ABSTRACT 
 
 
Title of Document: QUANTIFYING FINE SEDIMENT 
SOURCES IN THE NORTHEAST BRANCH 
OF THE ANACOSTIA RIVER, 
MARYLAND, USING TRACE ELEMENTS 
AND RADIONUCLIDES  
  
 Olivia Harcourt Devereux 
Masters of Science, 2006 
Directed By: Assistant Professor Brian A. Needelman, 
Department of Environmental Science and 
Technology 
 
 
Fine sediment sources were characterized in an urban watershed, the Northeast 
Branch of the Anacostia River, which drains to the Chesapeake Bay. Concentrations 
of 63 elements and two radionuclides were measured in possible sediment sources 
and suspended sediment collected at the watershed outlet during storm events. 
Methodology for selecting tracers was developed so the sediment fingerprinting 
method could effectively determine the relative quantity of sediment contributed by 
each source to the suspended fraction. The amount of enrichment of trace elements in 
sediment sources and suspended sediment was determined by calculating enrichment 
ratios, which are ratios of the normalized concentration of elements in the sample 
relative to their average normalized concentration in the Earth?s upper continental 
crust. Streambanks contributed the highest relative quantity of sediment in the fall 
and spring while upland areas contributed mostly during winter. Street residue 
contributed 12% on average and was the source most concentrated in 
anthropogenically enriched elements.  
 
 
 
 
 
 
 
 
QUANTIFYING FINE SEDIMENT SOURCES 
IN THE NORTHEAST BRANCH OF THE ANACOSTIA RIVER 
USING TRACE ELEMENTS AND RADIONUCLIDES 
 
 
 
By 
 
 
Olivia Harcourt Devereux 
 
 
 
 
 
Thesis submitted to the Faculty of the Graduate School of the  
University of Maryland, College Park, in partial fulfillment 
of the requirements for the degree of 
Masters of Science 
2006 
 
 
 
 
 
 
 
 
 
 
Advisory Committee: 
Assistant Professor Brian A. Needelman, Chair 
Dr. Allen C. Gellis 
Professor Bruce R. James 
Associate Professor Karen L. Prestegaard 
 
 
  
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
? Copyright by 
Olivia Harcourt Devereux 
2006 
 
 
 
 
 
 
 
 
 
 
 
  
 
Dedication 
 
To acknowledge their patience and confidence in me, this work is dedicated to my 
husband Erik Devereux and children Alexander Devereux and Mildred Devereux. 
 ii 
 
 
Acknowledgements 
 
Many persons offered vital support and assistance for this research. Professor Bruce 
James gave me the confidence to go to graduate school and become a scientist. Dr. 
Allen Gellis at the U.S. Geological Survey in Baltimore first suggested this project 
idea and provided direction throughout. Assistant Professor Brian A. Needelman, my 
advisor, gave me the opportunity and freedom to conduct this research. Dr. Jerry 
Ritchie, of the U.S. Department of Agriculture?s Agricultural Research Service 
Hydrology Lab in Beltsville, Maryland, helped me to understand radionuclides and 
performed the analyses of them. He also assisted with developing the concepts for the 
project proposal. William S.L. Banks, U.S. Geological Survey in Baltimore, 
Maryland, helped with centrifuging the samples. Dr. Jurate Landwehr, U.S. 
Geological Survey in Reston, Virginia, provided the mixing model. Professor Martin 
Rabenhorst, Dave Verdone (NRCS) and Eddie Earls (NRCS) trained me in field work 
and soil descriptions. Mr. Christopher Brosch was the undergraduate research 
assistant who assisted with lab work. Steve Burch, formerly with the Pedology 
Laboratory, gave me specific tips for making both lab and field work a little easier. 
Professor Richard Erdman shared his approach to scientific research, and a 
dairyman?s knowledge of the landscape. Professor Karen L. Prestegaard provided 
valuable advice at all times and constant intellectual guidance for which she has my 
deepest appreciation. This research received initial funding from the Maryland Water 
Resources Research Center. 
 iii 
 
 
Table of Contents 
 
 
Dedication..................................................................................................................... ii 
Acknowledgements......................................................................................................iii 
Table of Contents.........................................................................................................iii 
List of Tables ................................................................................................................ v 
List of Figures.............................................................................................................. vi 
Chapter 1: Introduction................................................................................................. 1 
Problem..................................................................................................................... 1 
Background............................................................................................................... 2 
Objectives ................................................................................................................. 3 
Description of Study Area ........................................................................................ 3 
Site Selection and Sediment Sampling ..................................................................... 6 
Tracer Selection ........................................................................................................ 7 
Hypotheses................................................................................................................ 8 
Chapter 2: Historical Patterns of Erosion and Sediment Storage Shown by Pb and Cs-
137................................................................................................................................. 9 
Introduction............................................................................................................... 9 
Study Site and Methods .......................................................................................... 11 
Study Site............................................................................................................ 11 
Sampling Methodology....................................................................................... 14 
Lab Analyses....................................................................................................... 15 
Results..................................................................................................................... 17 
Pb and Cs-137..................................................................................................... 17 
Clay Mineralogy ................................................................................................. 22 
Soil Survey.......................................................................................................... 23 
Discussion............................................................................................................... 23 
Chapter 3: Geogenic and Anthropogenic Sources of Fine Sediment in an Urban 
Watershed ................................................................................................................... 25 
Introduction............................................................................................................. 25 
Study Site and Methods .......................................................................................... 26 
Study Site............................................................................................................ 26 
Sample Locations................................................................................................ 28 
Sampling Methodology....................................................................................... 29 
Lab Analyses....................................................................................................... 30 
Results..................................................................................................................... 33 
Sediment Source Characterization...................................................................... 33 
Suspended Sediment Characterization................................................................ 35 
Comparison to the Earth?s Mean Crust Composition......................................... 37 
Discussion............................................................................................................... 40 
Chapter 4: Quantifying Fine Sediment Sources in the Northeast Branch of the 
Anacostia River........................................................................................................... 42 
Introduction............................................................................................................. 42 
Study Site and Methods .......................................................................................... 43 
 iii 
 
 
Watershed description......................................................................................... 43 
Sampling Strategy............................................................................................... 46 
Lab analyses........................................................................................................ 48 
Results..................................................................................................................... 54 
Suspended Sediment Characterization................................................................ 54 
Sediment Source Group Characterization........................................................... 56 
Discussion............................................................................................................... 62 
Chapter 5: Conclusions and Future Research ............................................................. 63 
Summary................................................................................................................. 63 
Connections to Other Research............................................................................... 65 
Future Research ...................................................................................................... 65 
Appendix A: Soil Descriptions ................................................................................... 67 
Appendix B: Distribution of Sediment Source Samples............................................. 80 
Appendix C: Raw Data ............................................................................................... 81 
Appendix D: Detection Limits.................................................................................... 85 
Appendix E: Sedigraphs and Hydrographs for Suspended Sediment Sample 
Collection Times......................................................................................................... 86 
Appendix F: Particle-Size Distributions ..................................................................... 87 
Appendix G: pH data .................................................................................................. 89 
Appendix H: Periodic Table ....................................................................................... 90 
References................................................................................................................... 91 
 
 
 iv 
 
 
List of Tables 
 
2-1 Soil series in the Northeast Branch watershed.......................................... 14 
2-2 Concentration of Cs-137 corrected for particle size ................................. 16 
2-3 Tracer variation......................................................................................... 18 
3-1 Suspended sediment sample characteristics.............................................. 31 
3-2 Element analytical methods ...................................................................... 32 
3-3 Tracers that differentiated hydrological subwatershed boundary............. 33 
3-4 Tracers that differentiate the landscape positions..................................... 35 
4-1 Suspended sediment sample characteristics.............................................. 49 
4-2 Element analytical methods ...................................................................... 50 
4-3 Bracketed tracers....................................................................................... 51 
4-4 Elemental concentrations in suspended sediment..................................... 54 
4-5 Tracers that differentiate sediment sources by hydrologic subwatershed 
................................................................................................................... 57 
4-6 Estimation of subwatershed sediment contribution using Ga, Ta, Ce, U 
................................................................................................................... 58 
4-7 Tracers that differentiate sediment sources by landscape position........... 59 
4-8 Estimation of landscape position sediment contribution using Ho, Sr, W 
................................................................................................................... 59 
4-9 Alternative tracers that differentiate sediment sources by hydrologic 
subwatershed............................................................................................. 60 
4-10 Alternative estimation of subwatersheds sediment................................... 61 
4-11 Alternative tracers that differentiate sediment sources by landscape 
position...................................................................................................... 61 
4-12 Alternative estimation of landscape position sediment contribution........ 61 
 v 
 
 
List of Figures 
1-1 Land use of study site ................................................................................. 4 
2-1 Northeast Branch sample sites.................................................................. 12 
2-2 Pb and Cs-137 concentrations................................................................... 18 
2-3 Clay mineralogy........................................................................................ 22 
3-1 Northeast Branch sample sites.................................................................. 27 
3-2 Tracers not bracketed by sediment sources .............................................. 36 
3-3 Enrichment by subwatershed .................................................................... 39 
3-4 Enrichment by landscape positions........................................................... 40 
4-1 Northeast Branch sample sites.................................................................. 44 
4-2 Daily average discharge............................................................................ 48 
4-3 Covariance between tracers in the Lanthanides group ............................. 56 
4-4 Covariance between tracers used in the mixing model............................. 57 
 
 vi 
 
 
Chapter 1: Introduction 
Problem 
Sediment is defined as particles derived from rocks and soils that are transported and 
deposited (USGS, 2004). Organic material is also included in the transport process 
(SSSA, 2006). In urban areas, sediment may be derived from natural hillslopes, yards 
in residential areas, floodplains, stream banks, stream beds, and from residue that 
accumulates on roads. Particulate matter may also be atmospherically deposited 
within the watershed. The transport process requires that sediment first become 
mobilized through water or wind erosion, and then transported from its source. 
Although significant amounts of transported sediment are stored within a river 
system, some of this sediment, particularly the fine-grained fraction, travels 
downstream to bays and oceans. 
 
Sediment is the leading water pollutant in the United States (Woodward and Foster, 
1997). The U.S. EPA has regulated sediment in waterways by establishing Total 
Maximum Daily Loads (TMDL). Suspended sediment diminishes light penetration in 
water, which affects submerged aquatic vegetation, and affects habitat for benthic 
biota such as blue crabs and rockfish. The health and extent of sub-aquatic vegetation 
is used as an indicator of the level of impairment caused by fine sediment in water 
bodies. The loss of aquatic life has a direct economic impact, particularly when it 
affects fisheries and the service industries that depend on seafood production.  
 
In addition to ecosystem problems caused by light attenuation due to suspended 
sediment in the water column, fine-grained sediment transport can also affect the 
concentration and distribution of sediment-bound contaminants. Some of the most 
common contaminants that adhere to sediment include heavy metals, nutrients 
(particularly P and K), and organic substances. For example, 57% of streams sampled 
in the U.S. exceed the EPA total phosphorus standards (U.S. Geological Survey, 
1999).  
 
Knowledge of sediment sources is essential in efforts to restore sediment-impaired 
watersheds. This is important for two reasons: 1) identification of sediment sources is 
the first step in reducing sediment pollution in water bodies, and 2) sediment 
characteristics (oxyhydroxide coatings, organic matter content, etc.) develop in situ in 
the landscape; these characteristics in turn influence chemical and physical properties 
(Marvin et al., 2004). These properties affect the type and amount of contaminants 
that may adsorb to sediments. In addition, sediment loss is not equally distributed in 
the landscape. Knowledge of sediment sources and their relation to geology and soil 
characteristics would be a significant step for science-based management practices to 
be targeted to zones at high risk for sediment loss. 
 1 
 
 
Background 
A sediment budget for a watershed consists of measurements of sediment input, 
sediment storage, and sediment output from the watershed. Sediment input into the 
watershed results from hillslope and channel erosion. Erosion measurements give a 
direct measurement of the contribution of sediment from that source to the sediment 
budget. For example, direct measurements of channel bank erosion are made by 
examining changes over time in channels using aerial photography, erosion pins, and 
sequential surveys of streams (Dunne and Leopold, 1978). Erosion measurements 
give a direct measurement of erosion rates from specific sources, but in most 
watersheds, significant amounts of the sediment will be stored within the basin. Thus, 
sediment erosion measurements can not be used to directly measure the sources of 
sediment that is exported from the basin.  
 
Sediment transport and storage in a watershed can be determined by direct 
measurements of suspended sediment load within a watershed system. The sediment 
load is measured at various places in a stream network, and sources are estimated 
from the various loads. However, extensive sediment monitoring networks are 
required to determine sediment loads, losses due to storage, and relationship between 
sediment loads and sediment sources. Detailed sediment budgets require 
measurements of both erosion from sources and transport of sediment within 
watersheds. This method is expensive and time-consuming (Dunne and Leopold, 
1978).  
 
Sediment fingerprinting is a method that was developed to determine the sources of 
the suspended fraction that is transported out of a watershed. This method compares 
sediment from a variety of different potential sources with suspended sediment 
collected at a downstream location. The method compares the amount of various 
tracers, which may be geochemical, mineralogical, or mineral magnetic, in various 
sediment sources to the suspended sediment. First developed in the 1970s, sediment 
fingerprinting determines sediment sources by using physical or chemical properties 
(Collins et al., 1997a; Walling and Woodward, 1992; de Boer and Crosby, 1995; 
Wallbrink et al., 1998; Walling et al., 1999; Rowan et al., 2000; Owens et al., 2000; 
Jansson 2002; Gruszowski et al., 2003; Jenkins et al., 2002; Motha et al., 2003; 
Papinicolaou et al., 2003; Motha et al., 2003; Rhoton et al., 2005; Walling, 2005; 
Carter et al., 2003; Gellis and Landwehr, 2006).  
 
Sediment source identification and quantification techniques have advanced 
substantially in the past five years. The use of the various tracers would not be 
possible if the tracer characteristics change through the erosion process (Motha et al., 
2002; Gao et al., 2005). Previously, researchers investigated the best way to use 
radionuclides as a characteristic of fingerprints (Wallbrink et al., 1998; Nagle and 
Ritchie, 2004). Using stable or radiogenic isotopes as part of a fingerprint has also 
been researched (Papanicolaou et al., 2003). Determining the effects of particle size 
has been thoroughly explored by a variety of authors (Motha et al., 2003; Marvin et 
al., 2004). While fingerprinting originally differentiated sources by land use, the 
 2 
 
 
impact of erosion of roads in rural areas has also been investigated (Gruszowski et al., 
2003). Clays also have been used as a component in fingerprints with success 
(Deboer and Crosby, 1995). The sediment fingerprinting method has become 
accepted as a valid method for identifying sediment sources (Walling, 2005). 
 
Sediment fingerprinting has not been used widely in watersheds with urban areas 
(Carter et al., 2003). The tracers that are commonly used in sediment fingerprinting 
research may not be effective in urban areas (Jeelani and Shah, 2006, Leung and Jiao, 
2006). Urban areas likely have a stronger anthropogenic component than rural or 
forested areas, resulting in a different chemical fingerprint. In addition, urbanization 
may decrease sediment storage areas because buildings are constructed on floodplains 
? resulting in impervious surfaces at just those locations where sediment could be 
stored as part of natural stream processes (Jacobson and Coleman, 1986). Also, 
impervious surface areas, such as parking lots, buildings, and concrete or asphalt 
roads, are not erodible so the areas that are sediment sources are necessarily different 
than those in rural or forested watersheds. 
Objectives 
The purpose of this research is to extend the sediment fingerprinting methodology to 
discriminate and quantify sediment sources in an urban watershed. In comparing the 
sources of sediment to suspended sediment, sources are grouped by subwatershed and 
also by landscape position. An adapted mixing model (Gellis and Landwehr, 2006) is 
used to determine the relative quantity of sediment derived from various sources.  
 
Testing the applicability of sediment fingerprinting methodology in urban areas is 
important for several reasons. Developed land is increasing at a rapid rate in many 
parts of the world. In the Washington, DC area, developed land increased 64% from 
2,544 to 4,167 km
2
 between 1986 and 2000 (Jantz et al., 2005). Legislation requiring 
water quality improvement has been used as the basis for establishing total maximum 
daily load (TMDL) requirements for sediment. There is a sediment TMDL 
established for the Anacostia River (U.S. EPA, 2006). Decisions on how best to 
manage and improve watersheds with sediment-impaired water require information 
about the sources of sediment.  
Description of Study Area 
The Chesapeake Bay watershed encompasses 168,000 km
2
 and includes parts of the 
states of New York, Pennsylvania, West Virginia, Maryland, Delaware, Virginia and 
all of the District of Columbia. The Anacostia River drains northern Washington D.C. 
and adjacent portions of Maryland to the Chesapeake Bay. The study area is the 
Northeast Branch of the Anacostia River watershed, which is northeast of 
Washington, D.C. in Maryland. The drainage area of the Northeast Branch watershed 
is 188.5 km
2
. 
 
 3 
 
 
Presently the Anacostia River watershed, which includes the Northeast Branch 
watershed, is 64% urban, 27% forested, 8% agricultural, <? % wetlands, and <? % 
other categories (Fig 1-1) (Maryland Department of Natural Resources, 2000). 
Impervious land is 33% (Maryland Department of Natural Resources, 2000). The 
Northeast Branch area of the Anacostia is the oldest developed portion of the 
watershed, developed primarily in the time period between 1960 and 1970 (Jacobson 
and Coleman, 1986). While construction projects are currently taking place, there are 
no current wide-spread Earth-moving projects.  
 
{
Streams
Urban
Forest
Agriculture
Recreational Grass
Sample_locations
Pysiographic_Boundary
02,2504,501,125 Meters
 
Fig: 1-1 Land Use of Study Site. Note that recreational grasses are ball fields.  
 
The Northeast Branch watershed spans two physiographic regions ? the Piedmont and 
the Coastal Plain. The physiographic province reflects the geology and topography. 
Coastal Plain geology primarily comprises silt-clay facies of the Potomac Group 
(Patuxent Formation), which is Early Cretaceous in age. The Potomac Group is 
 4 
 
 
fluvially deposited quartz, with some sandstone and chert. The Coastal Plain also 
includes Arundel Clay and much of the lower Patapsco Formation. The Coastal Plain 
in this area is highly-weathered Piedmont material from fluvial deposition. Upland 
deposits of the Coastal Plain are fluvial sediments of the ancestral Potomac River and 
are of the late Miocene and Pliocene periods. Terrace deposits are the product of 
stream erosion during the early Quaternary Period. These are primarily 
unconsolidated sediments. The Coastal Plain ranges in age from the Cretaceous to 
Quaternary periods (Markewich et al, 1987).  
 
Mapped units in the Piedmont, including the Laurel formation, are characterized by 
Precambrian crystalline metamorphic rocks with occasional intrusions of igneous 
rocks. The granite, gneiss, schist, and other crystalline rocks range in age from 
Precambrian to late Paleozoic. Piedmont upland deposits are high in vermiculite clay 
(Cleaves et al., 1968; Hack, 1977; Glaser, 1996; Csato et al., 2006). 
 
In addition to encompassing two physiographic regions, the Northeast Branch 
watershed is made up of two subwatersheds. The west subwatershed is influenced by 
the Piedmont geology and the east subwatershed is characterized by the Coastal Plain 
geology. The subwatershed boundary is mapped east of the physiographic boundary. 
In the west subwatershed the land use is urban?a combination of commercial and 
residential. There are numerous wooded parks along the streams. In the east 
subwatershed land use is industrial, commercial, residential, forested and agricultural. 
There are gravel pits along the headwaters of a stream in the east subwatershed. 
Exhausted gravel pits are used as dumps for yard waste and they are also sites for 
suburban development. Also located in the east subwatershed is rural land that has 
long been managed as farmland by USDA-ARS and a forested national park.  
 
Four main tributaries make up the Northeast Branch watershed. From west to east 
they are: Paint Branch, Little Paint Branch, Indian Creek, and Beaverdam Creek. The 
Little Paint Branch drains to the Paint Branch. Beaverdam Creek drains to Indian 
Creek. The confluence of Paint Branch and Indian Creek form the Northeast Branch 
of the Anacostia. The Piedmont streams include the upstream reaches of the Paint 
Branch and Little Paint Branch and are gravel and boulder bedded. Some of the 
boulders range to several meters in diameter. The floodplains are small, between three 
and six meters in many locations. The region is hilly and slopes are high and steep. 
The soils are more fine-grained than those in the Coastal Plain. The downstream 
reaches of these streams are characterized by fewer boulders and more sand.  
 
The Coastal Plain streams include the entirety of Indian Creek and Beaverdam Creek 
and the lower portions of Paint Branch and Little Paint Branch. The streams are sand, 
silt, and clay bottomed. Sand bars migrate through the two streams that originate in 
the Piedmont. The Coastal Plain is a low rolling plain. The Coastal Plain soils are 
sandier than those of the Piedmont. 
 
 5 
 
 
The downstream portions of the streams are channelized with cement, gabions, or 
granite blocks. The channelization inhibits sediment storage and increases stream 
flow velocity by decreasing friction.  
 
The climate is humid continental with an average temperature of 13.2 ?C and an 
average annual rainfall of 117 cm (NOAA 2005). There are four distinct seasons with 
mild winters. Rainfall is evenly distributed throughout the year; storms with broad 
aerial extent occur primarily in winter months. There are isolated thunderstorms in 
the summer that deliver rain quickly and create flashy stream flow. These summer 
thunderstorms typically do not span the entire watershed. 
Site Selection and Sediment Sampling 
Sediment sources were assumed to be the following: upland (non-channel) sites, 
stream bank sites, and sediment on paved surfaces. Therefore, a sampling strategy 
was developed to sample these sites throughout the stream system and within both 
physiographic provinces. Sediment source sample locations were located along 
reaches of each of the four tributaries and the main stem to represent both 
physiographic regions, both subwatersheds, multiple landscape positions, and each 
land use in the Northeast Branch watershed. Samples were collected from upstream 
and downstream reaches of the streams. The objective was to collect sediment source 
samples so that they could be grouped to maximize variation between groups. It was 
not known beforehand which grouping would result in maximum variation between 
groups.  
 
The high percentage of impervious surfaces necessitated more soil samples from 
banks because much of the upland area was paved. Of the bank samples, there were 
three sample sites along the Paint Branch, two along the Little Paint Branch, four 
along Indian Creek, five along Beaverdam, and three along the Northeast Branch. 
Bank sample sites were chosen where the stream width or depth changed, topography 
changed, there was a confluence, and/or significant erosion was visible. These 
geomorphic changes in the stream areas indicate a possible change in soils and 
sediments from the stream reaches above those areas. At five of these stream bank 
sites, samples were taken along a transect moving away from the bank to include the 
landscape positions of stream bank, floodplain, and upland. Depending on the 
topography, the upland position was either a backslope or a summit position. These 
five sites were selected to represent physiographic region and land use. 
 
The watershed included many urban areas that could contribute sediment with 
significantly different chemical characteristics to the stream. Additional sediment 
inputs in this urban area include street residue, sewage, and atmospheric particulate 
matter. Street residue was collected for both physiographic regions in the urbanized 
portions of the watershed. One construction site located adjacent to the stream and 
very near the suspended sediment collection site was sampled as were four point bars. 
It was outside the scope of this project to collect sewage or atmospheric particulate 
 6 
 
 
matter. However, atmospheric particulate matter data was available from other 
watersheds for comparison.  
 
Suspended sediment samples were collected at the outlet of the Northeast Branch 
watershed. The sediment was collected during storm events by pumping stream water 
into three 208 liter plastic-lined barrels in the back of a pick-up truck. The samples 
were driven to a continuous centrifuge and the sediment was recovered. Suspended 
sediment sampling was collected is such a way as to maximize the amount of 
sediment collected. Enough sediment had to be available to perform the lab analysis 
necessary to measure the concentration of the various tracers. This is especially true 
for the radionuclide analysis where error decreases with increased sample mass.  
 
The Northeast Branch watershed is a sediment-limited system with average turbidity 
of 11.8 and median turbidity of 2.9, as measured by U.S. Geological Survey from 
September 2005 to February 2006 using monochrome near infra-red LED light, 780-
900 nm, detection angle 90 +/ -2.5 degrees. Maximum turbidity occurs during the 
rising limb of the hydrograph of storm events (Appendix E). Thus, sampling was 
targeted for the rising limb of the hydrograph for storm events that covered the entire 
188 km
2
 watershed. One of the assumptions in this research is that travel times of the 
sediment are equal. By sampling at a single point in time, the sample may only 
represent sources that reached the gauge by that point in the storm. Thus, the furthest 
point from the gauge may not be included in the suspended sediment sample. 
Moreover, it is assumed that all sediment travels at an equal rate?that is organic 
particles travel at the same rate as micaceous particles or fine sand grains or clays. K 
concentration data from the sediment samples indicates that this may be false. 
However, this was not fully explored as part of the study. 
 
