, Report 28,11 -116.
Koiter, A. J., Owens, P. N., Petticrew, E. L., Lobb, D. A. (2013). The behavioral characteristics of
sediment properties and their implications for sediment fingerprinting as an approach for
identifying implications for sediment fingerprinting as an approach for identifying sediment
sources in river basins sediment sources in river basins. Earth-Science Reviews 125, 24-42.
Koltun, G.F., Eberle, M., Gray, J.R., and Glysson, G.D. (2006). User's manual for the graphical
constituent loading analysis system (GCLAS). U.S. Geological Survey Techniques and Methods,
book 4, chap. CI, 51, http://water.usgs.gov/software/GCLAS/.
Kondolf, G.M., and Matthews, W.V.G. (1991). Unmeasured residuals in sediment budgets: a cautionary
note. Water Resources Research, 27, 2483-2486.
Konert, M. and Vandenberghe, J. (1997). Comparison of laser grain size analysis with pipette and sieve
analysis: a solution for the underestimation of the clay fraction. Sedimentology, 44, 523-535.
Krein, A., Petticrew, E., Udelhoven, T. (2003). The use of fine sediment fractal dimensions and colour to
determine sediment sources in a small watershed. Catena, 53, 165-179.
Laceby, J.P., McMahon, J., Evrard, O., and Olley, J. (2015). A comparison of geological and statistical
approaches to element selection for sediment fingerprinting. Journal of Soils and Sediment, v.
75(10), 2117-2131. DOI 10.1007/sl 1368-015-1111-9.
Lamba, J., Thompson, A., Karthikeyan, K., and Fitzpatrick, F. (2015). Sources of fine sediment stored
in agricultural lowland streams, Midwest, USA. Geomorphology,
10.1016/j.geomorph. 2015.02.001, 44-53.
Lane, E. W. (1947). Report of the subcommittee on sediment terminology. American Geophysical Union
Transactions, 28(6), 936-938.
Larsen, M.C., Liu, Z., and Zou, X. (2012). Effects of earthworms on slopewash. surface runoff, and fine
litter transport on a humid tropical forested hillslope. Luquillo Experimental Forest. Puerto Rico.
chapter G in Murphy, S.F. and Stallard, R.F. (Eds.), Water quality and landscape processes of
96
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
four watersheds in eastern Puerto Rico: U.S. Geological Survey Professional Paper 1789, 179-
198.
Larsen, M. C., Gellis, A. C., Glysson, G. D., Gray, J. R., and Horowitz, A. J. (2010). Fluvial sediment in
the environment: a national problem. Proceedings, Second Joint Federal Interagency Conference,
Las Vegas, NV, June 7 -July 1, 2010, available at
http://acwi.gov/sos/pubs/2ndJFIC/Contents/OS_Larsen_9fisc_sediment_vision_3_4_2010.pdf.
15p.
Lawler, D. M. (1986). River bank erosion and the influence of frost: a statistical examination.
Transactions of the Institute of British Geographers 11, 227-242.
Lee, K. G. (2013). Estimation of reservoir storage capacity using multibeam sonar and terrestrial LiDAR,
Randy Poynter Lake, Rockdale County, Georgia, 2012. US Geological Survey Scientific
Investigations Map 3265.
Leopold, L. B. (1956). Land use and sediment yield. Williams, T.L., Jr. (Ed.) Man's Role in Changing the
Face of the Earth. Chicago, University of Chicago Press, 639-647.
Leopold, L. B., W. W. Emmett, and R. M. Myrick (1966). Channel and hillslope processes in a semiarid
area, New Mexico. U.S. Geological Survey Professional Paper 352-G.
Leopold, L. B. (2006). A view of the river. Cambridge, MA, Harvard University Press.
Libby, W. F., Anderson, E. C., and Arnold, J. R. (1949). Age determination by radiocarbon content:
world-wide assay of natural radiocarbon. Science, 709(2827), 227-228.
Lichter, J. M. and Costello, L. R. (1994). Measuring soil bulk density. Journal of Arboriculture 20(3),
160-164.
Little C.D., Jonas M. (2010). Sediment impact analysis methods (SIAM): overview of model capabilities,
applications, and limitations. Second Joint Federal Interagency Conference, Las Vegas, NV.
Loughran, R.J., Campbell, R.L., Shelly, D.J., and Elliott, G.I. (1992) Developing a sediment budget for a
small drainage basin in Australia. Hydrological Processes, v. 6, p. 145-158.