Since it is never known exactly how long a storm will last or the variation in intensity 
throughout a storm, not all storms were sampled at exactly the same point in the 
hydrograph (Appendix E). In addition, it was deemed dangerous to pull the truck 
close to the bank of the stream in icy conditions or darkness, so most sampling 
occurred during daylight.  
Tracer Selection 
The Northeast Branch of the Anacostia has different physiology, geology, and land 
use from other watersheds where sediment fingerprinting research has been done. 
Therefore I did not presume to know which source groups could be differentiated by 
use of tracers or which tracers would be most effective for source group 
identification. For this reason, I measured the concentrations of 63 elements and 2 
radionuclides in an initial set of sources and suspended sediment samples. Knowing 
the components of suspended sediment samples informed the selection of 
characteristics for a composite tracer fingerprint of source areas. Thus, I engaged in 
an iterative process whereby significant characteristics in the suspended sediments 
were determined at the same time as the sediment sources. Other sediment 
 7 
 
 
fingerprinting researchers tend to be more economical in their choice of tracers, but 
that was not an option because I did not know what I would find in this watershed.  
Hypotheses 
Specifically, the hypotheses tested were as follows. 
 
1 Chemical analyses of sediment can be used to determine the contributions of 
sediment sources in urban watersheds.  
2 Sediment source compositions used in the sediment fingerprinting of urban 
watersheds will include anthropogenically- enriched elements. 
3 The primary source of suspended sediment in the Northeast Branch is derived 
from stream bank erosion. 
 
In Chapter 2: Historical Patterns of Erosion and Sediment Storage, the objective was 
to conduct a soil survey informed by the distribution of Cs-137 and elemental Pb to 
show the patterns of erosion and sediment storage in this watershed. Variable land use 
change throughout the watershed was established using the distribution of Cs-137 and 
Pb. This preliminary site exploration was the basis for establishing options for 
grouping sediment sources. 
 
When testing the sediment sources and suspended sediment for the 63 elements and 2 
radionuclides, I found the first hypothesis to be supported but not the second. That is, 
chemical analyses were effective in determining sediment source groups, but those 
elements that were effective were not anthropogenic. So in Chapter 3, Geogenic and 
Anthropogenic Sources of Fine Sediment in an Urban Watershed, I show how to 
adjust the methodology to accommodate this anthropogenic chemical overprint from 
the variable land use change. Chapter 4, Quantifying Fine Sediment Sources, actually 
discriminates the sources and uses a mixing model to show the relative quantity of 
total sediment collected during the storm even was generated from each source. The 
banks indeed provided the most sediment, thereby supporting the third hypothesis.  
 8 
 
 
Chapter 2: Historical Patterns of Erosion and Sediment Storage 
Shown by Pb and Cs-137 
Introduction 
In urban areas, land use and land use history can be highly variable. Land use can 
change from forested to agriculture to urban. Erosion can occur when forests are 
clear-cut or burned, when tillage disturbs soil structure and fields are seasonally bare, 
and when construction moves or removes the soil. Within the same land use, 
management practices may vary and affect rates of erosion. Agricultural methods of 
tilling have changed from traditional moldboard plowing to conservation tillage or 
no-till (Brady and Weil, 2002). More erosion occurs under deep tillage as with the 
moldboard plow than under conservation or no-till agricultural practices. 
Urbanization resulted in an immediate erosive event as foundations were dug and 
land leveled for buildings. Once urbanized, much of the land is impervious and that 
which is permeable is largely unperturbed. These variations in land use and 
management each resulted in changes in erosion patterns (Wolman and Schick, 
1967). As land use shifted and erosion regimes changed, pulses of sediment moved 
through the streams (Brady and Weil, 2002; Wolman and Schick, 1967).  
 
For more than 30 years, Cs-137 has been used to determine rates of erosion and 
sedimentation (Ritchie, 2006). In the past 20 years, Cs-137 also has been used to 
identify the land use from which sediment originated (Ritchie, 2006). Cs-137 was 
atmospherically deposited during 1954-1975 with a significant increase during the 
year of 1964, all as a result of nuclear weapon tests. Cs-137 sorbs immediately to 
negatively charged soil particles and is not readily exchanged because it is 
monovalent, has a small hydrated radius, and a high ionic potential (Ritchie and 
McHenry, 1990). Cs-137 was deposited evenly over the land as atmospheric 
particulates, and then diluted as soil was mixed from land use change or by soil 
management practices.  
 
Surface concentrations of Cs-137 therefore reflect mixing or dilution of soils. 
Categorizing Cs-137 into levels of zero or not detectable, medium, or high allows one 
to understand how the land use changes and erosion regimes affect the Cs-137 
concentrations. The highest concentrations of Cs-137 are found in undisturbed areas 
such as forests, or where soils were translocated from undisturbed areas and not 
diluted (Nagle and Ritchie, 2004; Wallbrink et al., 1998). Conversely, a soil may have 
a Cs-137 concentration of zero under three scenarios: 1) the soil could be on a vertical 
surface where atmospherically deposited Cs-137 would not have come in contact with 
the soil surface, 2) the soil may be deep enough that it was not exposed during the 
years 1954-1975, or 3) a subsoil may have been exposed or relocated to a surface 
since 1975. Moderate Cs-137 concentrations are a result of some other perturbation?
either mixing or dilution. Moderate values of Cs-137 concentration are generally 
associated with agricultural land. This land is plowed so the soil is perturbed by the 
plow. While some farmers have plowed very close to streams, the actual banks are 
 9 
 
 
not plowed. Therefore, streambank soils may have moderate Cs-137 concentrations 
because of mixing by stream-generated erosion and floodplain deposition and/or by 
hillslope processes of erosion and footslope and valley deposition.  
 
While Cs-137 is commonly used to identify land use, it is less useful in identifying 
sediment sources in urban areas where land use has changed rapidly over time and is 
variable over small areas. Pb may provide complementary information useful for 
determining patterns of erosion and storage in urban areas (Juracek and Ziegler, 
2006).  
 
Pb in the environment comes from atmospheric sources as does Cs-137, but also from 
direct deposition (Callender, 2004). Natural sources of Pb are from weathering of the 
Earth?s upper continental crust and volcanic eruptions. Anthropogenic inputs are 
much greater than natural inputs in urban areas?17 mg kg
-1
 in the Earth?s upper 
continental crust versus 26 mg kg
-1 
in the average soil (Callender 2004). 
Anthropogenic inputs of Pb into the environment generally come from fossil fuel 
combustion?primarily coal, municipal waste incineration, cement manufacturing, 
sewage sludge discharge, fertilizer application, and animal wastes.  
 
Heightened anthropogenic inputs of Pb occurred over a specific time period. Pb was 
commonly used in paint and gasoline until 1978 (EPA, 2006). Pb particularly 
increased in the environment due to the burning of Pb alkyls in gasoline after 1940. 
U.S. regulation of Pb came as a result of an increase in lead poisoning, which has 
negative health effects on the human brain. The regulations led to a decrease from the 
1970s to the 1980s of Pb emissions to the air. In the 1970s, global primary Pb 
emission was 449,000 metric tons but only 332,000 metric tons in the 1980s 
(Callender, 2004). Once these sources of Pb were limited by federal regulations in 
1978, the amount of new additions of Pb to the environment sharply decreased. This 
peak of Pb in the environment may be used to track erosion and sediment deposition. 
 
The study area is a watershed where the land use was converted from forest to 
agriculture to urban. Sediment control measures were instituted in the first half of the 
1900s during the agricultural period, so the agricultural land became less erosive. The 
main period of urban development occurred in the 1950s (Moglen and Beighley, 
2002). There are minor renovations and some in-filling, but there currently are no 
extensive Earth-moving operations in this watershed. Land that remains in agriculture 
is 26% and is located on the U.S.D.A. Agricultural Research Station. Forested land 
comprises 12% and is located in a protected national park (not including riparian 
areas with trees); the remainder is urban. This known land-use history provides an 
ideal location to test the combined use of Cs-137 with elemental Pb to yield historical 
patterns of soil erosion and sedimentation.  
 
Therefore, I hypothesize that: 1) the pulses of sediment generated from land use 
changes were eroded and stored in different stream reaches; and 2) these historical 
erosion and sediment storage patterns can be determined by the concentrations of Cs-
137 and Pb in the streambanks. Thus, by comparing the distribution of Pb with Cs-
 10 
 
 
137, the historical patterns of erosion and alluviation may be revealed in an urban 
watershed.  
 
Study Site and Methods 
Study Site 
The Northeast Branch drains to the Anacostia River, which drains to the Chesapeake 
Bay. The drainage area is 188.5 km
2
. The Northeast Branch spans two physiographic 
regions?the Piedmont to the west and the Coastal Plain to the east (Fig 2-1). The 
physiology is differentiated by the geology. The Piedmont geology primarily 
comprises sand-gravel facies of the Potomac Group (Patuxent Formation), which is 
Early Cretaceous in age. The Potomac Group is fluvially deposited quartz, with some 
sandstone and chert. Upland deposits of the Piedmont are fluvial sediments of the 
ancestral Potomac River and are of the late Miocene and Pliocene periods. The 
upland deposits are high in vermiculite clay. Other mapped units in the Piedmont, 
including the Laurel formation, are characterized by crystalline metamorphic rocks 
with occasional intrusions of igneous rocks. The granite, gneiss, schist, and other 
crystalline rocks range in age from Precambrian to late Paleozoic (Glaser, 1996; 
Csato et al., 2005). The Coastal Plain geology is characterized as silt-clay facies of 
the Potomac Group, comprised of Arundel Clay and much of the lower Patapsco 
Formation. The Coastal Plain in this area is highly-weathered Piedmont material from 
fluvial deposition. Terrace deposits are the product of stream erosion during the early 
Quaternary Period. The Coastal Plain ranges in age from the Cretaceous to 
Quaternary periods. The Piedmont physiographic region encompasses all but the 
southernmost reaches of the Paint Branch and the northern third of the Little Paint 
Branch. The Coastal Plain physiographic region includes the southern portion of the 
Paint Branch, the southern two-thirds of the Little Paint Branch, Indian Creek, 
Beaverdam Creek, and the Northeast Branch. 
 
There is a subwatershed boundary that falls to the east of the physiographic region 
boundary. Four streams drain to the Northeast Branch with the Paint Branch located 
furthest west. The Little Paint Branch drains into the Paint Branch two-thirds the 
length of the Paint Branch. Indian Creek lies to the west of Beaverdam Creek, which 
is the easternmost stream and drains into Indian Creek three-quarters of the length of 
Indian Creek. The origin of the Northeast Branch is at the confluence of Paint Branch 
and Indian Creek.  
 
 11 
 
 
I1
P3
B3
P1
B2
L1
P2
I4
I3
I2
B1
B4
I5
L2
P4
IU2
PU4
IU1
PU5
BU1
PU2
PU3
PU1
PM5
PM6
PM7
B
ri
er
 D
it
c
h
L
i
t
t
l
e
 
P
a
i
n
t
 
B
r
a
n
c
h
B
e
a
v
e
r
d
a
m
 
C
r
e
e
k
In
d
ia
n
 C
re
e
k
B
ec
k 
B
ra
n
c
h
N
o
r
th
e
a
s
t 
B
r
a
n
c
h
 A
n
a
c
o
s
tia
 R
iv
e
r
P
a
i
n
t
 
B
r
a
n
ch
I
n
d
i
a
n
 
C
r
e
e
k
P
a
in
t 
B
ra
n
c
h
P
a
in
t 
B
r
a
n
c
h
P
a
i
n
t
 
B
r
a
n
c
h
P
a
i
n
t
 
B
r
a
n
c
h
I
n
d
i
a
n
 
C
r
e
e
k
L
i
t
t
l
e
 
P
a
i
n
t
 
B
r
a
n
c
h
I
n
d
i
a
n
 
C
r
e
e
k
L
i
t
t
l
e
 
P
a
i
n
t
 
B
r
a
n
c
h
L
i
t
t
l
e
 
P
a
i
n
t
 
B
r
a
n
c
h
P
a
in
t 
B
r
a
n
c
h
Beck Br
anch
I
n
d
i
a
n
 
C
r
e
e
k
Legend
Sample Sites
Sample Sites
Streams
Physiographic_Boundary
Subwatershed_Boundary
?
 
Fig. 2-1: Northeast Branch Sample Sites 
 
Not observable on a map scale are the stream valley features and character. This 
topography reflects the weathering of the different geology in each physiographic 
region. The Piedmont is characterized by deeply incised valleys with floodplains only 
six m wide in many areas. While the soils are more clayey, the streams in the 
Piedmont are boulder and gravel bedded because the clay has washed out of the 
streams. The terrain is steep and hilly. In contrast, the Coastal Plain terrain is low 
 12 
 
 
rolling hills with wide floodplains and streams are gravel, sand, silt or clay bottomed. 
The soils in the Coastal Plain are generally sandier than those in the Piedmont.  
 
Soil series mapped in the watershed reflect the geologic parent material from which 
the soil formed. This parent material provides spatial discrimination across the 
watershed. The taxonomic designation indicates the dominant particle size in that 
soil. The soils in the Piedmont have more clay while those in the Coastal Plain have 
more sand. The stream beds in the Piedmont have gravel and bedrock remaining and 
the Coastal Plain has finer particle sizes remaining. In general the reservoirs of fine 
grain sizes are in the stream beds in the Coastal Plain or upland areas in the Piedmont; 
reservoirs of sand-sized particles are in the upland areas of the Coastal Plain.  
 
The mineralogy is also designated in the taxonomic name. The soil series include 
those in Table 2-1, as well as the map units Mixed Alluvial Land in the Piedmont, 
Gravel and Borrow Pits in the Coastal Plain and Urban, Sandy and Clayey Land, and 
Silty and Clayey Land in the downstream mixed portion of the watershed (Soil 
Survey Staff, 1967). The new series names reflect the mesic temperature regime in 
the Mid-Atlantic region as opposed to the thermic temperature regime further south. 
The updated soil survey will re-map these units with the new series names and also as 
soils formed only in sedimentary or crystalline parent material as appropriate. 
Potobac, Issue or Widewater (formerly Iuka), Zekiah (formerly Bibb), and 
Longmarsh (formerly Johnstown) are formed in non-crystalline sediment. 
 
Land use designations of urban, rural or forested are drawn from the National Land 
Cover Dataset of 2001. The land use in the area that drains to the Paint Branch and 
Little Paint Branch is urban, a combination of commercial and residential. The land 
use in the area that drains to the Northeast Branch is urban. It was primarily 
developed in the 1960-70s, during the height of the Cs-137 and Pb deposition time 
frame. Overall urban areas in the watershed are 62%. There are numerous wooded 
parks along the stream. The Indian Creek area is similar with the addition of gravel 
pits located near the stream?s headwaters. Exhausted gravel pits currently are used for 
yard waste dumps. Atmospherically deposited Cs-137 and Pb would be diluted by 
these dumps.  
 
The land use in the area that drains to Beaverdam Creek is classified as rural; it is 
farmland owned by the USDA-ARS. Some residential areas are on the perimeter of 
this area. A small national forest is also included in this area. The forested portion 
would be expected to have the highest Cs-137 concentrations.  
 
Atmospheric particulate matter distribution on the land may have been affected by 
Hurricane Agnes, a Category 1 hurricane, which reshaped much of the fluvial system 
in 1972. After the storm, the streams became wider as the sediment carved out the 
stream banks (Prepas and Charette, 2004). This single event resulted in a change in 
stream character still evident today. In addition, storm water and sewer pipes run 
alongside the streams. When the pipes were installed and subsequently patched, the  
 13 
 
 
Table 2-1: Soil series in the Northeast Branch watershed 
Soil Series Physiographic 
Region 
Taxonomic Designation Revised Soil 
Series 
Taxonomic 
Designation 
Codorus Piedmont and 
Coastal Plain 
Fine-loamy, mixed, active, mesic 
Fluvaquentic Dystrudepts 
 
Hatboro Piedmont and
Coastal Plain 
Fine-loamy, mixed, active, 
nonacid, mesic Fluvaquentic 
Endoaquepts 
 
Comus Piedmont Coarse-loamy, mixed, active, 
mesic Fluventic Dystrudepts 
  
Bibb Piedmont and
Coastal Plain 
Coarse-loamy, siliceous, active, 
acid, thermic Typic Fluvaquents 
Zekiah Coarse-loamy, 
siliceous, active, acid, 
mesic Typic 
Fluvaquents 
Iuka Piedmont and
Coastal Plain 
Coarse-loamy, siliceous, active, 
acid, thermic Aquic Udifluvents 
Issue or 
Widewater 
Coarse-loamy, 
mixed, active, mesic 
Fluvaquentic 
Dystrudepts or Fine-
loamy, mixed, active, 
acid, mesic 
Fluvaquentic 
Endoaquepts 
Sassafras Piedmont and 
Coastal Plain 
Fine-loamy, siliceous, 
semiactive, mesic Typic 
Hapludults 
  
Sunnyside Piedmont Fine-loamy, siliceous, 
semiactive, mesic Typic 
Hapludults 
  
Keyport Piedmont Fine, mixed, semiactive, mesic 
Aquic Hapludults 
 
Chillum Coastal Plain Fine-silty, mixed, semiactive, 
mesic Typic Hapludults 
Johnstown Coastal Plain Fine-silty, mixed, superactive, 
mesic Pachic Argiustolls 
Longmarsh Coarse-loamy, 
siliceous, active, acid, 
mesic Fluvaquentic 
Humaquepts 
Matapeake Coastal Plain Fine-silty, mixed, semiactive, 
mesic Typic Hapludults 
  
Ochlockonee Coastal Plain Coarse-loamy, siliceous, active, 
acid, thermic Typic Udifluvents 
 
ground was backfilled, presumably with local soil. Combined sewer overflows occur 
during heavy rainfall which may provide soil additions and change soil 
concentrations of tracers. 
Sampling Methodology 
Sites were selected to represent land use and physiography. Sample sites (16) were 
located at points along streams where the stream width or depth or topography 
changed (Fig. 2-1). To inform the selection of the sampling sites, digitized 
physiographic region boundary, geology, stream maps, and soils layers were overlaid 
using ArcGIS software. A detailed topographical map created from a 2-m DEM aided 
in establishing a soil-landscape model of the hydrologically active zones.  
 14 
 
 
 
Samples of stream banks were collected by first cleaning off any plastered material 
from the bank?s surface with a stainless steel implement. Soil was collected from the 
banks by scraping no deeper than five cm and vertically aggregating the sample. With 
few exceptions, the banks were of similar texture and color vertically. Banks were 
defined as those areas immediately adjacent to the stream that contain average stream 
flow.  
Lab Analyses 
Sample preparation 
Samples were dried on paper plates in an isolated area to avoid contamination from 
the grinding of other samples. Sticks, leaves, roots, and garbage collected with the 
sample were manually removed. Samples were ground and dry-sieved to <63 ?m. 
This size fraction is the geological break between sand and silt and includes some 
very fine sands (63-50 ?m) in the USDA particle-size classification system. Samples 
were sieved to allow for comparison with a comparable size fraction for a companion 
study.  
 
Organic matter was not removed prior to particle-size analysis or radionuclide 
analysis as organic matter differences can be difficult to generalize (Walling 2005; 
Carter et al., 2003; Collins et al., 1997a). The sample preparation process for the 
elemental analysis by ICP included dissolution by HNO
3
. This dissolution oxidized 
the organic matter and left the metals as residuals. This yielded the total elemental 
load for this size class, but did not discriminate between the metals sorbed to organic 
versus inorganic sediment.  
 
Particle-size analysis 
Particles of <63 ?m were analyzed with a Horiba L250 laser particle size analyzer at 
ARS-USDA. This method was used because it allows any particle-size range to be 
defined, allowing maximum flexibility. The analysis was performed in duplicate and 
the mean is reported in Appendix F. A laser particle-size analyzer measures the ratio 
of velocity in a vacuum to the velocity of light in the various particles. A refractive 
index of the velocity of light in clay is used.  
 
The Cs-137 distribution in the watershed may only be understood when also 
accounting for the particle-size variation in the watershed. It is assumed Cs-137 sorbs 
immediately to the negatively charged soil and is not exchanged because it is 
monovalent, has a small hydrated radius, and a high ionic potential. Cs-137 adsorbs to 
and within the smaller clay particles which have greater surface area due to their sheet 
arrangement and interlayers (Ritchie et al., 1974). Thus, sand does not have sorption 
sites for Cs-137. Sieving a sandy sample and removing the sand-sized fraction would 
concentrate the particles highest in Cs-137 since the Cs-137 analysis is performed on 
a concentration per mass basis (Bq kg
-1
 soil). Nuclear activities are measured in 
Becquerels, the SI unit of radio-activity. However, as soil texture classes vary 
throughout this watershed, elimination of the >63 ?m misrepresented the bulk soil. 
 
 15 
 
 
The >63 ?m portion of 16 bank samples was weighed to correct for biases related to 
variable particle size in the watershed. Correction to the measured values of Cs-137 
were made by first determining the percent sand (>63 ?m) by weight. This was done 
by collecting a separate sample taken from the same location as the original, 
weighing, sieving, and then weighing the sand fraction. The percent <63 ?m 
multiplied by the Cs-137 Bq kg
-1
 was calculated to correct for particle size 
differences. 
 
Table 2-2: Concentration of Cs-137 corrected for particle size  
Site 
Cs-137 (Bq kg
-1
 
fine soil) 
Measured 
concentrations 
% >0.63 ?m by weight 
Cs_137 (Bq kg
-1
 bulk soil) 
Calculated concentration 
B2 
0.00 0.72 0.001
B1 
12.12 0.82 2.130
B3 
2.48 0.88 0.310
B4 
8.65 0.62 3.290
I2 
0.00 0.60 0.001
I3 
0.00 0.75 0.001
I4 
3.52 0.95 0.180
I5 
4.53 0.82 0.820
L1 
13.88 0.66 4.770
L2 
0.00 0.74 0.001
PM5 
4.28 0.90 0.450
PM6 
1.99 0.81 0.380
PM7 
0.00 0.89 0.001
P1 
0.00 0.87 0.001
P2 
0.00 0.88 0.001
Pdump 
9.51 0.90 0.930
 
The percent of silt and clay to the bulk soil is important in regard to Pb as well. Lead 
hydrolyzes and some portion of Pb may have been present on oxyhydroxide sand 
grains coatings. For the purpose of this study, I measured only the Pb in the size 
fraction that includes silt and clay (<63 ?m). The data were originally collected for 
comparison with fine sediment that remains suspended in the water column. For this 
portion of the study, the ideal data collection would have measured the Pb on the bulk 
soil. Since those data were not collected, and statistical correction is not possible 
given that the amount on sand grain coatings is difficult to estimate, only 
concentrations in the silt and clay fraction are reported.  
 
Elemental Analysis by ICP-MS 
Analysis of Pb was performed by Acme Analytical Laboratories Ltd. Pb was 
analyzed by ICP-mass spectrometry following a leach with 3 ml 2-2-2 HCl-HNO
3
-
H
2
O (aqua regia) at 95?C for 1 hour, diluted to 10 ml. A sample-prep blank was 
carried through all stages of preparation and analysis as the first sample. Two reagent 
blanks measured background concentrations. Standard reference materials prepared 
by Acme using internationally certified reference materials were used to monitor 
 16 
 
 
accuracy. Maximum acceptable error for precision and accuracy was ?5% at 
concentrations >100X the detection limit. The limit of detection is 0.01 mg kg
-1
. A 
duplicate sample returned the same results within the error for precision and 
accuracy. While some Pb is naturally occurring, the geogenic concentration was 
assumed to be low and evenly deposited. 
 
Radionuclide analysis 
Radionuclide analysis was performed to determine the concentration of Cs-137. This 
analysis was performed by gamma-ray spectrometry using a Canberra-2000 Genie-
2000 Spectroscopy System for 24 hours at the U.S. Department of Agriculture?s 
Agricultural Research Service Hydrology Lab. The system was calibrated and 
efficiency determined using an analytic mixed radionuclide standard, the calibrations 
are traced to the U.S. National Institute of Standards and Technology. The limit of 
detection is 1 Bq kg
-1
 for Cs-137. 
 