Lowdermilk, W.C. (1934). Acceleration of erosion above geologic norms. American Geophysical Union,
15th Transactions, Pt. 2, 505-509.
Lundstrom S. C., Mahan, S. A., Paces, J. B., Hudson, M. R., House, P. K., Malmon, D. V., Blair, J. L.,
Howard, K. A. (2008). Late Pleistocene aggradation and degradation of the lower Colorado
River: perspectives from the Cottonwood area and other reconnaissance below Boulder Canyon.
Reheis, M. C., Herschler, R., Miller, D .M. (Eds.) Late Cenozoic Drainage History of the
Southwestern Great Basin and Lower Colorado River Region: Geologic and Biotic Perspectives:
Geological Society of America Special Paper 439, 409-430. doi: 10.1130/2008.2439(19).
Mabit, L., Benmansour, M., Walling, D. E. (2008). Comparative advantages and limitations of fallout
radionuclides (137Cs, 210Pb and 7Be) to assess soil erosion and sedimentation. Journal of
Environmental Radioactivity, 99, 1799-1807.
Macaire J. J., Bellemlih S., Di-Giovanni, C., De Luca, P., Visset, L., Bernard, J. (2002). Sediment yield
and storage variations in the Negron River catchment (south western Parisian Basin, France)
during the Holocene Period. Earth Surface Processes andLandforms 27, 991-1009.
Magilligan, F.J. (1985). Historical floodplain sedimentation in the Galena River Basin, Wisconsin and
Illinois. Annals of the Association of American Geographers 75, 583-594.
97
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
Matisoff, G., Wilson, C. G., and Whiting, P. J. (2005). The 7Be/210Pbxs ratio as an indicator of suspended
sediment age or fraction new sediment in suspension. Earth Surface Processes andLandforms
30, 1191-1201.
Meade, R. H. (1994). Suspended sediments of the modern Amazon and Orinoco rivers. Quaternary
International 21, 29-39.
Meade, R. H., Weibezahn, F. H., Lewis Jr., W. M., and Perez Hernandez, D. (1990). Suspended-sediment
budget for the Orinoco River. F. H. Weibezahn, H. Alvarez and W. M. Lewis Jr. (Eds.). The
Orinoco River as an Ecosystem. Impresos Rubel, Caracas, 55-79.
Meals, D. W., Dressing, S. A., Davenport, T. E. (2010). Lag time in water quality response to best
management practices: a review. Journal of Environmental Quality 39, 85-96.
Milfred, C. J., Olson, G. W., Hole, F. D. (1967). Soil resources and forest ecology of Menominee County,
Wisconsin. University of Wisconsin, Geological and Natural History Survey, Soil Survey
Division, Bulletin 85, Soil Series No. 60: Madison, WI, 203 p. and 3 plates.
Miller, J. R., Orbock Miller, S. M. (2007). Contaminated rivers: a geomorphological-geochemical
approach to site assessment and remediation. Springer International Publishing: Dordrecht, The
Netherlands 418 p.
Miller, J. R., Mackin, G., and Orbock Miller, S. M. (2015). Application of Geochemical Tracers to
Fluvial Sediment. Springer International Publishing: Dordrecht, The Netherlands, 142 p.
Moody, J. A., Meade, R. H. (2013). Powder River: data for cross-channel profiles at 22 sites in
southeastern Montana from 1975 through 2013. US Geological Survey Data Set.
doi: 10.5066/F70Z719C.
Moody, J. A., Pizzuto, J. E., and Meade, R. H. (1999). Ontogeny of a floodplain. Geological Society of
America Bulletin 777(2), 291-303.
Montgomery, D. R. (2007). Dirt: the erosion of civilizations. Oakland, CA, University of California Press,
295 p.
Motha, J. A., Wallbrink, P. J., Hairsine, P. B., and Grayson, R. B. (2003). Determining the sources of
suspended sediment in a forested catchment in southern Australia: Water Resources Research,
39(3), 1056. DOI: 10.1029/2001 WR00794.
Mukundan, R., Walling, D. E., Gellis, A. C., Slattcry, M. C., and Radcliffc, D. E. (2012). Sediment
source fingerprinting: transforming from a research tool to a management tool. Journal of
American Water Resource Association 48, 1241-1257.
Munsell Color. (1975). Munsell soil color charts. Baltimore, MD, Munsell Color Division of Kollmorgen
Corp., variable pagination.
Nagel, David E.; Buffington, John M.; Parkes, Sharon L.; Wenger, Seth; Goode, Jaime R. (2014). A
landscape scale valley confinement algorithm: Delineating unconfined valley bottoms for
geomorphic, aquatic, and riparian applications. General Technical Report RMRSGTR-321. Fort
Collins, CO: U.S. Department of Agriculture, Forest Service, Rocky Mountain Research Station.
42 p.
Naglc, G.N., Fahey, T.J., Ritchie. J.C., and Woodbury. P.B. (2007). Variations in sediment sources and
yields in the Finger Lakes and CatskilIs regions of New York. Hydrological Processes 21(6), p.
828-838.
Natural Resource Conservation Service (NRCS) (2007). Gullies and their control: USDA-NRCS
Technical Supplement 14P, Part 654 National Engineering Handbook, 20 p.
98
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
Natural Resource Conservation Service (NRCS) (2002). Estimating soil loss from gully erosion. Field
Office Technical Guide Erosion Prediction, Section I-D-3.
Nelson, J. (2010). Digital terrain analysis with LiDAR for clean water implementation - Workshop April
7, 2010. Minnesota Department of Agriculture. Acquired during MN Dept. of Ag. Workshop in
Winona, MN April 7, 2010. A similar copy can be acquired from
http://www.mda.state.mn.us/protecting/cleanwater/pilotprojects/~/media/Files/protecting/waterpr
otection/lidarworkshopshow.ashx.
Nimz, G.J. (1998). Lithogenic and cosmogenic tracers in catchment hydrology, in Kendall, C., and
McDonnell, J.J., eds., Isotope Tracers in Catchment Hydrology: Elsevier, Amsterdam, The
Netherlands, p. 247-290.
O'Connor, J. E., Mangano, J. F., Anderson, S. W., Wallick, J. R., Jones, K. L., Keith, M. K. (2014).
Geologic and physiographic controls on bed-material yield, transport, and channel morphology
for alluvial and bedrock rivers, western Oregon. Geological Society of America Bulletin.
doi: 10.1130/B30831.1.
Oguchi, T. (1997). Late Quaternary sediment budget in alluvial-fan—source-basin systems in Japan.
Journal of Quaternary Science 72, 381-390.
Olsson, I. U. (1986). Radiometric dating. Berglund, B. E. (Ed.) Handbook of Holocene Palaeoecology
andPalaeohydrology. Chichester, U.K., John Wiley and Sons, 273-312.
Ostrom, M. E. (1971). Preliminary report on results of physical and chemical tests of Wisconsin silica
sandstones. Wisconsin Geological and Natural History Survey Information Circular 18, 61p.
Owens, P. N. and Walling, D. E., (2002). Changes in sediment sources and floodplain deposition rates
in the catchment of the River Tweed, Scotland, over the last 100 years: The Impact of Climate
and d-use Change. Earth Surface Processes andLandform 27(4), 302-423.
Owens, P. N., Walling, D. E. and Leeks, G. J. L. (1999). Use of floodplain sediment cores to investigate
recent historical changes in overbank sedimentation rates and sediment sources in the
catchment of the River Ouse, Yorkshire, UK. Catena 36, 21-47.
Page, M.J., Trustrum, N.A., and Dymond, J.R. (1994). Sediment budget to assess the geomorphic
effect of a cyclonic storm, New Zealand. Geomorphology, v. 9, no.3, p. 169-188.
Pedocchi, F., and Garcia, M. H. (2006). Evaluation of the LISST-ST instrument for suspended particle
size distribution and settling velocity measurements, Continental Shelf Research 2(5(8), 943-
958. doi: 10.1016/j.csr.2006.03.006.
Peppier, Marie C.; Fitzpatrick, Faith A. (2005). Methods for monitoring the effects of grazing
management on bank erosion and channel morphology, Fever River, Pioneer Farm, Wisconsin,
2004. Fact Sheet 2005-3134, 4 p.: ill.; maps.
Phillips, J.D. (1991). Fluvial sediment budgets in the North Carolina piedmont. Geomorphology 4, 231
-241.
Phillips, J. M., Russell, M. A., and Walling, D. E. (2000). Time-integrated sampling of fluvial
suspended sediment: a simple methodology for small catchments. Hydrological Processes,
74(14), 2589-2602.
Phippen, S.J., and Wohl, E. (2003). An assessment of land use and other factors affecting sediment
loads in the Rio Puerco watershed, Rio Puerco. Geomorphology, v. 52, p. 269-287.