Clay mineralogy 
Samples were analyzed to determine the type of clay. Unlike the other analyses 
performed in this study, x-ray diffraction is qualitative. X-ray diffraction was 
performed on 19 samples at the University of Maryland?s Pedology and X-Ray 
Diffraction Labs to determine differences in clay mineralogy throughout the 
watershed. Samples were centrifuged to separate the <2 ?m fraction. Samples were 
then separated into two parts and centrifuge-washed with K or Mg. The clay-oriented 
mounts were prepared using the filter-peel technique described in USGS File 01-041. 
A K and Mg mount were prepared for each sample. The K-saturated sample was 
analyzed at 25?C and subsequent heating to 300?C and 550?C. The Mg samples were 
placed in a desiccator over ethylene glycol and analyzed at 25?C. The scan was 2-30 
? .  
 
Soil Survey 
A soil survey was conducted to provide a pedological context for the study area and 
to inform an intuitive model that helps explain the chemical data. Auger borings were 
made by transects covering a catena of bank, floodplain, and upland areas. 
Morphological descriptions included: horizonation, texture, color, and redoximorphic 
features. Land cover and parent material were verified in the field. Thus, the landform 
and soil relationships were identified (Appendix A).  
Results 
Differences in the distribution of Pb and Cs-137 concentrations record the time period 
of deposition. As the soil eroded and moved as sediment from one part of the 
watershed to another, the mixing with other soils from other time periods can be 
estimated.  
Pb and Cs-137 
Within error and the limit of detection, only three sites were differentiated in Cs-137 
concentration, B1, B4, and L1. The concentrations of Pb and Cs-137 did not covary, 
 17 
 
 
indicating different depositional time periods or different depositional mechanisms 
such as direct versus atmospheric (Fig2-2).  
 
1
10
100
1000
-5.00 0.00 5.00 10.00 15.00
Cs-137 (Bq kg-1)
P
b
 (mg
 k
g
-1
)
 
Fig. 2-2, Pb and Cs-137 concentrations. 
Error for Pb is the size of the marker.  
 
The mean concentration of Cs-137 was 0.82 Bq kg
-1
 with a coefficient of variation 
(C.V.) of 175.41 Bq kg
-1
. The mean concentration of Pb was 43.41 mg kg
-1
 and the 
C.V. was 103.46 mg kg
-1
. Concentrations of Cs-137 and Pb varied by site (Table 2-3). 
A site-by-site description follows and is organized by stream from west to east.  
 
Table 2-3: Tracer variation  
Site Subwatershed 
Boundary 
Stream Distance 
above 
gauge (m) 
Cs-137 
(Bq kg
-1
 
bulk soil) 
Pb (mg kg
-1
 
fine 
soil)(detection 
limit=0.01 mg 
kg
-1)
B2 
East Beaverdam
14,855
0.00 14.40
B1 
East Beaverdam
12,173
2.13 24.10
B3 
East Beaverdam
10,168
0.31 18.50
B4 
East Beaverdam
9,468
3.29 27.20
I2 
East Indian Creek
11500
0.00 15.00
I3 
East Indian Creek
11499
0.00 5.80
I4 
East Indian Creek
10205
0.18 119.30
I5 
East Indian Creek
4744
0.82 188.20
L1 
West Little Paint Branch
17125
4.77 44.50
L2 
West Little Paint Branch
10494
0.00 46.50
PM5 
Mixed Northeast Branch
2458
0.45 63.60
PM6 
Mixed Northeast Branch
2286
0.38 39.60
PM7 
Mixed Northeast Branch
0
0.00 61.00
P1 
West Paint Branch
16385
0.00 16.70
P2 
West Paint Branch
9464
0.00 37.00
Pdump 
East Paint Branch
4744
0.93 773.20
 
Paint Branch 
The Paint Branch is located in the Piedmont, where stream beds are gravel and 
boulder bottomed and the narrow floodplains provide little opportunity for sediment 
 18 
 
 
storage. The sites moving downstream are P1, P2, and Pdump. Paint Branch has little 
variation in Pb or Cs-137. The Cs-137 concentrations in the banks of the first order 
streams of the Paint Branch were less than the limit of detection of 1 Bq kg
-1
 and 
measured 0 Bq kg
-1
. At sites P1 and P2, the bank is vertical and no opportunity 
existed for Cs-137 to have been atmospherically deposited on the vertical surface of 
these banks. The channels are stable and there has been little observable erosion. Site 
P1 is a county park and the downstream site P2 was developed pre-1950. The Pb 
concentration at the downstream site P2 is approximately 60% greater than at site P1. 
The drainage area to site P2 is also approximately 60% greater than site P1, indicating 
that Pb was transported and stored in the stream banks.  
 
Site Pdump is the furthest downstream sample taken from the Paint Branch. The Cs-
137 concentration was 0.93 Bq kg
-1
 compared with 0 Bq kg
-1
 at the upstream sites. At 
this site the Pb concentration was 773.20 mg kg
-1
?an order of magnitude higher than 
elsewhere. This site was significantly incised. It had ~2 meters of fill material over 
Cretaceous-aged sediments. The fill material was partially the result of the site being 
used as a garbage dump located directly alongside the stream. Historical photos and 
materials found at the site, such as a pipe with a manufacture date of 1859, confirm 
this. The opposite side of the bank, where there was no evidence of a garbage dump, 
measured 14.9 mg kg
-1
of Pb and 0 Bq kg
-1
 of Cs-137?levels commensurate with 
other samples taken from that stream (see Chapter 4 for data from additional 
samples). 
 
Little Paint Branch 
The Little Paint Branch headwaters are in the Piedmont and downstream is in the 
Coastal Plain, all of which is in the west subwatershed. The sites on the Little Paint 
Branch are L1 and L2, which had variable concentrations of Cs-137 in the banks. 
Upstream site L1 had 4.77 Bq kg
-1
 Cs-137 compared to 0 Bq kg
-1
 at downstream site 
L2. At the upstream sample site L1, the stream is incising and widening. Site L1 is 
located in a residential area that was developed around 1995. This site is immediately 
downstream from a bridge where the stream is piped underneath a road. The presence 
of a Bw horizon indicates a fairly young soil. It is possible that the entire surface and 
subsurface drainage network was changed as the residential area and an adjacent golf 
course were developed. Prior to development, the area was farmed. The stream may 
naturally migrate and may now be running through soils eroded from the hillslope. 
Construction practices moving subsoil to the surface combined with erosion via soil 
creep downslope may have diluted soil from the Cs-137 depositional period.  
 
The fine sediment to which the Cs-137 is sorbed appears to have been transported 
further downstream as indicated by the decrease in Cs-137 stored in the banks at site 
L2. At the downstream site L2, there is a floodplain over a pre-1950 soil. There is a 
lithologic discontinuity in some reaches of the stream bank that indicates that the 
floodplain overlies an older soil which overlies an even older soil. The cobble line in 
the bank sediments reveals an active floodplain where the stream migrates.  
 
 19 
 
 
The Pb concentrations of 44.5 Bq kg
-1
 at site L1 and 46.5 Bq kg
-1
 at site L were the 
same within the margin of error indicating soil stability prior to 1954 when Pb was 
being used in gasoline but Cs-137 was not yet deposited.  
 
The Little Paint Branch appears to have soil older than that which has Pb and Cs-137 
deposited on it. The newer sediment deposits are of a sandier texture, which has little 
Cs-137 sorption capacity.  
 
Indian Creek 
Indian Creek is located in the Coastal Plain where the soils are sandier and the 
streambeds are finer textured. Valleys are wide and the topography is not as steep as 
the Piedmont. Sites on Indian Creek include I2, I3, I4, and I5, which is downstream 
from the Beaverdam confluence. On Indian Creek, Cs-137 and Pb concentrations 
systematically increased in a downstream progression. Cs-137 was found at 0 Bq kg
-1
 
at both headwater sites I3 and I2, both sites have vertical banks. These sites are close 
to the headwaters so there is not a lot of sediment being transported. Site I3 was 
Cretaceous-age stratigraphy. At site I2, the soil is older than the 1950s. While the 
trees in the riparian buffer are fairly young (<60 years), the soil did not have an Ap 
horizon, so it was not farmed. The stream is in a bottom-land and is flooded too often 
to be farmed. While the banks are not high, they are vertical. Cs-137 would not have 
been deposited on these banks. 
 
Pb concentrations measured 5.8 mg kg
-1
 at site I3 and 15.0 mg kg
-1
 at site I2. The 
concentrations vary according to the distance to paved road and parking lots. While 
Pb was atmospherically deposited, it is possible that some gasoline leaked directly 
into the soil near the sample site from farm machinery.  
 
At site I4, the channel banks are vertical. The stream is widening and may be 
intercepting a soil on which radionuclides were deposited during the 1954-1974 time 
period. The moderate Cs-137 concentration of 0.18 Bq kg
-1
 at site I4 reflects the 
channel widening. The Pb dramatically increased from its upstream measured 
concentration to 119.3 mg kg
-1
. Since this site is in an industrial zone with many 
automobile repair businesses, it is possible that direct leaks to the soils and stream 
occurred.  
 
At site I5, both Cs-137 and Pb increase to 0.82 Bq kg
-1
 and 188.2 mg kg
-1
 
respectively over concentrations measured upstream. This site is located near where 
there historically has been industry, though now some of the industrial sites have been 
replaced by residential areas. Active erosion and sedimentation is illustrated by a 
half-exposed tire in a horizontal position in the bank of site I5; the tire could only be 
moved and deposited by the stream. Indian Creek widened during the current 
hydrological regime. The hydrologic regime change is caused by greater storm water 
fluxes, which are a result of increasing impervious surfaces. Levels of Cs-137 and Pb 
generally increase moving downstream showing sediment storage in the Indian 
Creek.  
 
 20 
 
 
Beaverdam Creek 
Beaverdam is located in the Coastal Plain and the sites on this stream are B1, B2, B3, 
and B4. Beaverdam Creek varies in terms of sediment storage and erosion sites. The 
pattern of different erosive reaches of the stream and storage stream reaches is most 
obvious on Beaverdam Creek. At sites B2 and B3 erosion was evident and at sites B1 
and B4 deposition was evident. The Cs-137 and Pb concentrations were higher at the 
depositional sites than at the erosional sites.  
 
Site B1 is the furthest upstream sample on the main branch of this creek. The soil 
profile for site B1 indicates that this site had undergone some soil structure formation, 
which would not be evident if the soils were solely unconsolidated sediments. 
Therefore, the stream cut or broadened the channel into an older soil. The area around 
this stream is a floodplain, as shown by both the soil and elevation. The area is 
wooded and could not have been farmed recently, although there is a farm upslope. 
Thus, the Cs-137 concentration of 2.13 Bq kg
-1
 supports the hypothesis that the 
stream cut a new channel or broadened the channel since Cs-137 was deposited. The 
pedological evidence of soil structure formation is that there has been deposition, or 
sediment stored, in the stream banks and does not indicate that sediment was 
deposited from an upstream site. Pb was measured at 24.10 mg kg
-1
.  
 
Site B3 has a typical floodplain profile with buried A horizons interspersed with 
newly forming soils and older, more developed soils. As the stream has either held its 
channel, allowing soil development, or meandered and deposited sediment over older 
soils while cutting into those older soils, sediment has been translocated. The area is 
forested with trees > 80 years old. Upslope is farmed, but not the floodplain. Bars 
were observed moving downstream and active erosion is occurring. This meandering 
and sediment storage has resulted in mixing, which is evident in the 0.31 Bq kg
-1
 of 
Cs-137 found in the banks. The Pb concentration of 18.5 mg kg
-1
 is less than that 
found upstream indicating active erosion of the fine sediment that had been in 
storage. 
 
Downstream, Beaverdam Creek opens up into a marsh where there is evidence of 
rapid changes in the path of the stream. In the seven months following sampling, the 
bank sample site became a small island in the middle of the stream. At site B4, the 
Cs-137 concentration increases to 3.29 Bq kg
-1
 which is greater than any upstream 
measurement. Either the bank is new and was from an undisturbed area, or the 
sediments forming the bank were translocated from an area of undisturbed land. The 
site is an area of sediment storage, and either explanation is equally defendable. Pb 
concentrations followed the same trend and were measured at 27.2 mg kg
-1
. 
 
Site B2 is at the headwater of a small tributary leading into Beaverdam Creek. It is 
sandwiched between a small road and large parking lot in an apartment complex and 
a highly trafficked freeway (Baltimore-Washington Pkwy). This tributary?s 
headwaters are drainage pipes for storm runoff. The soil around the stream is highly 
variable and disturbed. The 0 Bq kg
-1
 Cs-137 concentration reflects a subsoil fill was 
used. Pb was measured at 14.4 mg kg
-1
. 
 
 21 
 
 
Northeast Branch 
Sample sites on the Northeast Branch progress downstream from site PM5 where 
Indian Creek and Paint Branch come together, to site PM6 which is a backwater 
channel, to site PM7 where the banks are hardened with large, cut granite rocks. The 
hardening of the banks is effective and prevents stream bank erosion or channel 
widening. However, at sites PM5 and PM6, the stream has been able to incise as well 
as widen. Cs-137 was 0.45 Bq kg
-1
 and Pb was 63.6 mg kg
-1 
at site PM5. At site PM6, 
the backwater channel, Cs-137 decreased slightly to 0.38 Bq kg
-1
 and Pb decreased to 
39.6 mg kg
-1
. At site PM7, there was 0 Bq kg
-1
 Cs-137 and Pb was measured at 61.0 
mg kg
-1
. The banks that are now exposed were formed from channel widening into 
what had been a floodplain?where Cs-137 would have accumulated during the 
period of nuclear testing. Cs-137 values at sites PM5 and PM7 indicate mixing from 
an incising channel (where there was a zero concentration of Cs-137 as older soil is 
exposed on vertical bank faces) and a higher measurement of Cs-137 (from the 
widening channel into older soils). Incision without additional widening at site PM7 
is supported by the zero concentration of Cs-137 at site PM7.  
Clay Mineralogy 
The results of the clay mineralogy analysis did not show significant, identifiable 
differences throughout the watershed. For example Site P2, located in the Piedmont 
does not contrast significantly with Site B1, located in the Coastal Plain (Fig. 2-3). 
The scans show vermiculite, mica, kaolinite, quartz and some gibbsite and goethite 
present in both physiographic regions. The vermiculite peak had shoulders in several 
of the scans, indicating hydroxy-interlayered vermiculite. The size of the peak is not 
indicative of the dominance of that mineral and can not be used to calculate relative 
differences. This analysis was discontinued after 19 samples as the results did not 
distinguish among soil samples.  
 
0.400.500.600.700.800.901.001.201.501.902.503.60
Vermiculite
Micas,  V ermiculi te
Kaolinite
Micas,Vermiculite
Gibbsite
Vermiculite
Quartz
Geothite
Kaolinite
Quartz, Micas
d  s p a c i n g   (n m)
c 
o
 
u
 
n t
 
s   
p
 
e 
r
 
  s
 
e
 
 c o n 
d
 
c 
o
 
u
 
n t
 
s   
p
 
e 
r
 
  s
 
e
 
 c o n 
d
 
 
 
 22 
 
 
0.400.500.600.700.800.901.001.201.501.902.503.60
Vermiculite
Micas,Vermiculite
Kaolinite
Kaolinite
Micas,Vermiculite
Gibbsite
Quartz
Kaolinite
Quartz
c 
o 
u n
 t
 s
  
 
p e
 r
   
s
 
e  
c 
o
 
n 
d 
d  s p a c i n g   (n m)
c 
o 
u n
 t
 s
  
 
p e
 r
   
s
 
e  
c 
o
 
n 
d 
c 
o 
u n
 t
 s
  
 
p e
 r
   
s
 
e  
c 
o
 
n 
d 
c 
o 
u n
 t
 s
  
 
p e
 r
   
s
 
e  
c 
o
 
n 
d 
 
Fig.: 2-3 Clay mineralogy, P2 bank (top) and B1 bank (bottom) 
 
Soil Survey 
A site-by-site soil description illustrates the differences among the soils and streams 
in this watershed (Appendix A). As land development occurred, the effect on the 
streams differed according to the topography and particle size in the streambanks and 
bed. Those streams that were boulder and gravel bottomed have limited fine 
sediment, and the banks were not built during small floods. These streams became 
more deeply incised with time except where the stream bottoms were bedrock; these 
bedrock-bottomed streams widened. In the Coastal Plain, there were wider 
floodplains and these developed where the land was not impervious from urban 
development. However, urbanization increased the size of floods because storm water 
was efficiently eliminated from the impervious surfaces and piped into the streams at 
high velocities. With fewer small floods due to the storm water management system, 
channel widths and depths have enlarged to accommodate the more frequent higher 
discharges (Konrad et al., 2005). Thus, streams are not going overbank and depositing 
sediment loads so floodplains in the Coastal Plain are not being built as they were in 
the past. The soil descriptions show that each site varies based on land use history and 
stream character.  
Discussion 
While there were distinct land use differences in the watershed, the patterns of Cs-137 
and Pb concentrations in the streambanks did not differentiate these differences. This 
is because there is a distinct reach of the streams that provides sediment storage in 
this watershed. This reach falls between the headwater portion, which is erosional, 
and the downstream portion, which has no place for storage because banks are 
hardened and floodplain access restricted. The distributions of Cs-137 in the 
Northeast Branch watershed shows an increasing concentrations of Cs-137 in the 
progression downstream until the region where the channels are hardened and 
sediment storage is no longer possible. The exceptions to this increase are the two 
wetland areas along Beaverdam Creek (B1 and B4). Wetland stream channels 
 23 
 
 
frequently carry less discharge. This causes the stream flow to go overbank and create 
and/or maintain the riparian wetland. Sediment is carried and deposited in the riparian 
wetland.  
 
Sediment storage occurred in the middle swath running east to west in the watershed. 
The banks in these reaches were built from either transported bank materials from 
eroding banks moved downstream or overland flow moving upland materials to the 
floodplains. The banks are carved from these floodplains.  
 
In the downstream area of the watershed, Cs-137 concentrations were never above the 
watershed mean for the Northeast Branch downstream portion of the watershed. 
The streams were channelized and hardened in these downstream areas; there was no 
storage and Cs-137 concentrations reflected this by decreasing back to zero, as they 
were in deeply incised and vertical bank stream reaches.  
 
Pb can be a point-source pollutant whereas Cs-137 is only atmospherically deposited; 
the two confound each other. Pb concentrations seem to be a better indicator of level 
of traffic and distance to stream than time period of development (Juracek and 
Ziegler, 2006). This indicates that Pb in this watershed was primarily directly 
deposited. If Pb were atmospherically deposited, it would track time period of 
development. To fully understand historical trends in Pb and Cs-137 deposition, 
samples should be taken at depth increments in the banks along each reach of the 
stream.  
 
Correcting for particle-size differences in the watershed was necessary because the 
amount of Cs-137 and Pb or any other tracer is commonly measured in units of mass. 
By biasing the mass through removal of large particle sizes, some soils will have 
values that are diluted or concentrated in comparison to the land from which they 
came. This is only an issue where an understanding of the land is important. If the 
measurements are used only for comparison to other fine sediment sources such as 
the suspended load, then the patterns across the landscape are not relevant. 
Statistically correcting particle size to reflect bulk soil for Cs-137 concentrations 
assumes that Cs-137 does not adsorb to sand particles. I made this assumption 
because Cs-137 is monovalent and has a small hydrated radius. In contrast, Pb is 
known to sorb to sand particles as Pb hydrolyzes in the oxyhydroxide coatings. I was 
unable to make a statistical correction for Pb because of the lab procedure.  
 
Erosion was episodic and therefore difficult to track with depth integrated samples. 
Episodes were caused by land use and management changes as well as by single 
exceptional storm events like the hurricane (Nicholas et al. 1995).  
 
Atmospherically deposited Cs-137 shows that erosion and sedimentation varies by 
stream reach in this watershed. The directly deposited Pb was less useful to 
understanding the historical erosion and sedimentation in this urban watershed.  
 24 
 
 
Chapter 3: Geogenic and Anthropogenic Sources of Fine 
Sediment in an Urban Watershed 
Introduction 
Suspended sediment is a water quality problem because it blocks sunlight from 
permeating water, suppresses subaquatic vegetation growth, smothers the benthic 
biota that live among the plants, and carries contaminants (EPA, 2006c). When 
transported, sediment moves contaminants into and through waterways. Particles <63 
?m are the most important sediment fraction since it is these fine particles that stay in 
suspension and carry the bulk of the contaminants (Vanoni, 1975; Marvin et al., 2004; 
Tuccillo, 2006). Sources of fine suspended sediment may be present where there is 
little visible evidence of erosion. Also, the amount of fine sediment from sandy soils 
may be small, even if erosion is high. Sand-sized particles quickly settle and therefore 
do not carry toxins long distances or impact light attenuation in downstream water 
bodies. Fine sediment sources have been successfully quantified using a sediment 
fingerprinting approach (Collins et al., 1997a; Walling et al., 1999; Motha et al., 
2003; Walling, 2005).  
 
The sediment fingerprinting approach compares suspended sediment to sources, and 
determines the relative quantity from each source. One of the difficulties in 
quantifying sources of sediment is determining parameters for classifying the sources 
into groups. Previous researchers have parameterized sediment sources by land use 
(e.g., forested, agricultural, or industrial), geological formations, soil map units, or a 
combination of these classifications (Deboer and Crosby, 1995; Collins et al., 1997a; 
Walling et al., 1999; Motha et al., 2003; Walling, 2005; Gellis and Landwehr, 2006). 
 
Urban watersheds present unique challenges. Chemical or biological waste from 
human activities may sorb to the sediment that moves through the watershed 
(Boruvka et al., 2005). The spatial pattern of this anthropogenic imprint may vary 
among urban areas. In addition, the change in land use to urban does not necessarily 
happen simultaneously throughout the watershed. A non-linear rate of urbanization 
combined with a complicated spatial pattern of urbanization adds complexity to using 
land use as a sediment source parameter when a variable urban imprint is present. A 
method is needed to differentiate fine sediment in urban watersheds. 
 
I hypothesized that chemical analyses of sediment could be used to determine end-
member compositions of sediment in urban watersheds. Furthermore, I hypothesized 
that end-member compositions used in the sediment fingerprinting of urban 
watersheds included anthropogenically enriched elements. Also, I provide a method 
for differentiating fine sediment by the level of enrichment of trace elements 
compared to the mean composition of the Earth?s upper continental crust.  
 
Radionuclide and element concentrations are used in this study as sediment tracers. 
Elemental concentrations may originate from the bedrock, or from atmospheric 
 25 
 
 
inputs, and other anthropogenic activity. In this urban watershed, sediment sources 
include water-transported material from streets and the deposition of atmospheric 
particulate matter, in addition to eroded soil from fields and stream banks. 
Atmospheric particulate matter falls on soil surfaces and pavement with little seasonal 
variation (Almeida et al., 2005). Automobiles also produce particulates from tires and 
exhaust. All of these particulates accumulate and wash into streets and down storm 
drains that discharge to streams. The tracers used in previous research are generally 
fewer in number than those for which I tested (Collins et al., 1997a; Walling et al., 
1999; Motha et al., 2003; Walling, 2005; Gellis and Landwehr, 2006). I considered a 
broader number of trace elements because an urban watershed may have different 
tracers present than agricultural and forested watersheds, where sediment 
fingerprinting is more commonly used.  
Study Site and Methods 
Study Site 
The study area was the watershed of the Northeast Branch of the Anacostia River in 
Maryland located approximately 16 km northeast of Washington, D.C. (Fig. 3-1). 
Four main tributaries make up the Northeast Branch watershed. From west to east 
they are: Paint Branch, Little Paint Branch, Indian Creek, and Beaverdam Creek. The 
Little Paint Branch drains to the Paint Branch. Beaverdam Creek drains to Indian 
Creek. The confluence of Paint Branch and Indian Creek form the Northeast Branch 
of the Anacostia. The Anacostia River drains to the Chesapeake Bay. The drainage 
area of the Northeast Branch watershed is 188.5 km
2
. 
 
The Northeast Branch watershed spans two physiographic regions ? the Piedmont and 
the Coastal Plain. The different geological composition of the two physiographic 
regions results in distinguishable elemental concentrations that may be used as natural 
tracers. The physiographic province reflects the geology and topography. Coastal 
Plain geology primarily comprises silt-clay facies of the Potomac Group (Patuxent 
Formation), which is Early Cretaceous in age. The Potomac Group is fluvially 
deposited quartz, with some sandstone and chert. The Coastal Plain also includes 
Arundel Clay and much of the lower Patapsco Formation. The Coastal Plain in this 
area is highly weathered Piedmont material from fluvial deposition. Upland deposits 
of the Coastal Plain are fluvial sediments of the ancestral Potomac River and are of 
the late Miocene and Pliocene periods. Terrace deposits are the product of stream 
erosion during the early Quaternary Period. These are primarily unconsolidated 
sediments. The Coastal Plain ranges in age from the Cretaceous to Quaternary 
periods.  
 
 26 
 
 
I1
P3
B3
P1
B2
L1
P2
I4
I3
I2
B1
B4
I5
L2
P4
IU2
PU4
IU1
PU5
BU1
PU2
PU3
PU1
PM5
PM6
PM7
B
ri
e
r 
D
it
c
h
L
i
t
t
l
e
 
P
a
i
n
t
 
B
r
a
n
c
h
B
e
a
v
e
r
d
a
m
 
C
r
e
e
k
In
d
ia
n
 C
re
e
k
B
e
c
k
 B
ra
n
c
h
N
o
r
th
e
a
s
t 
B
r
a
n
c
h
 A
n
a
c
o
s
ti
a
 R
iv
e
r
P
a
i
n
t
 
B
r
a
n
c
h
I
n
d
i
a
n
 
C
r
e
e
k
P
a
in
t 
B
r
a
n
c
h
P
a
in
t 
B
r
a
n
c
h
P
a
i
n
t
 
B
r
a
n
c
h
P
a
i
n
t
 
B
r
a
n
c
h
I
n
d
i
a
n
 
C
r
e
e
k
L
i
t
t
l
e
 
P
a
i
n
t
 
B
r
a
n
c
h
I
n
d
i
a
n
 
C
r
e
e
k
L
i
t
t
l
e
 
Pa
i
n
t
 
B
r
a
n
c
h
L
i
t
t
l
e
 
P
a
i
n
t
 
B
r
a
n
c
h
P
a
i
n
t 
B
r
a
n
c
h
Beck Br
anch
I
n
d
i
a
n
 
C
r
e
e
k
Legend
Sample Sites
Sample Sites
Streams
Physiographic_Boundary
Subwatershed_Boundary
?
 
Fig. 3-1 Northeast Branch Watershed sample sites.  
 
Mapped units in the Piedmont, including the Laurel formation, are characterized by 
Precambrian crystalline metamorphic rocks with occasional intrusions of igneous 
rocks. The granite, gneiss, schist, and other crystalline rocks range in age from 
Precambrian to late Paleozoic. Piedmont upland deposits are high in vermiculite clay 
(Glaser, 1996; Csato et al., 2006). 
 27 
 
 
In addition to encompassing two physiographic regions, the Northeast Branch 
watershed is made up of two subwatersheds. The west subwatershed is influenced by 
Piedmont geology and the east subwatershed is characterized by Coastal Plain 
geology (Fig. 3-1). The subwatershed boundary is mapped east of the physiographic 
boundary and is located between Little Paint Branch and Indian Creek.  
 
In the west subwatershed the land use is urban?a combination of commercial and 
residential. There are numerous wooded parks along the stream. In the east 
subwatershed land use is industrial, commercial, residential, forested, and 
agricultural. Land use around Indian Creek includes industrial, commercial, and 
residential areas. The land use in the area that drains to Beaverdam Creek is rural and 
has long been managed as farmland by USDA-ARS. Some residential areas are on the 
perimeter of the land that drains to Beaverdam Creek. A forested national park is also 
located in the east subwatershed.  
 
Presently the Anacostia River watershed, which includes the Northeast Branch 
watershed, is 64.6% urban, 26.6% forested, 8.1% agricultural, 0.2% wetlands, and 
0.4% other categories (Maryland Department of Natural Resources, 2000). 
Impervious land is 33.2% (Maryland Department of Natural Resources, 2000). The 
Northeast Branch area of the Anacostia was developed primarily between 1960 and 
1970. While minor construction projects are currently taking place, there are no 
current wide-spread Earth-moving projects.  
Sample Locations 
Sample location were located along reaches of each of the four tributaries and the 
main stem to represent each subwatershed, landscape position, and land use of the 
Northeast Branch watershed (Fig 3-1, Appendix B).  
 
The high percentage of impervious surfaces necessitated more soil samples from 
banks because much of the upland area was paved. Of the bank samples, there were 
three sample sites along the Paint Branch, two along the Little Paint Branch, four 
along Indian Creek, five along Beaverdam, and three along the Northeast Branch. 
Bank sample sites were chosen where stream width or depth changed, topography 
changed, there was a confluence, or significant erosion was visible. 
 
These geomorphic changes in the stream areas indicated a possible change in soils 
and sediments. At five of these stream bank sites, samples were taken along a transect 
moving away from the bank to include the landscape positions of stream bank, 
floodplain, and upland. Depending on the topography, the upland position was either 
a backslope or a summit position. These five sites were selected to represent 
physiographic region and land use. 
 
Additional sediment inputs in this urban area include street residue, sewage, and 
atmospheric particulate matter. Street residue was collected for both physiographic 
regions in the urbanized portions of the watershed. One construction site located 
adjacent to the stream and very near the suspended sediment collection site was 
 28 
 
 
sampled as were four point bars. It was outside the scope of this project to collect 
sewage or atmospheric particulate matter. However, atmospheric particulate matter 
data was available from other watersheds for comparison. Suspended sediment 
samples were collected at the outlet of the Northeast Branch watershed (Fig. 3-1, 
PM7). The distribution of the sediment samples is presented in Appendix B. 
Sampling Methodology 
Sediment source samples 
A total of 35 sediment source samples were collected. The stream banks were defined 
as those areas immediately adjacent to a stream that contain the stream at average 
flow. Floodplains were defined as those areas that are broad flatlands on either side of 
a stream that accommodate maximum flood capacity. All other areas were classified 
as upland. 
 
For floodplain and upland areas, surface samples were collected 2-5 cm deep with a 
stainless steel implement. Bank samples were collected by first scraping away any 
plastered material, and then aggregating sediment vertically throughout the bank. 
With few exceptions, each bank profile was of similar texture and color vertically. 
Full descriptions of stream banks in this watershed are in Appendix A.  
 
Street residue was collected by sweeping the street 3.0 m on either side of a storm 
drain located on a street adjacent to a stream. Streets were swept with a clean plastic-
bristled broom from the center of the street to the curb on both sides of the street 
(Breault et al., 2005). Street residue samples were collected by aggregating residue 
from six different streets. To represent seasonal variation, samples were collected just 
after leaf fall in fall 2005 and in late spring 2006. The intervening winter was 
atypically dry with little snowfall, so salt was not broadly applied to streets. 
Accordingly, a winter sample was not collected.  
 
Suspended sediment samples 
Samples were collected when rainfall covered the entire watershed and when the 
sediment load was most concentrated. Samples of sediment were collected for 
comparison to the sources to determine the sources of suspended sediment and the 
characteristics of each source.  
 
The Northeast Branch watershed is located at 38? and 39? N, 76? and 77? W. The 
climate is humid temperate with an average temperature of 13.2 ?C and an average 
annual rainfall of 117 cm (NOAA, 2005). There are four distinct seasons with mild 
winters. Rainfall is evenly distributed throughout the year; storms with broad aerial 
extent occur primarily in winter months. There are isolated thunderstorms in the 
summer that deliver rain quickly and create flashy stream flow. These summer 
thunderstorms typically do not span the entire watershed.  
 
Suspended sediment collection was targeted for high sediment generating events. 
Suspended sediment is most concentrated during the initial runoff period of a storm 
 29 
 
 
when precipitating conditions have been wet (Appendix E). Sampling was targeted 
for saturated overland flow conditions. 
 
Weather forecasts were monitored via real-time forecasts (University of Maryland 
Department of Meteorology and NOAA). In addition, a rain gauge eight km from the 
site was monitored at least three times per day. When rainfall was close to 2.5 cm, the 
Northeast Branch was visited and assessed for sediment movement into the stream by 
visual assessment of the Tyndall effect in the stream water, gauge height, and amount 
of overland flow. Samples were not collected when unsafe conditions existed due to 
darkness or excessive ice on the roads.  
 
Sampling occurred at one location and depth under a bridge where the stream was 
well mixed. A trash pump was placed into the stream where the current held it above 
the stream bed but below the surface of the water. Particles <63 ?m are well mixed in 
the water column, so it was not necessary to collect at a particular depth (Vanoni, 
1975). The stream water was pumped into three 208-liter steel barrels lined with 
round-bottomed plastic bags.  
 
Eight suspended sediment samples were collected for one year (July 1, 2005 to June 
30, 2006) to account for seasonal differences in erosion and run off. Collection dates 
and flood information are in Table 3-1. The fine sediments are on average 25% of the 
total load carried in the stream. The larger particles do not travel far downstream, but 
settle out of the water column. 
Lab Analyses 
Sample Preparation 
Sediment source samples were brought back to the laboratory and air dried on paper 
plates in an isolated lab to minimize cross contamination. Visible leaves, twigs, and 
garbage in the samples were removed manually. Samples were then ground using a 
mortar and pestle. In order to avoid concentrating or diluting source areas based on 
particle size, the samples were dry sieved to < 63 ?m using a shaker, which restricted 
analysis to the dominant size class in the suspended sample.  
 
Suspended sediment samples were prepared by removing the stream water from the 
barrels within one week following collection. A Teflon-lined continuous centrifuge 
was used for sediment removal. The centrifuge was selected because the Teflon lining 
would prevent metal contamination from the machine itself. Suspended sediment 
samples were dried in an oven at 60? C. The suspended sediment samples were also 
ground and dry sieved to <63 ?m using a shaker. 
 
Organic matter was not removed from any of the samples prior to radionuclide 
analysis. Organic matter differences are complex and difficult to generalize, therefore 
even where significant organic matter is present, corrections are frequently not made 
(Walling, 2005). The sample preparation process for the elemental analysis by ICP 
includes dissolution by HNO
3
. This dissolution oxidizes the organic matter and leaves 
the metals as residuals. This yields the total elemental load for this size class, but does 
 30 
 
 
not discriminate among the metals sorbed to organic versus inorganic sediment (Sun 
et al., 2001; Impellitteri et al., 2002). In all of the source samples, total carbon was < 
9.9%. 
 
For the particle-size analysis, organic matter was removed by H
2
O
2
 only from the 
street residue samples. Without organic matter removal a great many particles were 
not spherical and blocked laser detection of the smaller particles, leading to an 
overestimate of particle size. These non-spherical particles are thought to be leaves 
and grass clipping that were crushed or ground by automobile traffic.  
 
Table 3-1: Suspended sediment sample characteristics. **at last available time before 
gauge failed.  
Date and 
time of 
suspended 
sediment 
collection 
Q at time 
of 
suspended 
sediment 
collection 
(m
3
 s
-1
) 
Date and 
time of 
hydrograph 
peak 
Q at 
hydrograph 
peak 
(m
3
 s
-1
) 
Sediment 
collected 
(g m
-3
) 
<0.63 ?m 
Sediment 
(g m
-3
) 
%<0.63 ?m 
sediment/bul
k sediment 
collected 
Sediment 
flux at 
time 
collected 
(g s
-1
, 
<0.63 
?m) 
10/7/2005 
at 12:30 
15.51 10/8, at 
7:00
55.87 23.05 8.21 36% 127.37
10/22/2005 
at 9:00 
0.83 10/22 at 
9:30
2.16 250.72 53.76 21% 44.67
12/25/2005
at 15:00 
0.33 12/25 at 
18.15
3.32 105.46 31.57 30% 10.52
1/18/2006 
at 9:00 
4.00 1/18 at
11:45
10.55 175.12 37.56 21% 150.15
2/4/2006 at 
16:00 
0.80 2/4 at 21:00 6.61 120.7 31.99 27% 25.58
4/23/2006 
at 6:05 
1.46 4/22 at
21:00
9.09 623.46 199.33 32% 290.63
5/11/2006 
at 20:30 
0.39 5/11 at
21:00
4.24 114.2 15.914 14% 6.24
6/25/2006 
at 11:45 
14.39 6/25 at
20:30 **
173.72** 293.09 62.23 21% 31613.37
 
Radionuclide analysis 
Radionuclide analysis was performed on sediment source samples and suspended 
sediment samples by gamma-ray spectrometry using a Canberra-2000 Genie-2000 
Spectroscopy System for 24 hours at the U.S. Department of Agriculture?s 
Agricultural Research Service Hydrology Lab. By running the sample for the long 
duration, lower levels may be quantified. The limit of detection is 1.00 Bq kg
-1
. The 
system was calibrated and efficiency determined using an analytic mixed radionuclide 
standard; the calibrations are traced to the U.S. National Institute of Standards and 
Technology.  
 
 31 
 
 
Elemental analysis by ICP-MS and ICP-ES 
The elemental analysis was performed by Acme Analytical Laboratories Ltd. ICP-
emission spectrometry was performed following a LiBO
2
/Li
2
B
4
O
7 
fusion at 980?C 
and dissolution in 0.5% HNO
3
. The lithium borate fusion lowers the boiling point 
allowing complete dissolution of the silica. Rare Earth and refractory elements were 
tested by ICP-mass spectrometry following the same fusion and dissolution process. 
Total carbon and sulfur were determined by LECO. Loss on ignition (LOI) was 
determined by weight difference after ignition at 1000?C. The elements and method 
for each are in Table 3-2. 
 
The elemental analyses were performed on all the sediment source samples as well as 
the suspended sediment samples. A sample-prep blank was carried through all stages 
of preparation and analysis as the first sample. Two reagent blanks measured 
background. Standard reference materials prepared by Acme using internationally 
certified reference materials were used to monitor accuracy. Maximum acceptable 
error for precision and accuracy was ?5% at concentrations >100X the detection 
limit. The limits of detection are in Appendix D. A duplicate sample returned the 
same results within the error for precision and accuracy. 
 
Table 3-2: Element analytical methods 
Method Chemical Tracer
LiBO
2
/Li
2
B
4
O
7
 fusion analysis by ICP-
ES. 
SiO
2
, Al
2
O
3
, Fe
2
O
3
, MgO, CaO, Na
2
O, 
K
2
O, TiO
2
, P
2
O
5
, MnO, Cr
2
O
3
, Ni, and 
Sc 
LECO Total C, Total S 
Rare Earth and Refractory Elements: 
LiBO
2
/Li
2
B
4
O
7
 fusion analysis by ICP-
MS 
Ba, Be, Co, Cs, Ga, Hf, Nb, Rb, Sn, Sr, 
Ta, Th, U, V, W, Zr, Y, La, Ce, Pr, Nd, 
Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb, Lu 
Leach with 3 ml 2-2-2 HCl-HNO
3
-H
2
O 
(aqua regia) at 95?C for 1 hour, diluted to 
10 ml, analyzed by ICP-MS. 
Mo, Cu, Pb, Zn, Ni, As, Cd, Sb, Bi, Ag, 
Au, Hg, Tl, Se 
 
Statistical Analyses 
Tracer concentrations of sediment source samples were examined for normality using 
SAS v. 9.1 (SAS Institute Inc., 2004). Analysis of probability plots and box plots 
indicated non-normally distributed data, and not all elements were skewed in the 
same direction. Normality was violated, so non-parametric tests were used. In 
addition, there were orders of magnitude among the tracers, so ranking the data 
allowed for comparison.  
 
The source samples were grouped to minimize variation within source groups while 
maximizing variation among source groups. I examined a Spearman correlation 
matrix for various groupings of elemental and radionuclide concentrations. The non-
parametric Kruskal-Wallis means comparison test was used to identify the tracers 
with significant differences among source groups. Tracers that failed this test 
(p<0.05) were removed.  
 32 
 
 
 
The sediment sources were grouped by subwatershed and landscape position. These 
groups were further parameterized on the level of enrichment in various elements 
compared to the mean elemental composition of Earth?s upper continental crust.  
Results 
Sediment Source Characterization 
The tracers did not clearly discriminate sediment sources when grouping by land use, 
upstream vs. downstream sections of the streams, or even by each stream. Results of 
the physiographic region sediment source grouping were similar to the subwatershed 
grouping. However, the subwatershed grouping was more powerful than the 
physiographic region. Significant differences were present between the west 
subwatershed and east subwatershed for the tracers in Table 3-3. The west 
subwatershed was enriched in all of these elements except for As, Cr
2
O
3
, SiO
2
, and 
Zn. The west subwatershed is influenced by the Piedmont physiographic region; the 
enrichment is predominantly from elements native to the metamorphic rocks in that 
region. In contrast, the east subwatershed includes the more highly weathered Coastal 
Plain physiography that is depleted in these elements. The soils in the east 
subwatershed are sandier, which is reflected in the elevated SiO
2
 concentration. 
 
The tracers that were significantly different when grouped by landscape position are 
listed in Table 3-4. The banks were enriched in Dy, Eu, Gd, La, Lu, Tb, and Tm. The 
floodplains were enriched only in Cs-137, while the point bars were enriched in Er, 
Hr, Ho, K-40, Sn, TiO
2
, Y, Yb, and Zr. Street residue was enriched in Cu and Mo. 
Upland samples were enriched only in P
2
O
5
.  
 
More tracers were elevated in the bank and point bar landscape positions than any 
other position (7 and 9 respectively compared to < 2 in any other group). The banks 
and point bars are the landscape positions where soil is most likely to move into 
stream flow. When examined on a pair-wise basis, the banks and point bars cannot be 
discriminated from each other. The point bars were not considered in further analysis. 
Moreover, point bars represent temporary storage of sediment from other sources.  
 
Table 3-3: Tracers that differentiated by hydrological subwatershed boundary (p < 
0.05). Bold indicates higher concentration.  
Tracer 
 
Units Kruskal-
Wallis 
West mean 
(no.=13) 
East mean 
(no=19) 
Al
2
O
3
wt. % 0.000 15.02 9.70
As mg kg
-1
0.037 3.02 4.52
Ba ? 0.000 508.88 310.69
Be ? 0.003 2.92 1.63
Bi ? 0.000 0.36 0.18
CaO wt. % 0.026 0.35 0.33
Ce mg kg
-1
0.000 154.77 91.71
 33 
 
 
Tracer 
 
Units Kruskal-
Wallis 
West mean 
(no.=13) 
East mean 
(no=19) 
Cr
2
O
3
wt. % 0.011 0.01 0.01
Cs mg kg
-1
0.020 4.00 2.87
Cu ? 0.008 35.85 31.32
Dy ? 0.002 11.82 7.66
Er ? 0.006 7.05 4.75
Eu ? 0.001 2.47 1.42
Fe
2
O
3
wt. % 0.002 5.15 3.64
Ga mg kg
-1
0.000 19.65 12.22
Gd ? 0.005 11.52 7.03
Hf ? 0.034 27.39 18.75
Ho ? 0.003 2.40 1.60
K
2
O wt. % 0.000 2.14 1.28
La mg kg
-1
0.002 64.05 40.58
Lu ? 0.002 1.12 0.73
MgO wt. % 0.002 0.72 0.46
MnO wt. % 0.048 0.10 0.07
Na
2
O wt. % 0.015 0.44 0.29
Nb mg kg
-1
0.000 25.64 19.74
Nd ? 0.000 62.77 37.95
Ni ? 0.010 20.37 17.54
Pr ? 0.000 16.57 10.14
Rb ? 0.002 91.49 62.03
Se ? 0.023 0.63 0.61
SiO
2
wt. % 0.001 62.58 72.40
Sm mg kg
-1
0.001 13.01 7.67
Sn ? 0.003 4.08 3.16
Ta ? 0.000 2.10 1.49
Tb ? 0.001 2.01 1.26
Th ? 0.002 16.18 11.43
TiO
2
wt. % 0.001 1.57 1.19
Tl mg kg
-1
0.000 0.24 0.12
Tm ? 0.007 1.10 0.74
U ? 0.000 5.87 3.96
V ? 0.001 105.42 78.53
Y ? 0.002 69.74 45.61
Yb ? 0.005 6.82 4.67
Zn ? 0.002 77.31 90.68
Zr ? 0.019 982.98 686.22
 
 34 
 
 
Table 3-4: Tracers that differentiate the landscape positions (p < 0.05). Bold indicates 
highest mean concentration.  
Tracer 
(mg kg
-1
) 
Units Kruskal-
Wallis 
Bank 
mean 
(N=18) 
Floodplain 
Mean 
(N=5) 
Point Bar 
Mean 
(N=4) 
Streets 
Mean 
(N=2) 
Upland 
Mean 
(N=14) 
Cu mg kg
-1
0.017 39.29 29.28 52.78 88.25 30.57
Dy mg kg
-1
0.027 10.51 8.28 10.34 6.58 7.90
Er mg kg
-1
0.012 6.42 4.94 6.56 3.82 4.79
Eu mg kg
-1
0.025 2.15 1.55 1.80 1.04 1.54
Gd mg kg
-1
  0.044 10.18 7.71 8.95 5.96 7.38
Hf mg kg
-1
0.002 28.13 17.48 38.48 19.60 16.24
Ho mg kg
-1
0.005 2.17 1.65 2.29 1.31 1.62
K-40  Bq kg
-1
0.001 540.62 960.42 1708.16 496.79 991.01
La mg kg
-1
0.019 56.57 44.12 46.43 31.30 44.80
Lu mg kg
-1
0.039 1.02 0.73 1.08 0.65 0.77
Mo mg kg
-1
0.031 1.47 0.86 1.48 2.25 0.81
P
2
O
5
wt. % 0.012 0.12 0.16 0.15 0.29 0.20
Cs-137  Bq kg
-1
0.001 3.32 27.55 0.86 9.51 18.46
Sn mg kg
-1
0.014 3.17 4.00 6.00 5.50 3.64
Tb mg kg
-1
0.033 1.76 1.36 1.73 1.02 1.34
TiO
2
wt. % 0.024 1.43 1.15 1.84 1.38 1.16
Tm mg kg
-1
0.013 1.02 0.77 1.00 0.61 0.73
Y mg kg
-1
0.023 61.54 49.00 67.10 41.00 47.66
Yb mg kg
-1
0.007 6.35 4.85 6.56 3.54 4.65
Zr mg kg
-1
0.003 983.26 650.92 1491.20 802.60 613.89
Suspended Sediment Characterization 
The suspended sediment had higher concentrations than the means of the sediment 
sources by both subwatershed and landscape position in MnO, Cu, Ni, Zn, Cd, Co, 
Au and Mo (Fig. 3-2). Many of these non-bracketed tracers were of anthropogenic 
origin; and the sediment sources were primarily characterized by those tracers that are 
geogenic in origin, not anthropogenic. A missing source of sediment was indicated 
since the sediment sources did not bracket the suspended sediment. When including 
as a source a known garbage dump site in a floodplain Zn, Cu, and Au became 
bracketed, but MnO, Cd, Co, Mo, and Ni remained more concentrated in the 
suspended sediment than in the sediment sources. Historical photos and materials 
found at the garbage dump site, such as a pipe with a manufacture date of 1859, 
indicate when the site was active. 
 
Cu, Mo, and P
2
O
5 
were more concentrated in the street residues than in the other 
landscape positions. The Cu and Mo also were more concentrated in the suspended 
sediment, and the P
2
O
5
 concentration in the street residues approached that found in 
the suspended sediment. Streets isolate atmospheric particulate matter and channel 
this flow directly into streams disproportionately to their surface area. Upland soils 
are a mix of geogenic and anthropogenic elements while streets and other impervious 
surfaces are direct depositional environments of atmospheric particulate matter. This 
 35 
 
 
suggests that atmospheric particulate matter is a missing sediment source end 
member.  
 