Pitlick, J. and Wilcock, P. R. (2001). Flow, sediment transport, and aquatic habitat in large rivers.
Geomorphic Processes and Riverine Habitat. Dorava, J., Fitzpatrick, F., Montgomery, D., and
Palcsak, B. (Eds.) Washington, D.C., American Geophysical Union, 185-198.
99
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
Pittam, N., Foster, I., Mighall, T. (2009). An integrated lake-catchment approach for determining
sediment source changes at Aqualate Mere, Central England. Journal ofPaleolimnol 42, 215—
232.
Poesen, J., Nachtergaele, J., Verstraeten, G., and Valentin,C. (2003). Gully erosion and environmental
change: importance and research needs. Catena, 50(2-4), 91-133.
Pollard, D., and DeConto, R.M. (2003). Antarctic ice and sediment flux in the Oligocene simulated by a
climate-ice sheet-sediment model. Palaeogeography, Palaeoclimatology, Palaeoecology, v.
198,p. 53-67.
Polyakov, V.O., Nearing, M.A., Shipitalo, M.J. (2004). Tracking sediment redistribution in a small
watershed: implications for agro-landscape evolution. Earth Surface Processes and Landforms,
v. 29, p. 1275-1291.
Porterfield, G. (1977). Computation of fluvial-sediment discharge. U.S. Geological Survey Techniques
of Water-Resources Investigations, book 3, chap. C3, 66 p.
Page, M.J., Trustrum, N.A., and Dymond, J.R. (1994). Sediment budget to assess the geomorphic
effect of a cyclonic storm, New Zealand. Geomorphology, v. 9, no.3, p. 169-188.
Pulley, S., Foster, I., and Antunes, P. (2015). The application of sediment fingerprinting to floodplain
and lake sediment cores: assumptions and uncertainties evaluated through case studies in the
Nene Basin, UK : Journal of Soils and Sediments, 15, 2132-2154.
Ramos-Scharron, C. E., MacDonald, L. H. (2005). Measurement and prediction of sediment production
from unpaved roads, St. John, USVI. Earth Surface Processes and Landforms, 30, 1283-1304.
Rasmussen, P. P., Gray, J. R., Glysson, G. D., and Ziegler, A. C. (2009). Guidelines and procedures for
computing time-series suspended-sediment concentrations and loads from in-stream turbidity-
sensor and streamflow data. U.S. Geological Survey Techniques and Methods, book 3, chap.
C4, 52 p.
Reid, L. M., Dunne, T. (1984). Sediment production from forest road surfaces. Water Resources
Research 70(11), 1753-1761.
Reid, L. M., and Dunne, T. (1996). Rapid evaluation of sediment budgets. Reiskirchen, Germany,
Catena Verlag GMBH, 164 p.
Renwick, W.H., Smith, S.V., Bartley, J.D., and Buddemeier, R.W. (2005). The role of impoundments in
the sediment budget of the conterminous United States. Geomorphology, 71, (1-2), 99-111.
Retallick, G. (1985). Laboratory exercises in paleopedology. Course Handout, University of Oregon,
Eugene, 74 p.
Revesz, Kinga, Qi, Haiping, and Coplen, T. B. 2012). Determination of the S 15N and S 13C of total
nitrogen and carbon in solids; RSIL lab code 1832, chap. 5 o/Stable isotope-ratio methods, sec.
C. Revesz, Kinga, and Coplen, T. B. (Eds) Methods of the Reston Stable Isotope Laboratory
(slightly revised from version 1.1 released in 2007): U.S. Geological Survey Techniques and
Methods, book 10,21 p., available only at http://pubs.usgs.gov/tm/2006/tmlQc5/.
Ritchie, J. C., and McHenry, J. R. (1990). Application of radioactive fallout Cesium-137 for measuring
soil erosion and sediment accumulation rates and patterns: a review. Journal of Environmental
Quality, 19, 215-233.
Ritchie, J.C., McHenry, J.R., and Gill, A.C. (1974). Fallout 137Cs in the soils and sediments of three
small watersheds. Ecology v. 55, p. 887-890.
Robichaud, P. R. and Brown, R. E. (2002). Silt fences: an economical technique for measuring hillslope
100
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
soil erosion. General Technical Report RMRS-GTR-94. Fort Collins, CO: U.S. Department of
Agriculture, Forest Service, Rocky Mountain Research Station. 24 p.