0.00
0.05
0.10
0.15
0.20
0.25
0.30
S
us
p
. 
S
e
d
.
S
tr
ee
ts
B
a
n
k
s
F
lo
o
d
p
la
in
s
P
o
in
t 
B
a
rs
U
p
la
n
d
E
ar
th
M
n
O (wt.
 %
)
 
0
20
40
60
80
100
120
140
S
u
sp
. 
S
e
d
.
S
tr
e
e
ts
B
a
n
ks
F
lo
o
d
p
la
in
s
P
o
in
t 
B
a
rs
U
p
la
n
d
E
a
rt
h
C
u
 (
m
g
/kg
)
 
 
0.00
10.00
20.00
30.00
40.00
50.00
60.00
70.00
80.00
90.00
100.00
S
us
p.
 S
ed
.
S
tr
ee
ts
B
an
ks
F
lo
o
dp
la
in
s
P
oi
n
t B
a
rs
U
pl
a
nd
E
ar
th
Ni (
m
g
/kg
)
 
0
50
100
150
200
250
300
350
400
450
S
us
p
. S
ed
.
S
tr
ee
ts
B
a
nk
s
F
lo
od
p
la
in
s
P
oi
nt
 B
ar
s
U
p
la
nd
E
ar
th
Zn (
mg/
k
g
)
 
 
0
10
20
30
40
50
60
S
u
sp
. S
e
d.
S
tr
ee
ts
B
an
ks
F
lo
o
dp
la
in
s
P
oi
n
t B
ar
s
U
pl
an
d
E
ar
th
C
o
 (mg/
k
g
)
 
0
2
4
6
8
10
12
S
u
sp
. S
e
d.
S
tr
ee
ts
B
an
ks
F
lo
o
dp
la
in
s
P
oi
n
t B
ar
s
U
pl
an
d
E
ar
th
C
d
 (mg/
k
g
)
 
0
1
2
3
4
5
6
7
8
9
10
S
u
sp
. S
e
d.
S
tr
ee
ts
B
an
ks
F
lo
o
dp
la
in
s
P
oi
n
t B
ar
s
U
pl
an
d
E
ar
th
M
o
 (mg/
k
g
)
 
0
5
10
15
20
25
S
u
sp
. S
e
d.
S
tr
ee
ts
B
an
ks
F
lo
o
dp
la
in
s
P
oi
n
t B
ar
s
U
pl
an
d
E
ar
th
Au (
?
g/
k
g
)
 
Fig. 3-2: Tracers not bracketed by the sediment sources. The mean concentration in 
the Earth?s upper continental crust is shown for reference (Rudnick and Gao, 2003). 
The x-axis is landscape position. Measurement error is represented by the size of the 
point.  
 
The measured concentration of Cu in suspended sediment was similar to Cu 
concentrations in atmospheric particulate matter collected from two sites in New 
York (17.00 and 7.30 ng m
-3
/yr), one in New Jersey (4.70 ng m
-3
/yr), several samples 
from a forested site in Maryland (650 ug/m
2
/yr), and also data on global emissions to 
 36 
 
 
the atmosphere (28,000 metric tons/yr) (Gao et al., 2002; Callender 2004; Scudlark et 
al., 2005; Almeida et al., 2005; Duan et al., 2006). These studies did not test for Mo 
or P
2
O
5
. 
 
Combined sewer overflows could be an additional source of contamination to the 
suspended sediment in the stream (Rauter et al., 2005; Tuccillo, 2006). Rainwater 
enters the sewage pipes from manhole covers adding to the volume of material in the 
pipes. The high pressure in the sewer pipes during storm events could force sewage 
out of holes and cracks which otherwise would not leak. Washington Suburban 
Sanitary Commission reported values of 61 mg kg
-1
 dry for Cu and 5.5 mg kg
-1
 dry 
for Mo in lime cake taken from their Parkway processing plant in 2004 (Washington 
Suburban Sanitary Commission, 2005). This Cu concentration does not approach that 
in the street residue or suspended sediment samples. This Mo concentration exceeds 
that found in either street residue or suspended sediment samples. Since Washington 
Suburban Sanitary Commission measured limed samples, the concentrations are not 
directly comparable as the sorption of these elements change under different pH 
levels. Neither sewage nor atmospheric particulate matter was directly sampled as 
part of this project. However, Washington Suburban Sanitary Commission data 
indicated that the source of these elements could not be entirely from sewage.  
Comparison to the Earth?s Mean Upper Continental Crust Composition 
The trace element concentrations in suspended sediment samples were not bracketed 
in the sources and were primarily anthropogenic. Therefore the compositions of 
sediment source groups may not be characterized by anthropogenic trace elements as 
I originally hypothesized. Tracers that best differentiated sediment sources were 
geogenic, not anthropogenic. Since the suspended sediment was not bracketed by the 
sources, a missing source of sediment is suggested?possibly atmospheric particulate 
matter. While atmospheric particulate matter has an even spatial distribution, the 
evidence of these particulates is uneven because soils dilute the particulates. Traffic 
and roads have an uneven spatial distribution that coincides with the undiluted 
atmospheric particulate matter. Sources may be further parameterized by 
distinguishing elements that originated from atmospheric particulate matter and/or 
traffic intensity and road density. This can be done by quantifying the level of trace 
element enrichment compared to the Earth?s mean upper continental crustal 
composition. If the missing source is of anthropogenic origin, then the tracers for that 
source will be enriched. 
 
To quantify enrichment of elements compared to the Earth?s mean upper continental 
crustal composition, source samples were compared to the mean composition of the 
Earth?s upper continental crust (Rudnick and Gao, 2003) and normalized to Titanium. 
TiO
2
 had the least variability among the source samples (TiO
2
 range = 1.75 wt. %), 
has chemical stability, and has minor contributions of anthropogenic pollution (Duan 
et al.; 2006). As such TiO
2
 is commonly used for normalizations. The following ratio 
was used:  
 
 37 
 
 
crust
sediment
,
?
?
?
?
?
?
?
?
?
?
?
?
=
y
x
y
x
FactorEnrichment
xcrust
 (3-1) 
 
Where:  
x = concentration of an element  
y = concentration of TiO
2
 
The source of x is considered anthropogenic where the enrichment factor is greater 
than 1. To select the most extreme values, only enrichment values > 2 were 
considered significant. This is an arbitrary threshold and was based on an 
examination of the data to ensure selectivity of those that are clearly enriched.  
An example of how this would work is illustrated below. 
 
In a sample, Cd=0.2 mg kg
-1 
and TiO
2
=1.24 mg kg
-1
. In the Earth?s mean upper 
continental crustal composition, Cd=0.09 mg kg
-1 
and TiO
2
=0.64 mg kg
-1
. The ratio 
of Cd/ TiO
2
in the sample = 0.16 and the ratio of Cd/ TiO
2
in the Earth?s upper crust is 
0.14. To determine the level of enrichment, divide the sample by the Earth?s upper 
crust, 0.16/0.14 to get 1.14. Since 1.14 is greater than unity, then Cd is enriched in 
this sample compared to the Earth?s mean upper continental crustal composition. 
 
Of the geogenically derived trace elements, Zr, Pr, Er, and Yb were enriched in the 
banks located in the west subwatershed. Nd and Hf were enriched in the downstream, 
mixed portion of the stream banks. No geogenic elements were enriched in the east 
subwatershed, although Zr and Hf approached the threshold enrichment factor of two. 
Fig. 3-3 shows elements enriched when comparing subwatersheds plus Nb, Cu, Zn, 
Sb, Au, Hg, and Pb, which were enriched only in the landscape position comparison.  
 
Of the anthropogenically derived trace elements, Ag and Se were enriched in the east 
subwatershed. Cd was enriched in the downstream mixed area. This portion of the 
watershed is highly channelized with cement-lined channels and rock gabions. The 
primary source of water to the streams in this portion is through storm water drains. 
Cd, a particulate produced by automobiles (Callender, 2004), is efficiently delivered 
through these drains as is evidenced by concentrations in the suspended sediment. 
 
 38 
 
 
0.1
1
10
Ti
O
2
Zr
Nb
Pr
Nd
Er
Y
b
Hf
Cu
Z
n
S
e
Ag Cd S
b
A
u
Hg P
b
E
n
r
i
chm
e
nt
 Fact
or
West Banks East Banks Mixed Banks
Geogenic Anthropogenic
 
Fig. 3-3 Enrichment of subwatersheds by a factor >2, normalized to Ti.  
 
Fig. 3-4 shows elements enriched when comparing landscape position. The elements 
Pr and Nd, which were enriched only when comparing subwatersheds, were included 
for comparison purposes. Of the geogenically derived trace elements, Zr, Er, and Yb 
were enriched in the stream banks. Nb and Hf are enriched in the point bar samples.  
 
Of the anthropogenically derived trace elements, Cu, Zn, Ag, Au, and Hg were 
enriched in the suspended sediment samples. Zn is commonly found in very local 
sources, such as from rusting bridges (M. Goldhaber, personal communication, 2006). 
The elevated Zn may also reflect a high density of roads since Zn is a component of 
automobile tires. The Au may be enriched due to the three coal-burning power plants 
still operating in the region. Coal mined in West Virginia is used in these power 
plants and has a relatively high concentration of Au compared to other coal sources 
(M. Goldhaber, personal communication, 2006).  
 
In the floodplain, Se is slightly more enriched than other landscape positions. In the 
street residue samples Cd, Sb, and Pb are slightly more enriched. The west 
subwatershed has a higher road density compared to the east subwatershed, which 
may explain why Pb was the only enriched anthropogenic element in the west. Pb is a 
product of leaded gasoline widely used until 1978. 
 
 39 
 
 
0.1
1
10
100
Ti
O
2
Zr
N
b
Pr
N
d
E
r
Yb
Hf
Cu
Z
n
Se Ag Cd Sb Au H
g
Pb
E
n
r
i
ch
m
e
n
t
 F
act
o
r
Bank Flood Plain Point Bar Susp.Sed. Streets Upland
Geogenic Anthropogenic
 
Fig. 3-4 Enrichment of landscape positions by a factor >2, normalized to Ti.  
Discussion 
Using an iterative methodology, I tested the sediment fingerprinting method and 
found geogenic spatial characterizations of sediment sources the best way to group 
the sediment sources. The use of chemical tracers successfully discriminated 
sediment sources in this urban watershed. . Anthropogenic tracers did not serve as 
tracers in this study because the suspended sediment samples showed the highest 
values of these elements; thus they were not bracketed by the sediment source groups. 
The only two anthropogenic tracers found significant in the landscape position source 
groupings were Cu and Mo?both elevated in the street residue. The street residue 
also was enriched in Sb and Pb compared to the mean composition of the Earth?s 
upper continental crust. These elements may not sorb as tightly to particulates at some 
of the pH values found in this region as do many of the other measured anthropogenic 
tracers (Impellitteri et al., 2002). Alternatively, they may have been sorbed onto 
organic matter, which was carried higher in the flow than the suspended sediment 
sampled in the stream.  
 
Sediment source groups most clearly were defined when element concentrations in 
the sediment sources were compared to the Earth?s upper continental crust. This 
method discriminated between an overprint on sediment that confounded spatial 
distribution differences. The non-bracketed suspended sediment and the few 
significant anthropogenic tracers in the sediment source groups were neither 
consistent nor easily explained, likely due to a vector of traffic intensity and road 
density combined with the dilution of atmospheric particulate matter by the soils.  
 
The elements that were not enriched compared to the Earth?s  upper continental crust 
were primarily of geogenic origin. These included SiO
2
, Al
2
O
3
, Fe
2
O
3
, MgO, CaO, 
 40 
 
 
Na
2
O, K
2
O, P
2
O
5
, MnO, Sc, Ba, Be, Co, Cs, Ga, Rb, Sn, Sr, Ta, Th, U, V, W, Y, La, 
Ce, Sm, Eu, Gd, Tb, Dy, Ho, Tm, Lu, Mo, Ni, As, Bi, and Tl. These trace elements 
are primarily geogenically derived materials. P
2
O
5
, Ni, As, and Sn being the most 
notable exceptions. P
2
O
5
 is used in fertilizer and may wash off highly fertilized urban 
yards. Yet, the anthropogenic overprint comprises most of the statistically significant 
differences between the suspended sediment and sediment sources. 
 
Trace elements that had high concentrations in the suspended sediment were elements 
commonly associated with anthropogenic origins. The elements with the highest 
concentrations were Zn, Ba, Zr, and K-40. While the amount of Zr and Ba was high 
compared to other tracers in the suspended sediment, the concentration was diluted 
compared to the sediment sources. In contrast, Zn was more highly concentrated in 
the suspended sediment. Of these four tracers, Zn and Ba are primarily of 
anthropogenic derivation. K-40 is a natural isotope of K which may come either from 
the highly micaceous material in the west subwatershed or from fertilizer applied to 
agricultural crops and urban lawns. Regardless of mass, the tracers of highest 
concentration in the suspended sediment indicated an anthropogenic imprint.  
 
There were several other elements in the suspended sediment that were not bracketed 
by the sediment sources but were not enriched compared to the Earth?s mean upper 
continental crustal composition. These included MnO, Ni, Mo, and Co. MnO, Mo, 
and Co are known to exchange during canopy fall through and in soils (Brady and 
Weil., 2002; Skudlark et al., 2005).  
 
Atmospheric particulate matter is delivered to impermeable surfaces and washed into 
the streets and down storm drains. The Washington Sewer and Sanitary Commission 
also reports a high concentration of Mo (Washington Suburban Sanitary Commission, 
2005). Of all of the sources measured, the street residues most closely approached the 
concentrations found in the suspended sediment. This indicates that the sediment 
most contaminated by metals is washing off streets and is not coming from point bars, 
stream banks, or upland areas. Thus, the pathway that delivers the most toxic 
sediment is one that is only spatially differentiated by measures of impervious surface 
area and the atmospheric particulate matter that falls on it along with variations in 
traffic intensity. While the highest mass of eroded material may be coming from 
banks, floodplains and other soil surfaces, this material is not the most contaminated. 
Rather the soils dilute the contaminants derived from atmospheric particulate matter 
and automobile traffic. 
 
The method of calculating enrichment of tracers and comparing to the mean 
composition of the Earth?s upper continental crust provides a concise summary of 
complex environmental factors. It also helps to explain the origin of the tracers, 
which provides another tool in teasing apart sources of sediment polluting our 
waterways. 
 41 
 
 
Chapter 4: Quantifying Fine Sediment Sources in the Northeast 
Branch of the Anacostia River 
Introduction 
Fine sediment negatively impacts light attenuation in water and diminishes subaquatic 
vegetation and the benthic biota that depend on vegetation. The EPA has instituted a 
Total Maximum Daily Load requirement for fine sediment going into the Chesapeake 
Bay (EPA, 2006c). Determining the sources of fine sediment is relevant to remedying 
the problem. 
 
The sediment fingerprinting method has been used by researchers extensively 
(Collins et al., 1997a; Walling and Woodward, 1992; de Boer and Crosby, 1995; 
Wallbrink et al., 1998; Walling et al., 1999; Rowan et al., 2000; Owens et al., 2000; 
Jenkins et al., 2002; Carter et al., 2003; Gruszowski et al., 2003; Motha et al., 2003; 
Papinicolaou et al., 2003; Walling, 2005; Gellis and Landwehr, 2006). This method 
provides a relative quantification of sediment from various source groups by 
comparing naturally occurring tracers in sediment sources to suspended sediment. 
These tracers may be measures of mineral magnetic susceptibility (Slattery et al., 
1995), concentrations of radionuclides (Walling and Woodward, 1992; He and 
Owens, 1995), concentrations of stable isotopes (C-13 and N-15) (Papanicolaou et al., 
2003; Gellis and Landwehr, 2006), or concentrations of trace elements (Wallbrink et 
al., 1998; Motha et al., 2002; Walling, 2005). Depletion or concentration in the 
elements discriminates between sediment sources. 
 
The sediment fingerprinting approach has primarily been applied in rural settings 
with land use as the primary classification of source groups and radionuclides used to 
discriminate sources. Cs-137 was atmospherically deposited 1954-1975 with a 
significant increase during the year of 1964 as a result of nuclear weapon tests. 
Surface concentrations of Cs-137 reflect land use, except where mixing or dilution of 
soils occurred. The highest concentrations of Cs-137 are found in those areas that are 
undisturbed, such as forested areas, or where soils were translocated from undisturbed 
areas and not diluted (Nagle and Ritchie 2004, Matisoff et al., 2002; Wallbrink et al. 
1998).  
 
In many urban watersheds, urbanization occurs at a variable rate (Jacobson and 
Coleman, 1986) and does not necessarily occur on continuous plots of land, but rather 
in small, scattered areas. Therefore, it may be difficult to use land use as a 
differentiating characteristic of sediment sources in urban areas.  
 
The purpose of this study was to test the sediment fingerprinting approach in an urban 
watershed. Sources of fine sediment were characterized to determine their sources. 
Specifically, the hypothesis was: The primary source of suspended sediment in this 
watershed is derived from stream bank erosion. Knowing the components of 
suspended sediment samples informed the selection of characteristics for a composite 
 42 
 
 
tracer of source areas. Thus, an iterative process whereby significant characteristics in 
the suspended sediments were determined at the same time as the measurements in 
the sediment sources.  
Assumptions 
Several assumptions are inherent in sediment fingerprinting research. Suspended 
sediment samples were only collected during storm events with full aerial extent. If it 
were to only rain on one portion of the watershed, than that one portion would 
necessarily be represented in the suspended sediment. It must rain over the entire 
watershed equally in order to attribute suspended sediment collected downstream to 
various upstream sources. Gauges were not located in every part of the watershed; 
however, larger storms that have a broad aerial extent were sampled. 
 
The timing of the sample collection has an impact on the sources identified. Travel 
times vary according to many factors such as channel roughness and distance from 
the suspended sediment collection site. If roughness is equal throughout the 
watershed, then this is not a factor. In terms of travel time, those sources closest to the 
collection site will be over-represented in samples collected early in the hydrograph. 
Likewise, headwater sources will be over-represented in samples collected in the 
falling limb of the hydrograph. The sediment sources attributed to the suspended 
sediment may not be generalized to the entire storm event, but only to those at the 
particular point in the hydrograph at which the suspended sample was collected. It is 
also assumed that all the tracers travel at the same rate. That is, K sorbed onto a 
particle travels at the same rate as a Si particle. This assumption was not explored as 
part of this study. 
 
I also assume that the tracers used in this study do not change or participate in any 
exchange reactions. I assume tracers in the solid phase were not retained or 
complexed in the soil differently throughout the watershed. I also assume each tracer 
is equally distributed among the sizes <63 ?m, which is the size that remains 
suspended in the water column (Vanoni, 1975). 
Study Site and Methods 
Watershed Description 
The study site was the Northeast Branch watershed, an 188 km
2
 watershed located in 
an urban area of Maryland near Washington D.C. The watershed spans two 
physiographic regions?the Piedmont to the west and the Coastal Plain to the East 
(Fig. 4-1). The Northeast Branch watershed includes two subwatersheds; the western 
subwatershed is approximately a third the total watershed area and is influenced by 
Piedmont physiography. The eastern subwatershed is influenced by Coastal Plain 
physiography. The different geological composition of the two physiographic regions 
means elemental concentrations are significantly different enough that they may be 
used as natural tracers. 
 43 
 
 
 
I1
P3
B3
P1
B2
L1
P2
I4
I3
I2
B1
B4
I5
L2
P4
IU2
PU4
IU1
PU5
BU1
PU2
PU3
PU1
PM5
PM6
PM7
B
ri
e
r 
D
it
c
h
L
i
t
t
l
e
 
P
a
i
n
t
 
B
r
a
n
c
h
B
e
a
v
e
r
d
a
m
 
C
r
e
e
k
In
d
ia
n
 C
r
e
ek
B
e
c
k
 B
ra
n
c
h
N
o
r
th
e
a
s
t
 B
r
a
n
c
h
 A
n
a
c
o
s
t
i
a
 R
i
v
e
r
P
a
i
n
t
 
B
r
a
n
c
h
I
n
d
i
a
n
 
C
r
e
e
k
P
a
in
t 
B
r
a
n
c
h
Pa
i
n
t 
Br
a
n
c
h
P
a
i
n
t
 
B
r
a
n
c
h
P
a
i
n
t
 
B
r
a
n
c
h
I
n
d
i
a
n
 
C
r
e
e
k
L
i
t
t
l
e
 
P
a
i
n
t
 
B
r
a
n
c
h
I
n
d
i
a
n
 
C
r
e
e
k
L
i
t
t
l
e
 
P
a
i
n
t
 
B
r
a
n
c
h
L
i
t
t
l
e
 
P
a
i
n
t
 
B
r
a
n
c
h
P
a
i
n
t
 
B
r
a
n
c
h
Beck Br
anch
I
n
d
i
a
n
 
C
r
e
e
k
Legend
Sample Sites
Sample Sites
Streams
Physiographic_Boundary
Subwatershed_Boundary
?
 
Fig. 4-1 Northeast Branch Watershed sample sites. 
 
 44 
 
 
There are four streams that drain to the Northeast Branch. From west to east they are: 
Paint Branch, Little Paint Branch, Indian Creek, and Beaverdam Creek. The 
confluence of the Paint Branch and Indian Creek forms the Northeast Branch, which 
drains to the Anacostia River. The Anacostia River in turn drains to the Chesapeake 
Bay. 
 
The Coastal Plain streams in this study include the entirety of Indian Creek and 
Beaverdam Creek and the lower portions of Paint Branch and Little Paint Branch. 
The streams are silt-and clay-bottomed. Sand bars migrate through the two streams 
that originate in the Piedmont. The Coastal Plain is a low rolling plain. Its geology in 
this watershed is silt-clay facies of the Potomac Group, comprised of Arundel Clay 
and much of the lower Patapsco Formation. The Coastal Plain in this area is highly 
weathered Piedmont material from fluvial deposition. Terrace deposits are the 
product of stream erosion during the early Quaternary Period. The Coastal Plain 
sediments range in age from the Cretaceous to Quaternary periods.  
 
Coastal Plain geology primarily comprises silt-clay facies of the Potomac Group 
(Patuxent Formation), which is Early Cretaceous in age. The Potomac Group is 
fluvially deposited quartz, with some sandstone and chert. The Coastal Plain also 
includes Arundel Clay and much of the lower Patapsco Formation. The Coastal Plain 
in this area is highly-weathered Piedmont material from fluvial deposition. Upland 
deposits of the Coastal Plain are fluvial sediments of the ancestral Potomac River and 
are of the late Miocene and Pliocene periods. Terrace deposits are the product of 
stream erosion during the early Quaternary Period. These are primarily 
unconsolidated sediments. The Coastal Plain ranges in age from the Cretaceous to 
Quaternary periods.  
 
The Piedmont streams include the upstream reaches of the Paint Branch and Little 
Paint Branch and are gravel- and boulder-bedded. Some of the boulders range to 
several meters in diameter. The floodplains range from three and six meters in width. 
The region is hilly and slopes are steep and have high elevations. The downstream 
reaches of these streams are characterized by fewer boulders and more sand.  
 
Mapped units in the Piedmont, including the Laurel formation, are characterized by 
Precambrian crystalline metamorphic rocks with occasional intrusions of igneous 
rocks. The granite, gneiss, schist, and other crystalline rocks range in age from 
Precambrian to late Paleozoic. Piedmont upland deposits are high in vermiculite clay 
(Glaser, 1996; Csato et al., 2006). 
 
Presently, impervious land in the Anacostia River watershed, which includes the 
Northeast Branch watershed, is 33% (Maryland Department of Natural Resources, 
2000). The Northeast Branch area of the Anacostia is the oldest developed portion of 
the watershed, developed primarily from between 1960 and 1970. While minor 
construction projects are currently taking place, there are no current wide-spread 
Earth-moving projects.  
 
 45 
 
 
Street residue provides an additional sediment input. Field observation indicates that 
drainage in decreasing order of magnitude is from storm drains piping runoff from 
impermeable surfaces directly into the streams, overland flow, and groundwater 
upwelling. The downstream portions of the streams are channelized with cement, 
gabions, or granite blocks. The channelization inhibits sediment storage and increases 
stream flow velocity by decreasing friction. 
 
The Northeast Branch watershed is located between 38? and 39? N and 76? and 77? 
W. The climate is humid continental with an average temperature of 13.2 ?C and an 
average annual rainfall of 117 cm (NOAA 2005). There are four distinct seasons with 
mild winters. Rainfall is evenly distributed throughout the year; storms with broad 
aerial extent occur primarily in winter months. There are isolated thunderstorms in 
the summer that deliver rain quickly and create flashy stream flow. These summer 
thunderstorms typically do not span the entire watershed.  
Sampling Strategy 
To determine the sources of suspended sediment leaving the Northeast Branch 
watershed, sites for potential sediment sources were selected at locations throughout 
the watershed (Fig. 4-1). These sites were located along reaches of each of the four 
tributaries and the main stem to represent each subwatershed and landscape position 
of the Northeast Branch Watershed.  
 
The high percentage of impervious surfaces in the watershed resulted in more soil 
samples from banks, because the paved areas, mostly located in upland areas, could 
not be sampled. Of the bank samples, there were three sample sites along the Paint 
Branch, two along the Little Paint Branch, four along Indian Creek, five along 
Beaverdam, and three along the Northeast Branch. Bank sample sites were chosen at 
locations where stream width or depth change, topography changes, there was a 
confluence, or visible erosion was present. These geomorphic changes in the stream 
areas indicated a possible change in soils and sediments. At five of these stream bank 
sites, samples were taken along a transect moving away from the bank to include the 
landscape positions of stream bank, floodplain, and upland. Depending on the 
topography, the upland position was either a backslope or a summit position. These 
five sites were selected to represent physiographic region.  
 