Roehl, J.E. (1962). Sediment source areas, and delivery ratios influencing morphological factors.
International Association ofHydrological Sciences, 59(1962), 202-213.
Roering, J. J., Mackey, B. H., Marshall, J. A., Sweeney, K. E., Deligne, N. I., Booth, A. M.,
Handwerger, A. L., and Cerovski-Darriau, C. (2013). You are HERE: connecting the dots with
airborne lidar for geomorphic fieldwork. Geomorphology 200, 172-183.
Rommens, T., Verstraeten, G., Bogman, P., Peeters, I., Poesen, J., Govers, G., Van Rompaey, A. and
Lang, A. (2006). Holocene alluvial sediment storage in a small river catchment in the loess area
of central Belgium. Geomorphology 77(1-2), 187-201.
Rosgen, D. (2006). Watershed assessment of river stability and sediment supply (WARSSS). Fort
Collins, CO, Wildland Hydrology.
Ross, K. M., Hupp, C. R., and Howard, A. D. (2004). Sedimentation in floodplains of selected
tributaries of the Chesapeake Bay. Bennett, S. J and Simon, A. (Eds.) Riparian vegetation and
fluvial geomorphology: American Geophysical Union, Water Science and Applications 8, p.
187-208.
Rovira, A., Batalla, R J., and Sala, M. (2005). Response of a river sediment budget after historical
gravel mining (the lower Tordera, NE Spain). River Research and Applications, v. 21, p. 829-
847.
Runkel, A. C., and Steenberg, J. R. (2012). Silica sand-rich bedrock layers in the Minnesota, Wisconsin,
and Iowa region of the central midcontinent: overview. Conference on the Silica Sand
Resources of Minnesota and Wisconsin: Precambrian Research Center Guidebook 12-01, p. 5-
17. Available at
http://www.d.umn.edu/prc/workshops/Guidebooks/Silica%20Sand%20Field%20Trip.pdf last
accessed June 3, 2016.
SAS Institute Inc. (2004). SAS/STAT9.1Users Guide. Cary, NC, SAS Institute Inc.
Schelske, C. L., Conley, D. J., Stoermer, E. F., Newberry, T. L., and Campbell, C. D. (1986). Biogenic
silica and phosphorus accumulation in sediments as indices of eutrophication in the Laurentian
Great Lakes. Hydrobiologia 143: 79-86.
Schenk, E.R., Hupp, C.R., Gellis, A. (2012). Sediment dynamics in the restored reach of the Kissimmee
River Basin, Florida: a vast subtropical riparian wetland. River Research and Applications 28,
1753-1767.
Schick, A.P., and Lekach, J. (1993). An evaluation of two ten-year sediment budgets, Nahal Yael,
Israel. Physical Geography, v. 14, p. 225-238.
Schoeneberger, P. J., Wysocki, D. A., Benham, E. C., Soil Survey Staff (2012). Field book for
describing and sampling soils, Version 3.0. Lincoln, NE, Natural Resources Conservation
Service, National Soil Survey Center.
Schreuder, H. T., Ernst, R., & Ramirez-Maldonado, H. (2004). Statistical techniques for sampling and
monitoring natural resources. USDA Forest Service, Rocky Mountain Research Station.
RMRS-GTR-126.
Schumm, S. A. (1977). The Fluvial System New York, NY, John Wiley & Sons, Inc.
Sharpley, A., Jarvie, H. P., Buda, A., May, L., Spears, B., and Kleinman, P. (2013). Phosphorus legacy:
overcoming the effects of past management practices to mitigate future water quality
impairments. Journal of Environmental Quality 42, 1308.
101
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
Shipp, A. and Cordy, G. E. (2002). The USGS role in TMDL assessments. U.S. Geological Survey Fact
Sheet 130-01.
Shenk, G. W. and Linker, L. C. (2013). Development and application of the 2010 Chesapeake Bay
watershed total maximum daily load model. Journal of American Water Resource Association,
49 (2013), pp. 1042-1056.
Singh, S., and Masuku, M.B. (2014). Sampling techniques and determination of sample size in applied
statistics research: an overview. International Journal of Economics, Commerce and
Management, II, (XI), 1-22.
Simm, D. J. and Walling, D. E. (1998). Lateral variability of overbank sedimentation on a Devon
floodplain. Hydrological Sciences, 43, (5), 715-732.
Simon, A., and Hupp, C. R. (1986). Channel evolution in modified Tennessee channels. Proceedings of
the Fourth Federal Interagency Sediment Conference, 2, (5), 71-82.
Simon, A. (1989). A model of channel response in disturbed alluvial channels. Earth Surface Processes
and Landforms, 14, 11-26.
Simon, A., Curin, A., Darby, S. E., Langendoen, E. J. (2000). Bank and near-bank processes in an
incised channel. Geomorphology 35, 183-217.
Simon, A., Dickerson, W., and Heins, A. (2004). Suspended-sediment transport rates at the 1.5-year
recurrence interval for ecoregions of the United States: transport conditions at the bankfull and
effective discharge? Geomorphology 58, 243-262.
Simonson, T. D., Lyons, J., Kanehl, P. D. (1993). Guidelines for evaluation of fish habitat in Wisconsin
steams. USD A Forest Service General Technical Report NC-164, St. Paul, Minnesota, 36 p.
Slattery, M. C., Walden, J., and Burt, T. P. (2000). Fingerprinting suspended sediment sources using
mineral magnetic measurements: a quantitative approach. Foster, I., (Ed.) Tracers in
Geomorphology, New York, NY, John Wiley and Sons, 309-322.
Slaymaker, O., Souch, C., Menounos, B., and Filippelli, G. (2003). Advances in Holocene mountain
geomorphology inspired by sediment budget methodology. Geomorphology v. 55, p. 305-316.
Smith, H. G., and Dragovich, D. (2008). Sediment budget analysis of slope-channel coupling and in-
channel sediment storage in an upland catchment, southeastern Australia. Geomorphology 101,
643-654.
Smith, H. U., and Blake, W.H. (2014). Sediment fingerprinting in agricultural catchments: a critical re-
examination of source discrimination and data corrections. Geomorphology, 204, 177-191.
Soil Survey Division Staff (1993). Soil survey manual. Soil Conservation Service. U.S. Department of
Agriculture Handbook 18.
Springer, G.A., Dowdy, H.S., and Eaton, L.S. (2001). Sediment budgets for two mountainous basins
affected by a catastrophic storm: Blue Ridge Mountains, Virginia. Geomorphology, v. 37, p.
135-148.
Starkel, L. (2011). Paradoxies in the development of gullies. Landform Analysis, 17, 11-13.
Stewart, H. A., Massoudieh, A., and Gellis, A. (2014). Sediment source apportionment in Laurel Hill
Creek, PA, using Bayesian chemical mass balance and isotope fingerprinting. Hydrological
Processes, onlinelibrary.wiley.com, 15 p.
Strahler, A. N. (1956). Quantitative slope analysis. Geological Society of America Bulletin, 67,
571-596.
102
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
Strahler, A. N. (1957). Quantitative analyses of watershed geomorphology. Transactions of the American
Geophysical Union, 38 (6), 913-920.
Strahler, A. N. (1958). Dimensional analysis applied to fluvially eroded landforms. Geological Society of
America Bulletin, 69, 279-300.
Sutherland, R. A. (1989). Quantification of accelerated soil erosion using the environmental tracer
caesium-137. Land Degradation & Rehabilitation, (1), 199-208.
Sutherland, R. A. (1991). Caesium-137 and sediment budgeting within a partially closed drainage basin:
Zeitschriftfur Geomorphologie, 35, (1), 47-63.
Sutherland, R.A., and Bryan, R.B. (1991) Sediment budgeting—A case study in the Katiorin drainage
basin, Kenya. Earth Surface Processes and Landforms, v. 16, p. 383-398.
Swanson, F. J., Janda, R. J., Dunne, T., Swanston, D. N. (1982). Sediment budgets and routing in forested
drainage basins: U.S. Department of Agriculture, Forest Service, PNW Forest and Range
Experiment Station. General Technical Report, PNW-141, Portland, OR, 165.
Taylor, J. K. (1990). Statistical techniques for data analysis. Chelsea, MI, Lewis Publishers, Inc, 200.
Thorne, C. R. (1993). Guidelines for the use of stream reconnaissance sheets in the field. Contract Report
HL-93-2, U. S. Army Corps of Engineer, Waterways Experiment Station, Vicksburg, MS, 91.
Trauth, M. H., Bookhagen, B., Mueller, A., Strecker, M. R. (2003). Erosion and climate change in the
Santa Maria Basin, NW Argentina during the last 40,000 yrs. Journal of Sedimentary Research
73, 82-90.