The stream banks were defined as those areas immediately adjacent to a stream that 
contain the stream at average flow. Floodplains were defined as those areas that are 
broad flatlands on either side of a stream that accommodate maximum flood capacity 
of a 100-year storm. All other areas were classified as upland. 
 
Sediment source samples were collected from upland soils at 2-5 cm depth of soil 
with a stainless instrument. Bank samples were collected by scraping away any 
material plastered from previous stormflow from the bank and vertically integrating 
the samples. Stream banks in most areas of this watershed were not higher than one 
meter and were usually shorter.  
 
 46 
 
 
Street residue was collected for both physiographic regions in the urbanized portions 
of the watershed. In addition, one construction site located adjacent to a stream and 
near the suspended sediment collection site was sampled. Street residue was collected 
by sweeping the street 3.0 meters on either side of a storm drain located on a street 
adjacent to a stream. Streets were swept with a clean plastic-bristled broom from the 
center of the street to the curb on both sides of the street (Breault et al., 2005). Street 
residue samples were collected by aggregating residue from six different streets. To 
represent seasonal variation, a sample was collected just after leaf fall, in the fall of 
2005 and another in the late spring of 2006. The winter was atypically dry with little 
snowfall, so salt was not broadly applied to streets. Accordingly, a winter sample was 
not collected.  
 
Four point bars located in the downstream mixed areas of the watershed were also 
sampled to determine if the bed sediment carried a different chemical signature. 
Sediment was collected from the top five cm from the upstream end of the bars where 
the grain size is finer.  
 
Suspended sediment samples were collected in a well-mixed area of the stream just 
above the U.S. Geological Survey gauge site. The sampling site was upstream of two 
tributaries that drain into the USGS gauge (Latitude 38?57'36.9", Longitude 
76?55'33.5", Fig. 4-1). These tributaries are primarily concrete storm water channels. 
Because they are concrete channels, the travel time from these areas to the gauge site 
is less than 0.75 hour on average (K.L. Prestegaard, personal communication, 2005). 
These fluxes would not have been represented in suspended sediment samples 
because they move through rapidly.  
 
Suspended sediment samples were collected when rainfall covered the entire 
watershed. Weather forecasts were monitored via the real-time forecasts (Department 
of Meteorology University of Maryland, NOAA). In addition, a rain gauge eight km 
from the site was monitored at least three times per day. When rainfall was close to 
2.5 cm, the Northeast Branch was visited and assessed for an increase in discharge 
and sediment movement into the stream. Increased discharge was most likely when 
the conditions preceding the storm had been wet. Samples were not collected when 
unsafe conditions existed due to darkness or excessive ice on the roads.  
 
Suspended sediment samples were collected during the rising limb of the hydrograph, 
which is when the sediment load was most concentrated (Appendix E). An additional 
sample was collected on April 23, 2006 during the falling limb of the hydrograph.  
 
A flow-proportional method was employed by placing a trash pump into the stream 
where the current held it above the stream bed but below the surface of the water. 
Particle sizes <63 ?m are well mixed in the water column, so it was not necessary to 
collect at a particular depth (Vanoni, 1975). The stream water was pumped into three 
208-liter steel barrels lined with round-bottomed plastic bags.  
 
 47 
 
 
Eight suspended sediment samples were collected for the duration of one year to 
account for seasonal differences in erosion and run-off. The sampling year was July 
1, 2005 through June 30, 2006. Collection dates and storm information are in Table 4-
1, Fig 4-2, and Appendix E. The fine sediments are on average 25% of the total load 
carried in the stream. The larger particles do not travel far downstream, but settle out 
of the water column (Vanoni, 1975). 
Daily Average Discharge, NE Branch Anacostia River
1
10
100
1000
10000
9/5/2005 10/25/2005 12/14/2005 2/2/2006 3/24/2006 5/13/2006 7/2/2006 8/21/2006
Di
s
c
h
a
r
g
e
,
 c
f
s
 
Fig. 4-2: Daily average discharge, Northeast Branch of the Anacostia River. Arrows 
indicate date suspended sediment sampled. 
Lab Analyses 
Sample Preparation 
Sediment source samples were air-dried in an isolated lab to decrease the potential of 
cross-contamination. Visible leaves, twigs, and garbage in the samples were removed 
manually. Each sample was ground with a mortar and pestle and sieved to <63 ?m 
using a shaker.  
 
Suspended sediment samples were prepared by removing the stream water from the 
barrels within one week following collection. A Teflon-lined continuous centrifuge 
was used for sediment removal. The centrifuge was selected because the Teflon lining 
would prevent metal contamination from the machine itself. Suspended sediment 
samples were dried in an oven at 60? C. The suspended sediment samples were also 
ground and dry-sieved to <63 ?m using a shaker. 
 
Organic matter was not removed from any of the samples prior to radionuclide 
analysis. Organic matter differences are complex and difficult to generalize, therefore 
even where significant organic matter is present, corrections are frequently not made 
(Walling, 2005). The sample preparation process for the elemental analysis by ICP 
 48 
 
 
includes dissolution by HNO
3
. This dissolution oxidizes the organic matter and leaves 
the metals as residuals. This yields the total elemental load for this size class, but does 
not discriminate among the metals sorbed to organic versus inorganic sediment (Sun 
et al., 2001; Impellitteri et al., 2002). In all of the source samples, total carbon was < 
9.9%. 
 
Table 4-1: Suspended sediment sample characteristics. **at last available time before 
gauge failed.  
Date and 
time of 
suspended 
sediment 
collection 
Q at time 
of 
suspended 
sediment 
collection 
(m
3
 s
-1
) 
Date and 
time of 
hydrograph 
peak 
Q at 
hydrograph 
peak 
(m
3
 s
-1
) 
Sediment 
collected 
(g m
-3
) 
<0.63 ?m 
Sediment 
(g m
-3
) 
<0.63 ?m 
sediment/bulk 
sediment 
collected 
Sediment 
flux at 
time 
collected 
(g s
-1
, 
<0.63 
?m) 
10/7/2005 at 
12:30 
15.51 10/8, at 
7:00
55.87 23.05 8.21 36% 127.37
10/22/2005 
at 9:00 
0.83 10/22 at 
9:30
2.16 250.72 53.76 21% 44.67
12/25/2005 
at 15:00 
0.33 12/25 at 
18.15
3.32 105.46 31.57 30% 10.52
1/18/2006 at 
9:00 
4.00 1/18 at
11:45
10.55 175.12 37.56 21% 150.15
2/4/2006 at 
16:00 
0.80 2/4 at 21:00 6.61 120.7 31.99 27% 25.58
4/23/2006 at 
6:05 
1.46 4/22 at
21:00
9.09 623.46 199.33 32% 290.63
5/11/2006 at 
20:30 
0.39 5/11 at
21:00
4.24 114.2 15.914 14% 6.24
6/25/2006 at 
11:45 
14.39 6/25 at
20:30 **
173.72** 293.09 62.23 21% 31613.37
 
Radionuclides 
Radionuclide analysis was performed on sediment source samples and suspended 
sediment samples by gamma-ray spectrometry using a Canberra-2000 Genie-2000 
Spectroscopy System for 24 hours. The long duration of the spectrometry allows 
lower concentrations to be detected. The system was calibrated and efficiency 
determined using an analytic mixed radionuclide standard; the calibrations are traced 
to the U.S. National Institute of Standards and Technology.  
 
Elemental analysis by ICP-MS and ICP-ES 
The elemental analysis was performed by Acme Analytical Laboratories Ltd. ICP-
emission spectrometry was performed following a LiBO
2
/Li
2
B
4
O
7 
fusion at 980?C 
and dissolution in 0.5% HNO
3
. The LiBO
2
/Li
2
B
4
O
7 
fusion process lowers the boiling 
point allowing for more complete dissolution. Rare Earth and refractory elements 
were tested by ICP-mass spectrometry following the same fusion and dissolution 
process. Total carbon and sulfur were determined by LECO. Loss on ignition (LOI) 
 49 
 
 
was determined by weight difference after ignition at 1000?C. The elements and 
method for each are in Table 4-2. 
 
A sample-prep blank was carried through all stages of preparation and analysis as the 
first sample. Two reagent blanks measured background. Standard reference materials 
prepared by Acme using internationally certified reference materials were used to 
monitor accuracy. Maximum acceptable error for precision and accuracy was ?5% at 
concentrations >100X the detection limit (Appendix D).  
 
Table 4-2: Element analytical methods 
Method Chemical Tracer
LiBO
2
/Li
2
B
4
O
7
 fusion analysis by ICP-
ES. 
SiO
2
, Al
2
O
3
, Fe
2
O
3
, MgO, CaO, Na
2
O, 
K
2
O, TiO
2
, P
2
O
5
, MnO, Cr
2
O
3
, Ni, and 
Sc 
LECO Total C, Total S 
Rare Earth and Refractory Elements: 
LiBO
2
/Li
2
B
4
O
7
 fusion analysis by ICP-
MS 
Ba, Be, Co, Cs, Ga, Hf, Nb, Rb, Sn, Sr, 
Ta, Th, U, V, W, Zr, Y, La, Ce, Pr, Nd, 
Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb, Lu 
Leach with 3 ml 2-2-2 HCl-HNO
3
-H
2
O 
(aqua regia) at 95?C for 1 hour, diluted to 
10 ml, analyzed by ICP-MS. 
Mo, Cu, Pb, Zn, Ni, As, Cd, Sb, Bi, Ag, 
Au, Hg, Tl, Se 
 
Sediment Source Identification 
Tracer concentrations were examined for normality using SAS v. 9.1 (SAS Institute 
Inc., 2004). Unequal variances were evident from the probability plot and box plots. 
Normality was violated, so non-parametric tests were used. In addition, there were 
orders of magnitude differences between the tracers and ranking the data allowed for 
clearer comparison.  
 
Source groups were determined based on a Spearman-correlation matrix of elemental 
and radionuclide concentrations. In addition, the landscape model derived from a soil 
survey (Appendix A) helped to inform the grouping of sediment sources. The sources 
of fine sediment in the Northeast Branch watershed were grouped in two ways?by 
subwatershed and landscape position. Each source grouping and subsequent 
comparison with suspended sediment was considered separately. 
 
Those tracers that were more enriched in the sediment source samples than in the 
mean composition of the Earth?s upper continental crust were removed following 
procedures outlined in a companion study (Chapter 3). The tracers were removed 
because some did not vary on a spatial pattern and others did not bracket the 
suspended sediment. Those that did not vary on a spatial pattern included: Hg, Pb, 
Ag, Se, Hf, Yb, Er, Nd, Pr, Nb, Zr, and Sb. Removing these enabled focus on the 
parameters that best described the sediment sources on a spatial basis. Those tracers 
in the suspended sediment whose concentrations were not bracketed by the sediment 
source areas were an indication of a missing source of sediment. These tracers were 
largely from atmospheric particulate matter (Chapter 3). Street residue concentrations 
 50 
 
 
approached concentrations measured in the suspended sediment, but were not always 
greater than the suspended sediment. I discarded these tracers from statistical 
consideration. The tracers that were removed because the tracer concentrations were 
higher in the suspended sediment than in the sources identified for this study 
included: MnO, Cu, Zn, Co, Ni, As, Cd, and Au. Those that are bracketed are shown 
in Table 4-3. 
 
Table 4-3: Bracketed Tracers 
  Sediment Sources Suspended Sediment 
Tracer Units Mean 
Sample 
Variance Min. Max. Mean 
Sample 
Variance Min. Max. 
Cs-137 Bq/kg 15.34 330.62 0.00 61.11 3.55 5.56 0.00 6.09
K-40 Bq/kg 760.47 288315.30 4.26 2342.38 1315.55 596058.68 118.30 2660.29
SiO
2
% 72.11 47.74 56.27 78.94 63.59 9.99 59.09 70.32
Al
2
O
3
% 9.93 10.06 6.45 19.33 12.28 2.72 10.08 14.52
Fe
2
O
3
% 3.74 0.76 2.84 5.70 5.66 0.27 4.75 6.36
MgO % 0.40 0.02 0.20 0.65 1.03 0.05 0.77 1.47
CaO % 0.26 0.03 0.03 0.70 0.83 0.12 0.50 1.65
Na
2
O % 0.27 0.01 0.14 0.45 0.31 0.00 0.26 0.38
K
2
O % 1.27 0.06 0.80 1.68 1.50 0.03 1.28 1.80
TiO
2
% 1.19 0.13 0.91 2.39 1.08 0.01 0.95 1.16
P
2
O
5
% 0.14 0.00 0.07 0.29 0.24 0.00 0.16 0.32
MnO % 0.05 0.00 0.01 0.14 0.19 0.01 0.07 0.28
Cr
2
O
3
% 0.01 0.00 0.01 0.02 0.02 0.00 0.01 0.02
Ni mg/kg 25.00 137.38 5.00 45.00 66.88 232.70 47.00 95.00
Sc mg/kg 10.43 5.80 8.00 17.00 13.13 1.55 12.00 15.00
LOI % 10.47 27.95 5.90 22.30 13.11 4.93 10.30 17.60
Total C % 3.00 5.58 0.13 7.73 4.11 1.63 3.08 6.73
Total S % 0.04 0.00 0.01 0.12 0.09 0.00 0.05 0.19
Sum % 99.85 0.03 99.38 99.98 99.85 0.00 99.73 99.92
Ba mg/kg 294.72 3132.50 184.70 377.80 342.81 2849.48 288.00 430.50
Be mg/kg 1.79 0.95 1.00 4.00 2.25 0.21 2.00 3.00
Co mg/kg 13.60 58.67 1.80 23.60 42.11 117.80 27.30 55.60
Cs mg/kg 2.94 0.87 1.80 4.90 3.54 0.13 3.10 4.10
Ga mg/kg 12.59 12.16 8.10 21.20 16.11 4.44 12.90 18.50
Hf mg/kg 18.89 147.52 5.20 55.60 10.63 3.64 7.70 14.20
Nb mg/kg 19.86 11.61 16.50 29.70 18.44 1.91 16.10 20.00
Rb mg/kg 61.29 175.08 37.40 82.90 74.58 48.13 64.40 86.20
Sn mg/kg 3.07 1.76 2.00 6.00 4.75 0.50 4.00 6.00
Sr mg/kg 53.54 48.00 40.30 62.70 66.13 4.01 63.00 69.10
Ta mg/kg 1.54 0.06 1.10 2.20 1.33 0.02 1.20 1.50
Th mg/kg 11.59 5.30 9.30 17.10 10.13 1.01 8.70 11.40
U mg/kg 3.96 0.71 2.70 6.50 3.54 0.04 3.30 3.90
V mg/kg 81.29 353.30 56.00 129.00 105.38 59.13 93.00 119.00
W mg/kg 1.94 0.07 1.50 2.40 1.98 0.08 1.70 2.50
Zr mg/kg 691.31 230242.12 211.00 2182.40 413.73 5259.05 301.40 545.50
Y mg/kg 46.70 112.46 32.40 76.10 43.88 4.03 40.50 46.40
 51 
 
 
  Sediment Sources Suspended Sediment 
Tracer Units Mean 
Sample 
Variance Min. Max. Mean 
Sample 
Variance Min. Max. 
La mg/kg 40.56 34.63 29.30 51.00 40.21 19.86 34.80 49.80
Ce mg/kg 90.89 247.59 64.30 127.40 98.60 73.82 87.70 112.20
Pr mg/kg 10.01 3.30 7.14 14.33 11.09 0.92 9.93 12.56
Nd mg/kg 37.14 43.92 24.90 49.80 40.96 18.86 35.90 49.30
Sm mg/kg 7.66 2.23 5.10 9.70 8.58 0.41 7.80 9.80
Eu mg/kg 1.44 0.09 0.85 1.83 1.55 0.03 1.38 1.86
Gd mg/kg 7.13 2.10 4.40 9.78 7.42 0.41 6.52 8.56
Tb mg/kg 1.27 0.07 0.85 1.93 1.31 0.01 1.17 1.43
Dy mg/kg 7.82 1.98 5.33 11.17 7.53 0.19 6.79 7.99
Ho mg/kg 1.62 0.13 1.16 2.67 1.48 0.00 1.35 1.55
Er mg/kg 4.88 1.20 3.64 7.79 4.15 0.09 3.78 4.81
Tm mg/kg 0.75 0.03 0.54 1.20 0.63 0.01 0.55 0.77
Yb mg/kg 4.79 1.32 3.43 7.85 3.88 0.11 3.41 4.48
Lu mg/kg 0.75 0.04 0.47 1.32 0.63 0.00 0.58 0.66
Mo mg/kg 0.98 0.42 0.20 3.00 3.01 6.13 1.40 9.00
Cu mg/kg 31.70 505.17 16.50 100.50 95.53 683.00 52.10 129.50
Pb mg/kg 31.15 315.76 5.80 70.70 58.65 73.47 46.20 70.00
Zn mg/kg 60.86 1149.52 21.00 149.00 299.88 5576.13 194.00 411.00
Ni mg/kg 15.36 59.30 2.60 30.70 48.16 196.58 30.70 75.10
As mg/kg 4.48 1.68 3.00 7.00 5.90 0.43 4.90 7.20
Cd mg/kg 0.21 0.02 0.10 0.50 3.15 10.96 0.70 10.80
Sb mg/kg 0.31 0.03 0.10 0.60 1.10 0.18 0.40 1.90
Bi mg/kg 0.16 0.00 0.10 0.20 0.29 0.01 0.20 0.50
Ag mg/kg 0.15 0.02 0.10 0.60 0.26 0.01 0.10 0.40
Au ?g/kg 4.27 32.06 0.50 20.40 9.13 30.63 3.50 21.10
Hg mg/kg 0.08 0.00 0.03 0.25 0.20 0.00 0.12 0.28
Tl mg/kg 0.11 0.00 0.10 0.20 0.19 0.00 0.10 0.20
Se mg/kg 0.59 0.04 0.50 1.20 0.63 0.01 0.50 0.70
 
The non-parametric Kruskal-Wallis means comparison test was used to identify the 
remaining tracers with significant differences between source groups. Tracers that 
failed this test (p < 0.05) were removed. The Mann-Whitney test, which is a non-
parametric means comparison test between two variables, was used to compare 
sediment source groups on a pair-wise basis. Sediment source groups that showed no 
significant differences in this test were eliminated.  
 
Others have successfully used discriminant function analysis to further discriminate 
which of the tracers are significant between each of the source groups (Owens et al., 
2000; Walling, 2005). It was not an option in this study because the data violated four 
of the five assumptions of discriminant function analysis: 1) normality, 2) no outliers, 
3) homoescadacity, 4) no skewness, 5) no multi-collinearity (Poulson and French, 
2006). Many of the tracers were highly correlated with each other, as would be 
expected with the elements that make up the minerals in the sediment. In addition, 
there were several outliers, presumably from specific outcroppings of a unique 
 52 
 
 
geologic material or of a point-source pollutant. These outliers impact the mean and 
increase variability. Thus, the pooled variances result in erroneous statistical 
significance. The data also does not have homoescedascity. This could affect within-
group variations of the source groups. Lastly, the data was not normally distributed, 
which was caused by skewness and outliers. In lieu of discriminant function analysis 
to determine between group variations, I used Mann-Whitney and tested the 
differences in tracers in each pair of sources.  
 
Once the tracers identifying each source group were statistically verified, a mixing 
model was used to attribute relative quantities of suspended sediment to sources for a 
particular storm event. Mixing models have been used in groundwater research for 
decades when looking for sources in aquifers. In sediment fingerprinting, various 
mixing models have been used (Rowan et al., 2000; Collins et al., 1997b; Walling et 
al., 1999). The model I used was developed specifically for small sample sizes (Gellis 
and Landwehr, 2006). The model solves the system of linear equations whereby the 
sum of the source contributions equals unity and the standard deviation of the tracer 
properties in the suspended sediment are related to the source group. The source 
contributions account for the unknown proportion of contributions of each source 
group. The function was minimized by the root mean square of the variance of the 
tracers in the source groups. The error term is a standard deviation; therefore the 
tracers are normalized in the mixing model. Each source and tracer is scaled by the 
standard deviation of the mixture so that they are comparable. This model is unique 
because it is based on normalizing a standard deviation of a mixture as opposed to 
directly relating the values of each variable; thereby giving this model more statistical 
power (J.M. Landwehr, personal communication, 2006). Tracers which approach zero 
concentration may be used because the formula is a standard deviation and does not 
depend on dividing directly by the tracer concentration. This is particularly useful 
when using elemental tracers that are commonly found in very low concentrations.  
 
?
?
?
=
=
=
?
?
?
?
?
?
?
?
?
?
?
?
?
=
T
t
S
S
S
S
VARstfs
fsAstVt
T
E
1
1
2
1
)
1
(
  (4-1) 
 
 
Where:  
T=total no. of tracers 
t =a specific tracer 
Vt=value of the tracer t in the suspended sediment 
fs=fractional sources, must sum to one 
S=total no. of source groups 
s=source group 
Ast=average of a specific tracer (t) in the source group (s) 
VARst= variance of a specific tracer (t) in the source group (s) 
 53 
 
 
 
With three source groups, the same error value can be generated by a variety of 
different parameter combinations. This indicates that a single optimized solution is 
only one of many statistically equivalent solutions that each may yield different 
results in terms of source contribution (Rowan et al., 2000). Some of the sources 
necessarily have a greater degree of sensitivity. This may be caused by the tracers 
used to identify this source group or by a higher number of samples in a particular 
source group. Thus, the optimal model is only one of many outcomes that have a 
small error. Since the model can have the same optimization, it was necessary to 
compare the different explanations and use the explanation that makes the most sense 
given the study site (Owens et al., 2000). Propagating all uncertainties to the final 
source contribution could be achieved through a Bayesian Monte Carlo simulation 
approach (Rowan et al., 2000). However, that is outside the scope of this project. In 
lieu of additional statistical tests, results were compared with the landscape model, 
developed during a soil survey (Chapter 2). This landscape model provided 
information about erodibility mechanisms such as mass wasting, soil structure, 
particle size differences, and channel roughness. The source contributions with 
minimum error matched predictions from the landscape model.  
Results 
Suspended Sediment Characterization 
The elemental concentrations in the suspended sediment are shown in Table 4-4. The 
suspended sediment collected during storm events was most concentrated in 
anthropogenic elements. The most variation in suspended sediment between storm 
events was in Cd. 
 
Table 4-4: Elemental concentrations in suspended sediment.  
Trace Elements Units Mean STD CV Min Max 
Sc mg kg
-1
13.13 1.25 9.50 12.00 15.00
Total S wt. % 0.09 0.05 54.02 0.05 0.19
Ba mg kg
-1
342.81 53.38 15.57 288.00 430.50
Be ? 2.25 0.46 20.57 2.00 3.00
Co ? 42.11 10.85 25.77 27.30 55.60
Cs ? 3.54 0.36 10.13 3.10 4.10
Ga ? 16.11 2.11 13.08 12.90 18.50
Hf ? 10.63 1.91 17.95 7.70 14.20
Nb ? 18.44 1.38 7.50 16.10 20.00
Rb ? 74.58 6.94 9.30 64.40 86.20
Sn ? 4.75 0.71 14.89 4.00 6.00
Sr ? 66.13 2.00 3.03 63.00 69.10
Ta ? 1.33 0.13 9.67 1.20 1.50
Th ? 10.13 1.00 9.92 8.70 11.40
U ? 3.54 0.19 5.43 3.30 3.90
V ? 105.38 7.69 7.30 93.00 119.00
 54 
 
 
Trace Elements Units Mean STD CV Min Max 
W ? 1.98 0.28 14.00 1.70 2.50
Zr ? 413.73 72.52 17.53 301.40 545.50
Y ? 43.88 2.01 4.58 40.50 46.40
La ? 40.21 4.46 11.08 34.80 49.80
Ce ? 98.60 8.59 8.71 87.70 112.20
Pr ? 11.09 0.96 8.66 9.93 12.56
Nd ? 40.96 4.34 10.60 35.90 49.30
Sm ? 8.58 0.64 7.50 7.80 9.80
Eu ? 1.55 0.16 10.19 1.38 1.86
Gd ? 7.42 0.64 8.65 6.52 8.56
Tb ? 1.31 0.09 6.53 1.17 1.43
Dy ? 7.53 0.44 5.82 6.79 7.99
Ho ? 1.48 0.07 4.57 1.35 1.55
Er ? 4.15 0.30 7.25 3.78 4.81
Tm ? 0.63 0.07 11.48 0.55 0.77
Yb ? 3.88 0.34 8.70 3.41 4.48
Lu ? 0.63 0.03 4.36 0.58 0.66
Mo ? 3.01 2.48 82.17 1.40 9.00
Cu ? 95.53 26.13 27.36 52.10 129.50
Pb ? 58.65 8.57 14.61 46.20 70.00
Zn ? 299.88 74.67 24.90 194.00 411.00
Ni ? 48.16 14.02 29.11 30.70 75.10
As ? 5.90 0.65 11.06 4.90 7.20
Cd ? 3.15 3.31 105.11 0.70 10.80
Sb ? 1.10 0.43 38.87 0.40 1.90
Bi ? 0.29 0.10 34.47 0.20 0.50
Ag ? 0.26 0.12 45.25 0.10 0.40
Au ? 9.13 5.53 60.65 3.50 21.10
Hg ? 0.20 0.06 29.07 0.12 0.28
Tl ? 0.19 0.04 18.86 0.10 0.20
Se ? 0.63 0.07 11.31 0.50 0.70
 
Minor Elements Units Mean STD CV Min Max 
Na2O wt. % 0.31 0.04 14.21 0.26 0.38
MnO ? 0.19 0.08 39.69 0.07 0.28
Cr
2
O
3
? 0.02 0.00 20.16 0.01 0.02
 
Major Elements Units Mean STD CV Min Max 
SiO
2
wt. % 63.59 3.16 4.97 59.09 70.32
Al
2
O
3
? 12.28 1.65 13.43 10.08 14.52
K
2
O ? 1.50 0.17 11.51 1.28 1.80
Fe
2
O
3
? 5.66 0.52 9.19 4.75 6.36
MgO ? 1.03 0.21 20.67 0.77 1.47
CaO ? 0.83 0.35 41.86 0.50 1.65
TiO
2
? 1.08 0.08 7.07 0.95 1.16
LOI ? 13.11 2.22 16.93 10.30 17.60
 55 
 
 
 
Radionuclide 
Isotopes 
Units Mean STD CV Min Max
Cs-137 Bq kg-1 3.55 2.36 66.46 0.00 6.09
K-40 Bq kg-1 1315.55 772.05 58.69 118.30 2660.29
Sediment Source Group Characterization 
Source Groups: East and West Subwatershed 
Those tracers that were not significant at p < 0.05 in the Mann-Whitney test were the 
two radionuclides, Cs-137 and K-40, as well as P
2
O
5
, CaO, Sr, W, and Mo. These 
were removed from further consideration, in addition to the anthropogenically 
enriched elements discussed in Chapter 3. The downstream portions of the Northeast 
Branch watershed, where the west and east subwatersheds mix, were not considered 
in the Mann-Whitney test. The mixed portion of the watershed is not an end-member 
and therefore sediment source can not be ascribed to the mixed portion using a 
mixing model.  
 
Several of the significant tracers covaried (Fig. 4-3). To strengthen the model, 
elements were selected that were neither highly correlated nor necessarily coexist 
with one another because of chemical properties. Tracers with the lowest p-value 
were selected from different groups of the periodic table, e.g.: one element from the 
lanthanides and one from the alkali Earth metals (Appendix H). When there was more 
than one element in a group with a very low p-value, the tracer with the largest 
atomic radius, lowest ionic potential, and lowest electron affinity was selected. The 
tracers that were selected based on these criteria included Ga, Ta, Ce, and U (Table 4-
5). None of these four elements naturally bind with one another and all are stable in 
the environment. Each was graphed against the other and none co-varied.  
 
0
50
100
150
200
250
mg/
k
g
Ce Pr Nd Sm Eu Gd Tb
Dy Ho Er Tm Yb Lu
 
Fig. 4-3: Covariance between tracers in the Lanthanide Group. X-axis is 
sample. 
 56 
 
 
 
Landscape Position Alternative Tracers
-5
0
5
10
15
20
25
Co
n
cen
t
r
at
io
n
CaO Cr2O3 Gd
 
Landscape Position Tracers
0
10
20
30
40
50
60
70
80
90
100
110
120
130
140
mg
/k
g
Ho Sr W
 
 
Subwatershed Tracers
0
30
60
90
120
150
180
210
mg
/
k
g
Ce Ga Rb U
 
Subwatershed Alternative Tracers
0
100
200
300
400
500
600
700
800
mg/
k
g
Ba Bi Sm Th
 
Fig: 4-4: Covariance between tracers used in the mixing model. X-axis is sample. 
 
Table 4-5: Tracers that Differentiate Sediment Sources by Hydrological 
Subwatershed Boundary 
Tracer Kruskal-
Wallis 
West 
Mean 
(N=10) 
West 
CV 
(N=10) 
East 
mean 
(N=15) 
East CV 
(N=15) 
Periodic Group 
Ce (mg kg
-1
) 0.000 163.98 19.88 89.02 14.81 Lanthanide
Ga (mg kg
-1
) 0.012 20.27 22.97 12.24 23.72 Metal
Ta (mg kg
-1
) 0.000 2.11 12.19 1.47 11.61 Transition metal
U (mg kg
-1
) 0.000 5.99 20.78 3.85 10.89 Actinide
 
Results of the mixing model that was used to determine the relative quantity of fine 
sediment attributable to each source indicate that storm events varied considerably 
according to season, rainfall intensity and rainfall volume (Table 4-6). Suspended 
sediment was attributable to the east subwatershed for all events except the October 
22, 2005 storm event and the February 9, 2006 storm event. In the October 22, 2005 
storm event, the rainfall volume was low and a 200-year storm had occurred two 
weeks prior. The February 9, 2006 storm event is the only event sampled where there 
had been sizeable rainfall and runoff within the previous 24 hours. Both the October 
22 and February 9 event had suspended sediment that could be attributed to the west 
subwatershed.  
 
Travel paths are further from the west to the gauge than from the east to the gauge. 
Sediment measured at a particular point in the hydrograph may represent a sample 
based more on travel time than a preference for either subwatershed to contribute fine 
sediment. Since the two events contributing suspended sediment from the west had 
preceding conditions that moved sediment, it is possible that sediment in temporary 
storage only made it to the gauge in time for sampling with the addition of a 
 57 
 
 
subsequent storm event. Without the subsequently sampled storm, the west suspended 
sediment would move through during the falling limb of the hydrograph.  
 
Table 4-6: Estimation of Relative Sediment Source Contributions by Hydrological 
Subwatershed using tracers Gallium, Tantalum, Cerium, and Uranium. 
Suspended 
Sediment 
West West Flux  East East Flux  Error 
Date % 
Contributing
(g/s) % 
Contributing
(g/s)  
10/7/2005 0% 0.00 100% 127.37 0.377
10/22/2005 31% 13.80 69% 30.86 0.748
12/25/2005 0% 0.00 100% 10.52 0.661
1/18/2006 0% 0.00 100% 150.15 0.957
2/4/2006 5% 1.30 95% 24.28 0.213
4/23/2006 0% 0.00 100% 290.63 0.784
5/11/2006 0% 0.00 100% 6.24 0.676
6/25/2006 0% 0.00 100% 31613.37 0.729
            
average 5% 1.89 96% 4031.68 0.643
fall average 16% 6.90 85% 79.12 0.563
winter average 2% 0.43 98% 61.65 0.61
spring average 0% 0.00 100% 10636.75 0.73
 
Source Groups: Landscape Position 
The Kruskal-Wallis test performed on point bars, banks, floodplains, upland areas, 
and street residue identified 18 significant tracers. Non-significant tracers, including 
SiO
2
, Al
2
O
3
, Fe
2
O
3
, MgO, Sn, Na
2
O, K
2
O, Ba, Be, Cs, Ga, Nb, Rb, Mo, Ta, Th, U, V, 
Ce, Bi, and Tl, were removed from further analysis. 
 
A pair-wise comparison of source groups using the Mann-Whitney test showed no 
tracers were found significantly different between the floodplain and the streets or the 
floodplain and the upland areas. Only one tracer was found in a significantly different 
concentration between the floodplains and the banks?Cs-137. A floodplain is similar 
to a bank since it includes the gradual addition of the same sediments that form a 
bank. Therefore, floodplains were not considered as a separate source group.  
 
The Mann-Whitney test identified only one tracer as significantly different between 
point bars and streets?Cs-137. Point bars are made up of stream bed material, which 
is temporary storage of sediment. This source group is also confounded by an overall 
larger particle size (Appendix F). Sieving the point bars to the <0.63 ?m particle size 
did result in enough sample for analysis, but the mean particle-size of the point bars 
was much higher than in any other sample. Therefore, point bars could not be 
considered comparable independent sources and were not considered as a separate 
source group.  
 
 58 
 
 
The source groups that were possible to discriminate, and therefore available to 
attribute suspended sediment to, were the banks, upland, and street residue landscape 
positions. The tracers that were significant when comparing the bank and upland 
were: Cs-137, K-40, TiO
2
, P
2
O
5
, Y, La, Eu, Gd, Tb, Dy, Ho, Tm, Lu, and Sm. The 
tracers that significantly differentiated the bank from the street residue were: P
2
O
5
, 
La, Eu, Tb, Dy, Ho, Tm, CaO, Sm, Sr, and W. The tracers that differentiated the 
upland from the streets were CaO, Cr
2
O
3
, Sr, and W.  
 
To eliminate multi-collinearity and to ensure that at least one tracer uniquely 
separated each of the three source groups, I narrowed the set of tracers to Ho, Sr, and 
W (Table 4-7). These same criteria of low p value, small hydrated radius, low ionic 
potential, and low electron affinity were again used. None co-occur together 
chemically and when graphed by sample site, they do not covary.  
 
Table 4-7: Tracers that differentiate the Banks, Upland and Street Residue 
Tracer (mg 
kg
-1
 g
-1
) 
Kruskal-
Wallis 
Bank 
Mean 
(N=18) 
Bank 
CV 
Streets 
Mean 
(N=2) 
Street 
CV 
Upland 
Mean 
(N=14) 
Upland 
CV 
Periodic 
Group 
Ho (mg kg
-1
) 0.005 2.17 34.46 1.31 1.63 1.62 24.66 Lanthanide
Sr (mg kg
-1
) 0.043 58.23 19.37 117.75 22.04 57.12 17.63 Alkali Earth 
Metal
W (mg kg
-1
) 0.033 1.97 20.27 1.15 6.15 1.75 20.39 Transition 
Metal
 
 
Table 4-8: Estimation of Relative Sediment Source Contributions by Landscape 
Position using Tracers: Holmium, Strontium, and Tungsten 
Date of 
Suspended 
Sediment 
Collection 
Banks % 
Contributing 
Banks 
flux 
(g/s) 
Streets % 
Contributing
Streets 
Flux 
(g/s) 
Upland % 
Contributing 
Upland 
Flux (g/s) 
Error 
10/7/2005 98% 124.82 2% 2.55 0% 0.00 0.904
10/22/2005 85% 37.97 15% 6.70 0% 0.00 0.515
12/25/2005 11% 1.16 14% 1.47 75% 7.89 0.253
1/18/2006 12% 18.02 12% 18.02 76% 114.11 0.138
2/4/2006 90% 23.03 10% 2.56 0% 0.00 0.356
4/23/2006 85% 247.04 15% 43.59 0% 0.00 0.394
5/11/2006 82% 5.11 18% 1.12 0% 0.00 0.748
6/25/2006 0% 0.00 15% 4742.01 85% 26871.36 0.089
                
average 58% 2335.43 13% 508.23 30% 1189.90 0.424
fall average 92% 78.71 9% 7.31 0% 0.00 0.707
winter 
average 
38% 23.41 12% 7.45 50% 31.23 0.249
spring 
average 
56% 5924.67 16% 1701.88 28% 3010.20 0.41
 
 59 
 
 
The mixing model was applied to this source grouping (Table 4-8). The banks were 
the dominant contributing source in the fall and spring. The upland contributed more 
fine sediment to the stream in the winter and during the June 26 storm event. This 
event was the largest storm of the year and was a 200-year storm. The seasonal 
variation for the bank and upland samples indicates that leaf cover protects the upland 
areas from eroding fine sediment in the fall and spring. Leaf cover affects raindrop 
erosion. These data indicate different erosion mechanisms are in effect in different 
seasons of the year. More sloughing and mass wasting occurs in winter while 
overland erosion predominates in the summer.  
 
The street residue contribution was consistently around 12%. It was not affected by 
seasonal variation. Its coverage, infiltration, and imperviousness are static. The one 
exception was the October 7, 2006 storm event. This event was a 30-year storm of 25 
hours, 18 cm accumlated rainfall. The lower relative contribution reflects limited fine 
sediment supply and dilution. Our finding for street residue contribution is consistent 
with other researchers who found that roads may contribute 10-20% of sediment 
during smaller, low intensity storms (Burton and Pitt, 2002).  
 
Any source with a limited availability of fine sediment will reach a maximum 
contribution. Once that maximum contribution has been reached, calculating relative 
contributions of fine sediment will necessarily make it appear that any source already 
at its maximum in terms of available sediment is contributing less on a proportional 
basis. This is a dilution effect.  
 
Alternative Tracer Selection: Subwatershed  
To test the sensitivity of the criteria used to select the tracers and the mixing model, 
alternative tracers were selected from among those that were significantly different 
between the source groups. Using the same criteria of low p, one tracer per periodic 
group, small hydrated radius, lowest electron affinity, and the lowest ionic potential 
among those in that periodic group, we ran the mixing model with Ba, Bi, Sm and Th 
(Table 4-9).  
 
Table 4-9: Alternative Tracers that Differentiate Sediment Sources by Hydrological 
Subwatershed Boundary 
Tracer Kruskal-
Wallis 
West 
Mean 
(N=10) 
West CV 
(N=10) 
East 
mean 
(N=15) 
East CV 
(N=15) 
Periodic 
Group 
Ba (mg kg
-1
) 0.0001 508.88 31.39 310.69 21.93
Alkali Earth 
Metal 
Bi (mg kg
-1
) 
0.0002 0.36 47.29 0.18 35.24
Metal 
Sm (mg kg
-1
) 
0.0001 13.01 38.44 7.67 20.13
Lanthanide 
Th (mg kg
-1
) 
0.0004 16.18 23.87 11.43 18.57
Actinide 
 
Selecting two different types of storms based on the hydrograph (Appendix E), these 
alternative tracers show different results when used in the mixing model (Table 4-10). 
These sources attributed to the suspended sediment vary significantly when a 
different set of tracers is used. Yet, both sets of tracers result in lower error than when 
 60 
 
 
all of the possible tracers are used. The error is <1.0 with few tracers as opposed to 
errors <5.0 with all possible tracers. Fewer tracers produce better results, but the 
criteria for selecting those tracers and the behavior of the tracers in the erosion and 
transport process requires more investigation. 
 
Table 4-10: Alternative estimation of Relative Sediment Source Contributions by 
Hydrological Subwatershed using tracers Barium, Bismuth, Samarium, and Thorium. 
Suspended 
Sediment 
West West Flux  East East Flux  Error 
Date % 
Contributing
(g/s) % 
Contributing
(g/s)  
10/7/2005 31% 39.48 69% 87.88 0.886
1/18/2006 26% 39.04 74% 18.93 0.670
 
Alternative Tracer Selection: Landscape Position  
To determine the sensitivity of the model to the selection of tracers, I used the same 
criteria of low p value, lowest ionic potential, largest atomic radius, and lowest 
electron affinity within the tracers still available from that periodic group. As an 
alternative, CaO, Cr
2
O
3
, and Gd were selected for analysis (Table 4-11).  
 
Table 4-11: Alternative tracers that differentiate the Banks, Upland and Street 
Residue 
Tracer (mg kg
-1
 
g
-1
) 
Kruskal-
Wallis 
Bank 
Mean 
(N=18) 
Bank 
CV 
Streets 
Mean 
(N=2) 
Street 
CV 
Upland 
Mean 
(N=14) 
Upland 
CV 
Periodic 
Group 
CaO (wt. %) 0.046 
0.34 67.83 4.42 21.30 0.67 141.87
Alkali Earth 
Metal 
Cr
2
O
3
 (wt. %) 0.043 
0.02 99.78 0.02 10.35 0.01 32.75
Transition 
Metal 
Gd (mg kg
-1
) 0.018 
10.18 43.42 5.96 3.68 7.38 40.60
Lanthanide 
 
Table 4-12: Alternative Estimation of Relative Sediment Source Contributions by 
Landscape Position using Tracers: Calcium, Chromium, and Gadolinium 
Date of 
Suspended 
Sediment 
Collection 
Banks % 
Contributing 
Banks 
flux 
(g/s) 
Streets % 
Contributing
Streets 
Flux 
(g/s) 
Upland % 
Contributing 
Upland 
Flux (g/s) 
Error 
10/7/2005 
65% 82.79 11% 14.01 24% 30.57 
0.178
1/18/2006 
90% 135.13 10% 15.01 0% 0.00 
0.237
 
Using these alternative tracers overall reduced error to <1.0 when compared to using 
all possible tracers where error <4.0. However, using Ca, Cr, and Gd produced 
different results than when Ho, Sr, and W were used. Further investigation into the 
behavior of the tracers is required.  
 61 
 
 
Discussion 
Testing the sediment fingerprinting approach in this urban watershed provided 
substantial challenges. Tracers of geogenic origin provide more significant 
differences among sources than tracers of anthropogenic origin. The process by which 
these tracers were chosen required an iterative approach of measuring the 
concentrations of 63 different elements. Only when beginning with so many tracers, 
and measuring the concentrations in both the sources and suspended sediment, could 
those appropriate for this watershed be selected. For future sediment fingerprinting 
research in urban areas, it is recommended to use geogenic tracers appropriate to the 
mineralogy and geology of the region. 
 
There may be statistically equivalent solutions to the mixing model when using a 
large number of tracers. The solution to this problem is to eliminate the covarying 
tracers so there is only one solution to the mixing possible that produces the lowest 
error. The data show that fewer tracers among those that show significant differences 
in the sediment source groups are clearly more powerful. In selecting tracers, multi-
collinearity is a factor that must be considered. Where several tracers covary, only 
one may be selected. Otherwise, source groups that are overrepresented by specific 
tracers will be overrepresented in the results of the mixing model. The criteria for 
selecting specific tracers is critical to ensuring accurate results from the mixing 
model. The behavior of the tracers in the erosion and transport process must be 
conservative.  
 
Finally, a careful interpretation of the attribution of the quantity of sediment from the 
sources must account for a dilution effect. The limited supply of sediment from some 
sources may weight a proportional contribution. This was seen in this data in that the 
street residue was consistently about 12% except where the rainfall volume was 
highest.  
 
Overall, this method is viable for urban watersheds. The adaptations outlined in 
Chapter 3 and careful statistical analysis allow the sediment fingerprinting method to 
be effectively deployed to quantify sediment sources in the predominantly urban 
Northeast Branch Watershed.  
 62 
 
 
Chapter 5: Conclusions and Future Research 
Summary 
This research identifies sediment sources of the suspended fraction in the Northeast 
Branch of the Anacostia River. After reviewing the existing literature on sediment 
source identification, and gathering some preliminary data from the watershed, the 
following three hypotheses were formed:  
 
1. Chemical analyses of sediment can be used to determine the contributions of 
sediment sources in urban watersheds.  
2. Sediment source compositions used in the sediment fingerprinting of urban 
watersheds will include anthropogenically enriched elements. 
3. The primary source of suspended sediment in the Northeast Branch is derived 
from stream bank erosion. 
 
To test these hypotheses in the Northeast Branch watershed, I conducted a soil survey 
to understand historical alluvial deposition patterns under land use change (forested, 
agricultural, urbanization). I mapped soil morphological variation (buried A horizons, 
texture, carbon, clay illuviation, and redoximorphic features) (Appendix A). Sites 
were informed by the use of GIS analysis using: stream maps (both NRCS and 
USGS), hydrological watershed boundary (USGS and NRCS), geology maps 
(Maryland Geological Society), land use (NLCD), topography (Digital Raster 
Graphic, USDA and NRCS), and the Soil Survey (Soil Survey Staff, 1967). In 
addition, I used Cs-137 and Pb concentrations to determine historical alluvial 
deposition patterns under land use change (Chapter 2). 
 
I collected samples from potential sources of sediment at more than 25 sites in the 
watershed including from streambanks, point bars, flood plains, upland areas, and 
street residue (Appendix B) and suspended sediment samples from the Northeast 
Branch at the outlet of the watershed during eight storm events over the course of 12 
months (Appendix E). These samples were analyzed for elemental and radionuclide 
concentrations (Appendix C). Additional data used in the analysis includes mean 
composition of the Earth?s upper continental crust (Rudnick and Gao, 2003). Overall, 
I found tremendous variation in land use over space and time. The soils and geology 
also had significant differences, but were spatially more continuous. Therefore, these 
more continuous differences in soils and geology allow discrimination among 
sediment sources. 
 
In order to evaluate whether elements commonly contributed from anthropogenic 
sources could be used to identify urban contributions to total suspended sediment, I 
used an enrichment ratio method to quantify anthropogenically enhanced elemental 
concentrations in watershed sediment. This allowed elements that are not a result of 
land use changes or point source contamination to be identified (Chapter 3). These 
 63 
 
 
elements were also evaluated to determine whether they formed a distinct end-
member that perhaps defined an urban signature in the watershed. 
 
A multivariate mixing model related the sum of contributions from all the different 
sediment sources to the suspended sediment. In using this model, the goal is to 
minimize error while creating a vector of sources equal to 100%. The error term is 
standard deviation; therefore the tracers are normalized in the mixing model. To 
avoid issues of multi-collinearity, I selected those tracers that were most significantly 
different. Among those, I also selected tracers within each periodic group represented 
that had the smallest hydrated radius and lowest electron affinity (Chapter 4). 
 
In terms of the hypotheses, my findings are as follows. First, there is support for using 
chemical analyses of sediment to determine sources in urban watersheds. The 
chemical tracers effectively discriminated among the sediment source groups. 
Second, sediment source compositions primarily were composed of geogenic 
elements, but the suspended sediment was primarily characterized by anthropogenic 
elements. Moreover, the suspended sediment concentrations were not bracketed by 
the sources, indicating a previously unidentified source group. This result did not 
support the second hypothesis. Further analysis showed that street residue closely 
approached the concentrations in the suspended sediment. Street residue is the only 
source that does not significantly dilute atmospheric particulate matter. This indicates 
that atmospheric particulate matter is one of the sources of the suspended sediment. 
This method allowed for identification of a missing source of sediment?atmospheric 
particulate matter. It also showed street residue as the most contaminated source of 
sediment (e.g., Hg, Pb, and Cd). 
 
Third, on a mass basis, stream bank erosion was the primary source of suspended 
sediment in the Northeast Branch. As just noted, the primary constituents of the 
sediment were geogenic, but the contaminated sediment was related to street residue. 
Solving the mixing model provided the relative quantity of sediment generated by 
each source in the watershed. The eastern portion of the subwatershed contributed the 
most sediment to the fine sediment fraction. This is likely caused by the difference in 
the length of the travel path to station where the suspended sediment was collected. 
The banks contributed the most sediment during the fall and spring while the upland 
areas contributed the most in the winter. Thus, seasonal differences were 
significant?even in this urban watershed. The street residue contribution was 
consistently around 12% regardless of seasonal variation. Sediment storage on streets 
has a maximum capacity. A dilution effect on the percent contributed from the streets 
was evident for those storm events that generated more stream discharge. 
 
This research also examined the use of radionuclides as sediment tracers. My results 
show that Cs-137 is not suitable for suspended sediment identification in this 
watershed. This is because of the rate of land use changes in the watershed. Cs-137 
revealed patterns of erosion and storage moving downstream (Chapter 2). In the 
headwaters there was evidence of erosion while the mid-reaches of the streams 
showed erosion and sediment storage. The downstream reaches of the streams 
 64 
 
 
showed erosion but little storage. This means that the sediment being eroded from the 
headwaters is not the same that passes through the gauge station on the Northeast 
Branch. Rather, it is sediment eroded from the mid-reaches that continues 
downstream because of a lack of sediment storage. 
Connections to Other Research 
Very few other studies have examined sources of suspended sediment in a dominantly 
urban watershed. One related study (Carter et al., 2003), examined sewage treatment 
waste solids and ?road dust? as sources of sediment in the Rivers Aire and Calder 
watershed in central England. This watershed has stable land use patterns, well-
identified land uses (cultivated, uncultivated and woodland), and good water quality 
upstream. The study area is not comparable to common urban/suburban watersheds in 
much of the United States. My research shows that the methods are applicable to a 
more common watershed environment in the Northeast of the United States. To 
successfully apply those methods, I used spatially defined source groups rather than 
land use. 
Future Research 
Future research could compare the current trace element concentrations to historic 
Chesapeake Bay sediment depth profiles (pre-1770) to illustrate anthropogenic 
differences specifically for this region rather than using the mean composition of the 
Earth?s upper continental crust. Differences in the methods used to measure trace 
elements will have to be controlled. 
 