Trimble, S. W. (1974). Man-induced soil erosion on the southern Piedmont, 1700-1970. Ankeny, IA, Soil
Conservation Society of America, 180.
Trimble, S. W. (1983). A sediment budget for Coon Creek Basin in the Driftless Area, Wisconsin, 1853-
1977. American Journal of Science, 283 (5), 454-474.
Trimble, S. W., Lund, S. W. (1982). Soil Conservation and the reduction of erosion and sedimentation in
the Coon Creek Basin, Wisconsin. IJ. S. Geological Survey Professional Paper, 1234, 1-35.
U.S. Department of Agriculture, Soil Conservation Service, Soil Survey Staff, (1951). Soil survey
manual, U.S. Department of Agriculture, Agricultural Handbook No. 18, Washington, D.C., 503.
USDA Forest Service. (2005). 3.0 Phase 3 Field Guide-soil measurements and sampling 29 p. Available
at http://www.fia.fs.fed.us/library/field-guides-methods-proc/
U.S. Department of Agriculture Natural Resources Conservation Service, (2007). Sediment impact
assessments, Chapter 13. National Engineering Handbook Part 654 Stream Restoration Design.
USEPA (1999). Protocol of developing sediment TMDLs. United States: Environmental Protection
Agency, EPA-841-B-99-004.
USEPA (2001). EPA Requirements for Quality Assurance Project Plans, EPA QA/R-5. March 2001.
https://www.epa.gov/quality/epa-qar-5-epa-requirements-quality-assurance-project-plans
USEPA (2002). Guidance on Choosing a Sampling Design for Environmental Data Collection, for Use in
Developing a Quality Assurance Project Plan, EPA QA/G-5S. EPA/240/R-02/005. December,
http ://www.epa.gov/quality/qa_docs .html.
USEPA (2008). Handbook for Developing Watershed Plans to Restore and Protect Our Waters. Office of
Water, Nonpoint Source Control Branch. EPA 841-B-08-002, Chapter 8. Estimate Pollutant
Loads, available at
103
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
http://water.epa.gov/polwaste/nps/upload/2008_04_18_NPS_watershed_handbook_ch08.pdf. last
accessed Feb 21, 2015.
USEPA (2015). Aquatic Ecosystem Restoration Efforts Manage Sediment and Improve Ecosystem
Services and Functions in the Big Spring Run Watershed. USEPA NONPOINT SOURCE
SUCCESS STORY< 84l-F-15-001JJJ, 2 p. https://www.epa.gov/sites/production/files/2016-
0 l/documents/pa_bigspringrun.pdf
USEPA (2016). National Summary of State Information, Causes of Impairment in Assessed Rivers and
Streams: Accessed at
http://ofmpub.epa.gOv/tmdl_watersl0/attains_nation_cy.control#total_assessed_waters.
U.S. Geological Survey (2014). Reconfigured Channel Monitoring and Assessment Program—Monitoring
methods: Colorado Water Science Center Web page, accessed MONTH, DAY, YEAR,
http://co:water.usgs.gov/projects/rcmap/monitormethods.html (last accessed March 19, 2015).
Van der Perk, M. and Jetten, V. G. (2006). The use of a simple sediment budget model to estimate long-
term contaminant export from small catchments. Geomorphology 79, 3-12.
Van Metre, P. C., Wilson, J.T., Fuller, C.C., Callender, Edward, and Mahler, B.J. (2004). Collection,
analysis, and age-dating of sediment cores from 56 U.S. lakes and reservoirs sampled by the U.S.
Geological Survey, 1992-2001. U.S. Geological Survey Scientific Investigations Report 2004-
5184, 180 p.
Vanoni, V. A., (Ed.) (2006). Sedimentation engineering—Chapter IV, Sediment sources and sediment
yields. American Society of Civil Engineers Manuals and Reports on Engineering Practice no.
54, 261-294.
Wainer, H. (1976). Robust statistics: a survey and some prescriptions. Journal of Educational and
Behavioral Statistics, 7(4), 285-312.
Wallbrink, P.J., Roddy, B.P., and Olley, J.M. (2002). A tracer budget quantifying soil redistribution on
hillslopes after forest harvesting. Catena, v. 47, p. 179-201.
Walling, D. E. (1983). The Sediment Delivery Problem. Journal of Hydrology, 65, 209-237.
Walling, D. E. (2005). Tracing suspended sediment sources in catchments and river systems. Science of
the Total Environment, 344, (1), 159-184.
Walling, D. E., & He, Q. (1997). Use of fallout 137Cs in investigations of overbank sediment deposition
on river floodplains. Catena, 29, 263-282.
Walling, D. E., and Woodward, J. C., 1992, Use of radiometric fingerprints to derive information on
suspended sediment sources, in erosion and sediment transport monitoring programmes in river
basins: Proceedings of the Oslo Symposium, August 1992, IAHS Publication Number 210, p.
153-164.
Walling, D. E., He, Q. and Blake, W. (1999).Use of 7Be and 137Cs measurements to document short- and
medium-term rates of water-induced soil erosion on agricultural land. Water Resources Research
35. doi: 10.1029/1999WR900242. issn: 0043-1397.
Walling, D. E., Collins, A. L., and Sichingabula, H. M. (2003). Using unsupported lead-210
measurements to investigate soil erosion and sediment delivery in a small Zambian Catchment.
Geomorphology, 52, 193-213.
Waters, T. F. (1995). Sediment in streams: sources, biological effects, and control. American Fisheries
Society Monograph 7. Bethesda, MD, American Fisheries Society.
104
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
Walter, R. C., and Merritts, D. J. (2008). Natural streams and the legacy of water-powered mills. Science
319, 299-304.
Wasson, R. J., Mazari, R.K., Starr, B., and Clifton, G. (1998). The recent history of erosion and
sedimentation on the Southern Tablelands of southeastern Australia— sediment flux dominated by
channel incision. Geomorphology, v. 24, p. 291-308.
Wasson, R.J., Caitcheon, G., Murray, A.S., McCulloch, M., and Quade, J. (2002). Sourcing Sediment
Using Multiple Tracers in the Catchment of Lake Argyle, Northwestern Australia. Environmental
Management v. 29, p. 634-646.
Wijdenes, D. J. O., and Bryan, R. (2001). Gully-head erosion processes on a semi-arid valley floor in
Kenya: A case study into temporal variation and sediment budgeting. Earth Surface Processes
andLandforms, 26, 911-933.
Wilkinson, B. H., and McElroy, B. J. (2007). The Impact of Humans on Continental Erosion and
Sedimentation. Geological Society of America Bulletin 119(1-2), 140-56.
Wilkinson, S., Hancock, G., Bartley, R., Hawdon, A., Keen, R., 2013. Using sediment tracing to assess
processes and spatial patterns of erosion in grazed rangelands, Burdekin River basin, Australia.
Agriculture Ecosystems and Environment 180, 90-102.
Williamson, J. E., Haj, A. E., Stamm, J. F., Valder, J. F., and Prautzch, V. L. (2014). Fluvial sediment
fingerprinting—literature review and annotated bibliography. U.S. Geological Survey Open-File
Report 2014-1216.
Wilson, J. P., and Lorang, M. S. (2000). Chapter 6, Spatial Models of Soil Erosion and GIS. Spatial
Models and GIS. New Potential and New Models, 83-86.
Wolf, K. L., Ahn, C., Noe, G. B. (2011). Microtopography enhances nitrogen cycling and removal in
created mitigation wetlands. Ecological Engineering, 37, 1398-1406.
Wolman, M. G. (1959). Factors influencing erosion of a cohesive river bank. American Journal of
Science 257, 204-216.
Wolman, M. G., Leopold, L. B. (1957). River floodplains: some observations on their formation. U.S.
Geological Survey Professional Paper 282-C, 87-107.
Wynn, T. (2006). Streambank retreat—a primer: watershed update, January-March 2006. American
Water Resources Association Hydrology and Watershed Management Technical Committee, 4( 1),
1-14.
Young, B. M., Fitzpatrick, F. A., and Blount, J. D. (2015). Stream geomorphic and habitat data from a
base-line study of Underwood Creek, Wisconsin, 2012. U.S. Geological Survey Data Series 947.
Ypsilantis, W. G. (2011). Upland soil erosion monitoring and assessment: an overview. Bureau of Land
Management, National Operations Center, Denver, CO, Tech Note 438 35.
105
-------
A Manual to Identify Sources of Fluvial Sediment
EPA/600/R-16/210
PRESORTED
STANDARD POSTAGE
& FEES PAID EPA
PERMIT NO, G-35
United States
Environmental Protection
Agency
Office of Research
and Development
(8101R)
Washington, DC
20480
Official Business
Penalty for Private Use$300
106
------- |