Historic deposition of tracers could also be addressed by partitioning the source 
samples into various depths and tracking historical movement of sediment. Other 
analyses could track the movement of metals more clearly by partitioning the organic 
matter into a separate fraction for analysis. Bonding to organic material by metals 
varies significantly depending on the speciation of the metals and other elements 
present. Additional research should tease apart how much of these tracers are bound 
to organics or sediment. 
 
There may also be differences in sorption of the tracers to sediment associated with 
the fluctuating pH in this natural environment. However, the tracers used for the 
sediment mixing model were not ones that show a strong dependence upon pH within 
the range of pH values found in natural and urban stream channels. In this analysis, I 
dissolved the sediment rather than using multiple leachates; the enrichment ratio 
analysis suggests that meal contamination was overprinted onto much of the 
sediment. Partial dissolution of the sediment could be conducted to determine this.  
 
Because atmospheric deposition appeared to be an end-member composition, further 
analyses should measure the composition of atmospheric particulate matter deposited 
in the watershed and the composition of sewage. The components of fertilizers used 
on urban lawns and on the farms at Beltsville Agricultural Research Center should 
 65 
 
 
also be determined. By examining what tracers co-vary with K, K-40, and P
2
O
5
, it can 
be determined how much of these tracers are anthropogenic. 
 
Finally and most importantly, researchers should address travel times by sampling 
over the course of the hydrograph using a flow-integrated sampler (Phillips et al., 
2000).Such as sampler, placed into the stream at the beginning of a storm event, and 
would collect sediment over the entire storm event. 
Conclusion 
This research began with a concern for sediment as a pollutant in the Chesapeake 
Bay. On a mass basis most sediment is originating from stream banks. However, I 
found that the most contaminated sediment (e.g., Pb, Hg, and Cd) was coming from 
atmospheric particulate matter deposited onto roadways, parking lots and other 
impervious surfaces, which are then washed into streams. 
 
Land in urban watersheds in the Northeast United States increasingly is covered with 
roadways, parking lots, and other impervious surfaces. Any policy that seeks to 
mitigate water-born pollution from such lands must recognize the need for soils to 
dilute atmospherically deposited material rather than allow it to wash into rivers and 
streams. 
 
Overall, the sediment fingerprinting method is useful in an urban watershed with 
modifications. Since the physical, chemical, and biological processes in soils and 
streams are integrated by our watersheds, these systems continue to be worth studying 
to discover the sources of sediment as well as the sources of the most contaminated 
sediment. 
 66 
 
 
Appendix A: Soil Descriptions 
 
Paint Branch 
Site P1 (Fairland): 
Date of description:  3/3/06 
Slope: complex concave 
Vegetation/Land Cover:  Intermixed hardwood and shrub near stream. Ball fields on 
summit.  
Bank Height:  87 cm, of which 3 cm was submerged by the stream 
 
Site P1. Undercutting of bank and hanging roots gave evidence of active erosion. The 
channel was narrow (~3 feet). 
Horizon Depth 
(cm) 
Texture Mineralogy Concentrations/ 
Depletions: Quantity  
and Contrast 
Roots 
 
Comments 
 
B/A 87 Sandy 
Loam 
micaceous Concentrations: Iron 
masses along old root 
channels, common, 
distinct.  
Depletions: prominent 
common 
Few Unconsolidated 
material. 2% 
rock frags, few, 
large.  
 
   
Site P1 stream and bank 
 
Site P2 (Powder Mill Road): 
Date of description:  3/3/06 
Slope: simple, linear. Very steep incline to terrace or summit 
position 
Vegetation/Land Cover:  Intermixed hardwoods and conifers  
Bank height: 200 cm, of which 16 cm is submerged by the stream. 
 67 
 
 
 
Site P2. The stream channel was greater than 6 meters and was a gravel/boulder 
bottom.  
Horizon Depth 
in cm 
Texture Mineralogy Concentrations/ 
Depletions: Quantity  
and Contrast:  
Roots 
 
Comments 
 
B/A 200 loam Less mica 
than 
upstream, 
but still 
present 
none Many
roots 
20% 
cobbles 10 
cm or 
larger in 
diameter 
 
  
Site P2 stream and bank 
 
Site P3, P4, Pdump: 
Currently under construction; access to bank is limited. Past observation indicates that 
the high bank material is similar to Site I3 with a layer of ironstone causing a perched 
water table. 
 
  
Site P3 and P4 (both banks) of stream during 3/06 renovation 
 
Little Paint Branch 
Site L1 (near residential area on golf courses, <10 years old): 
Date of description:  3/3/06 
 68 
 
 
Slope  simple concave 
Vegetation/Land Cover: swamp-trees, shrubs  
Bank Height:  162 cm, of which 5 cm was submerged by the stream. 
 
Site L1. Undercutting of bank and hanging roots gave evidence of active erosion. 
Gravel bottom stream with moderate channel width (~4 meters). 
Horizon Depth 
in cm 
Boundary Texture Mineralogy Concentrations/ 
Depletions: 
Quantity and 
Contrast 
Roots 
 
A 142 Gradual Silt loam Micaceous 
 
None common
Bw 157  Silt loam 
but 
increasing 
sands 
Mixed  common
 
    
Site L1 stream and bank 
 
Site L2 (Cherry Hill Road): 
Date of description:  3/3/06 
Slope: simple, linear 
Vegetation/Land Cover: Intermixed hardwoods and conifers 
Bank height: 184 cm, of which 10 cm is submerged by the stream. Bank 
slopes slightly back to the floodplain  
 
 69 
 
 
Site L2. Hanging roots gives evidence of active erosion. At other locations along this 
reach of the stream there is an Ab horizon.  
Horizon Depth 
in cm 
Boundary 
 
Texture Mineralogy Concentrations/ 
Depletions: 
Quantity 
And Contrast 
Roots Comment 
C1 15 Abrupt Sandy 
loam 
(clay=16%)
mixed None few  
C2 40 Abrupt  mixed   cobble 
line 
2Bt 105 Abrupt Silty clay 
loam  
mixed Common, distinct 
iron concentrations 
  
 
   
Site L2 stream and bank. Note cobble line in bank ? from top of ground 
 
Indian Creek 
Site I3 (still a farm): 
Date of description:  3/3/06 
Slope: complex concave.  
Vegetation/Land Cover  Intermixed hardwoods and conifers along stream edge. 
Crop uplands.  
Bank height: 180 cm, of which 9 cm is submerged by the stream. 
 
 70 
 
 
Site I3. This site shows the Arundel Clay from the Cretaceous Period.  
Horizon Depth 
in cm 
Boundary 
 
Texture Mineralogy Concentrations/ 
Depletions: 
Quantity and  
Contrast 
Roots Comments 
A 40 Abrupt Loamy 
coarse 
sand 
mixed None many  
C1 85 Abrupt Loamy 
coarse 
sand 
mixed    
C2 135 Abrupt Silty 
clay 
loam 
mixed   yellow-
brown/white 
striated or 
banded 
C3 185  Silty 
clay 
loam 
mixed   reddish 
 
a)   b)  
 
c)  d)  
Site I3 stream (a) and bank (b), close up of shadowed upper bank (c), and 
close up of lower bank (d). 
 
Site I2 (Black Angus farm): 
Date of description:  3/3/06 
Slope: simple concave.  
Vegetation/Land Cover: Intermixed hardwoods and conifers near stream, crop on 
terrace.  
Bank height: 125 cm, of which 1 cm is submerged by the stream. 
 71 
 
 
 
Site I2. Cows have access to the stream and cut into the banks. In addition, the 
October 8, 2005 storm incised the bank approximately 0.5 meter. 
Horizon Depth 
in cm 
Boundary 
 
Texture Mineralogy Concentrations/ 
Depletions: 
Quantity and 
Contrast 
Roots 
 
A 60 Gradual Silt loam 
(clay=7%) 
mixed None  
Bt 124  Silt loam 
(Clay=15%)
mixed   
 
  
Site I2 stream and bank 
 
Site I4 (industrial area): 
Date of description:  3/3/06 
Slope: buildings and parking lots on what may have once been a 
floodplain.  
Vegetation/Land Cover: Urban?cement with lots of garbage 
Bank height: 120 cm, of which 1 cm is submerged by the stream. Bank 
slopes away from stream.  
 
Site I4. The stream has a gravel bottom with bars and the channel is approximately 4 
meters. I noted that there are tags tied to the few shrubs that say ?wetlands?.  
Horizon Depth 
in cm 
Boundary 
 
Texture Mineralogy Concentrations/ 
Depletions: 
Quantity and 
Contrast:  
Roots 
 
Comments 
 
Bw 120  loamy 
sand 
 
mixed None None  All very 
recent fluvial 
deposit. 
Terraced  
 
 
 72 
 
 
   
Site I4 stream and bank 
 
Site I5 (Berwyn Heights): 
Date of description: 3/4/06 
Slope: simple and concave.  
Vegetation/Land Cover; Stream banks are shrubs with some hardwoods. Terrace is a 
playground. Upslope are houses.  
Bank height: 100 cm of which 12 cm is submerged by the stream. 
 
Site I5. On opposite side of stream there are some gravel deposits. The steam is 
approximately 20 feet wide and the bottom is gravelly.  
Horizon Depth 
in cm 
Boundary 
Abrupt, 
clear, 
gradual 
Texture Mineralogy Concentrations 
Quantity 
Contrast:  
Roots 
 
Bw 88  Silt loam 
(clay=8%)
Some mica 
present 
None  Few 
 
a)   b)  c)  
Site I5 stream (a), close-up of opposite bank (b), and bank described above (c). Note 
that the tire was deposited in the bank, and is now being eroded and exposed. 
 
Beaverdam Creek 
Site B1 (OPEIII): 
Date of description: 3/4/06 
Slope: simple concave 
Vegetation/Land Cover: swamp-trees, shrubs and corn upslope 
Bank height: 30 cm, of which 5 cm is submerged by the stream. In some 
places, the bank height is less than 5 cm.  
 73 
 
 
 
Site B1 
Horizon Depth 
in cm 
Boundary 
 
Texture Mineralogy Concentrations/ 
Depletions: 
Quantity and 
Contrast:  
Roots 
 
Bt 10 Clear Clay 
loam 
siliceous Concentrations: 
Few, distinct 
iron masses. No 
depletions 
few 
Btg 25  Clay 
loam 
siliceous Concentrations: 
Few, distinct 
iron masses. 
Depletions: 
Matrix color is 
gleyed 
 
 
 
Site B1 floodplain auger boring. No available picture of bank 
described above. 
 
Site B3 (Plant Materials Center): 
Date of description: 3/4/06 
Slope: complex concave 
Vegetation/Land Cover: swamp-trees, shrubs and hay upslope 
Bank height: 150 cm, of which 56 cm is submerged by the stream.  
 
 74 
 
 
Site B3. In many places, there is an exposed buried A. There are bars and the stream 
bottom is sand. Many fallen logs criss-cross the stream. Active incision evident.  
Horizon Depth 
in cm 
Boundary 
 
Texture Mineralogy Concentrations/ 
Depletions: 
Quantity and 
Contrast:  
Roots 
 
A 25 Clear Silt loam 
(Clay=7%) 
mixed  common
Bt 40 Clear Loam 
(Clay=15%)
mixed Many distinct 
iron spherical 
concentrations. 
No depletions 
 
Ab 60 Gradual Loam 
(Clay=22%)
mixed Many distinct 
iron spherical 
concentrations. 
No depletions 
 
Bw 94 Gradual Sandy loam Siliceous   
 
 
Site B3 stream. Note buried A horizon. 
 
Site B4 (marsh on Research Road) 
Date of description: 3/6/06 
Slope: simple linear 
Vegetation/Land Cover: Shrubs. There are some hardwoods upslope. 
Bank height: 54 cm, of which 14 cm is submerged by the stream.  
 
 75 
 
 
Site B4 
Horizon Depth 
in cm 
Boundary 
 
Texture Mineralogy Concentrations/ 
Depletions: 
Quantity and 
Contrast 
Roots 
 
Bw/C 40  Sandy clay 
loam 
(clay=24%)
mixed Common 
distinct spherical 
iron 
concentrations. 
Gleyed matrix at 
water?s edge 
many 
 
   
Site B4 stream and close-up of redoximorphic features in the bank 
 
Site B2 (Mandan Road at BW Pkwy): 
Date of description: 3/4/06 
Slope: Densely residential and very close to road. Unable to 
determine slope. 
Vegetation/Land Cover: swamp-trees, shrubs 
Bank height: 35 cm, of which 3 cm is submerged by the stream.  
 
Site B2. Silty bottom stream that is about 0.5 meter across at a maximum. 
Horizon Depth 
in cm 
Boundary 
 
Texture Mineralogy Concentrations/ 
Depletions: 
Quantity and 
Contrast:  
Roots 
 
C 1 Abrupt Sand siliceous none  
Bw 27 Clear Silt 
loam 
(15% 
clay) 
mixed   
Bt 30 Clear Clay 
loam 
(30% 
clay) 
mixed   
 76 
 
 
 
 
Site B2 stream  
 
Northeast Branch of the Anacostia River 
Site PM5 (just below confluence of Indian and Beaverdam Creeks): 
Date of description: 3/6/06 
Slope: simple linear 
Vegetation/Land Cover; Stream banks are intermixed hardwoods and conifers. 
Upslope is urbanized.  
Bank height: 150 cm, of which 10 cm is submerged by the stream. 
 
Site PM5 
Horizon Depth 
in cm 
Boundary 
 
Texture Mineralogy Concentrations/ 
Depletions: 
Quantity and 
Contrast 
Roots 
 
Comments 
 
C 140 clear Sandy 
loam 
mixed none many 1 cm 
diameter 
mica flakes 
Ab? 150 Possible buried A at waters edge. Unable to determine source of organic 
material in the horizon because it was under water.  
 
 77 
 
 
a) b) 
Site PM5 stream. 
 
Site PM6 (near 94
th
 Aero squadron): 
Date of description: 3/6/06 
Slope: simple concave.  
Vegetation/Land Cover; Stream banks are shrubs with some hardwoods. Upslope is 
urban.  
Bank height: 156 cm, of which 95cm is submerged by the stream. 
 
Site PM6. Just above this is a berm to slow velocity. Stream has a gravel bottom 
interspersed with large granite rocks that the stream has moved. The rocks evidently 
came from an attempt at armoring the banks. There are hanging roots that indicate 
active erosion of the channel.  
Horizon Depth 
in cm 
Boundary 
Abrupt, 
clear, 
gradual 
Texture Mineralogy Concentrations/ 
Depletions: 
Quantity and 
Contrast 
Roots 
Few 
common  
Comments 
 
C 2 Abrupt Sandy Lots of
mica and 
sand. 
none   
A/B 59  Sandy 
loam 
(clay=13%)
mixed  many  
 
 
Site PM7 (near gauge): 
Date of description: 3/6/06 
Landscape: there is a floodplain approximately 2.5 meters wide. The 
terrace is a park and playground. 
Vegetation/Land Cover: Urban.  
Bank height: 110 cm of which 5 cm is submerged by the stream. 
 78 
 
 
 
Site PM7, gauge site. Heavy plastering of sediment, branches up to an inch in 
diameter, leaves oriented vertically on bank, and garbage such as plastic bags. Gravel 
bottom bed with lots of algae. Algal growth probably supported by sewage overflows. 
Just above this is a berm to slow velocity. 
Horizon Depth 
in cm 
Boundary 
 
Texture Mineralogy Concentrations/ 
Depletions: 
Quantity and 
Contrast:  
Roots 
 
C 105  Loamy 
sand 
siliceous none common
 
 
  
 
Site PM7 stream and bank 
 79 
 
 
Appendix B: Distribution of Sediment Source Samples 
 
Source Type Landscape 
Position 
Land Use No. of 
Samples 
Upland Urban 1 
 Rural 1
Forested 1
Floodplain Urban 0 
 Rural 1
Forested 1
Bank Urban 11 
Rural 3
Forested 2
Point Bar Urban 3 
 Rural 0 
Coastal Plain Physiographic Region 
Forested 1
Upland Urban 1 
 Rural 0
Forested 1
Floodplain Urban 0 
 Rural 0
Forested 1
Bank Urban 3 
Rural 0
Forested 1
Point Bar Urban 0 
 Rural 0 
Piedmont Physiographic Region 
Forested 0
Storm Events (same sample site, sampled 
over time) 
8
Construction site   1 
Street Residue (6 sites aggregated, 
sampled over time) 
2 
Total 43
 
 80 
 
 
Appendix C: Raw Data 
 
 
 
 81 
 
 
 
 
 82 
 
 
 
 
 83 
 
 
 
 
 
 84 
 
 
Appendix D: Detection Limits 
Detection limits for trace elements 
 
Detection Limits for Radionuclides 
Cs-137  -  1.0 Bq kg
-1 
K-40  -  10.0 Bq kg
-1
 85 
 
 
Appendix E: Sedigraphs and Hydrographs for 
Suspended Sediment Sample Collection Times 
 
October 7 - 8, 2005
0
100
200
300
400
500
2.
75
5
.
7
5
8
.
7
5
1
1.
75
1
4.
75
17
.
7
5
20
.
7
5
2
3.
75
26
.
7
5
2
9.
75
3
2.
75
35.
75
38
.
7
5
41
.
7
5
4
4.
75
4
7.
75
Tu
r
b
idity
0
10
20
30
40
50
60
Q (m
3
 s
-1
)
 
October 21-23, 2005
0
20
40
60
80
100
120
2
.
75
5
.
75
8
.
75
11.
75
14.
75
17.
7
5
20.
75
23.
75
26.
75
29.
75
32.
75
35.
75
38.
75
41.
75
44.
7
5
4
7.
75
50.
75
53.
75
56.
75
59.
75
62.
75
65.
7
5
68.
75
71.
7
5
Tu
r
b
idit
y
0
0.5
1
1.5
2
2.5
Q (
m
3
 s
-1
)
 
 
December 25-26, 2005
0
20
40
60
80
100
120
140
160
2.
7
5
5
.7
5
8.
7
5
1
1.
7
5
1
4.
7
5
1
7
.7
5
2
0.
7
5
2
3.
7
5
2
6
.7
5
2
9
.7
5
3
2.
7
5
3
5.
7
5
3
8
.7
5
4
1.
7
5
4
4.
7
5
4
7.
7
5
Tu
rb
i
d
i
t
y
0
0.5
1
1.5
2
2.5
3
3.5
4
Q (m
3
 s
-1
)
 
January 17-19, 2006
0
2
4
6
8
10
12
0 6 12 18 24 30 36 42 48 54 60 66 72
Q (
c
m
s
)
 
 
February 3-5, 2006
0
1
2
3
4
5
6
7
8
0 6 12 18 24 30 36 42 48 54 60 66 72
Q (
c
m
s
)
 
April 22-23, 2006
0
1
2
3
4
5
6
7
8
9
10
0 6 12 18 24 30 36 42 48
Q (
c
m
s
)
 
 
May 11-12, 2006
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
0 6 12 18 24 30 36 42 48
Q (
c
m
s
)
 
June 24-26, 2006
0
50
100
150
200
250
300
350
400
0 6 12 18 24 30 36 42 48 54 60 66 72
Q (
c
m
s
)
 
 
The x-axis is hours. Note the y-axis varies depending on storm size. The dashed line 
is turbidity measured as monochrome near infra-red LED light, 780-900 nm, 
detection angle 90 +/ -2.5 degrees. The solid line is discharge.  
 86 
 
 
Appendix F: Particle-Size Distributions 
 
 
West Subwatershed
0
10
20
30
40
50
60
70
80
90
100
0.1 1 10 100 1000
Diameter (um)
%
 Fi
ner
 
Cu
m
u
l
a
t
i
v
e
0
2
4
6
8
10
12
14
Abank Undersize (%) ABank2 Undersize (%) ABank3 Undersize (%) ABank4 Undersize (%)
A Point Bar Undersize (%) Cbank Undersize (%) Cfloodplain Undersize (%) C Upland Undersize (%)
F bankUndersize (%)  G Bank Undersize (%) P Bank Undersize (%) P Upland Undersize (%)
U Floodplain Undersize (%) Abank Frequency (%) ABank2 Frequency (%) ABank3 Frequency (%)
ABank4 Frequency (%) A Point Bar Frequency (%) Cbank Frequency (%) Cfloodplain Frequency (%)
C Upland Frequency (%) F bankFrequency (%)  G Bank Frequency (%) P Bank Frequency (%)
P Upland Frequency (%) U Floodplain Frequency (%)
 
 
 
 
East Subwatershed
0
10
20
30
40
50
60
70
80
90
100
0.1 1 10 100 1000
Diameter (um)
%
 Finer
,
 C
u
mulat
i
ve
0
2
4
6
8
10
12
14
E Bank Undersize (%) I Upland Undersize (%) H Bank Undersize (%) I Bank Undersize (%)
I Floodplain Undersize (%) L Bank Undersize (%) M Bank Undersize (%) N Bank Undersize (%)
O Bank Undersize (%) R Bank Undersize (%) O Floodplain Undersize (%) O Upland Undersize (%)
Q Bank Undersize (%) Q Floodplain Undersize (%) Q Point Bar Undersize (%) Q Upland Undersize (%)
Z Upland Undersize (%) V Upland Undersize (%) X Upland Undersize (%) E Bank Frequency (%)
I Upland Frequency (%) H Bank Frequency (%) I Bank Frequency (%) I Floodplain Frequency (%)
L Bank Frequency (%) M Bank Frequency (%) N Bank Frequency (%) O Bank Frequency (%)
R Bank Frequency (%) O Floodplain Frequency (%) O Upland Frequency (%) Q Bank Frequency (%)
Q Floodplain Frequency (%) Q Point Bar Frequency (%) Q Upland Frequency (%) Z Upland Frequency (%)
V Upland Frequency (%) X Upland Frequency (%)
 
 87 
 
 
 
Suspended Sediment
0
10
20
30
40
50
60
70
80
90
100
0.1 10 1000
Diameter (um)
% F
i
n
e
r
,
 
Cu
m
u
l
a
t
i
v
e
0
2
4
6
8
10
12
14
12/25/05 Undersize (%) 10/22/05 Undersize (%) 10/7/05 Undersize (%) 5/13/06 Undersize (%)
4/23/06 Undersize (%) 2/4/06 Undersize (%) 1/18/06 Undersize (%) 12/25/05 Frequency (%)
10/22/05 Frequency (%) 10/7/05 Frequency (%) 5/13/06 Frequency (%) 4/23/06 Frequency (%)
2/4/06 Frequency (%) 1/18/06 Frequency (%)
 
 
 88 
 
 
Appendix G: pH data 
 
 
Source: USGS, Maryland Water Science Center, USGS 01649500 North East Branch 
Anacostia River at Riverdale, MD, http://waterdata.usgs.gov/md/nwis/.  
 89 
 
 
Appendix H: Periodic Table 
 
(http://universe-review.ca/index.htm, 2006) 
 90 
 
 
References 
Almeida, S.M., C.A. Pio, M.C. Freitas, M.A. Reis, and M.A. Trancoso. 2005. Source 
apportionment of fine and coarse particulate matter in a sub-urban area at the 
western European coast. Atmospheric Environment 39:3127-3138. 
Boruvka, L., O. Vacek, and J. Jehlicka. 2005. Principal component analysis as a tool 
to indicate the origin of potentially toxic elements in soils. Geoderma 
128:289-300. 
Brady, N.C., R.R. Weil. 2002. The nature and property of soils. 13
th
 Ed. Prentice 
Hall. New Jersey. 
Breault, R.F., K.P. Smith, and J.R. Sorenson. 2005. Residential street-dirt 
accumulation rates and chemical composition, and removal efficiencies by 
mechanical- and vacuum-type sweepers, New Bedford, Massachusetts, 2003-
04. U.S. Geological Survey Scientific Investigations Report 2005-5184. 
Burton, G. A., Jr. and R. Pitt. 2002. Stormwater effects handbook: a toolbox for 
watershed managers, scientists, and engineers. Available on-line. ISBN 0-
87371-924-7. Lewis Publishers. 
Callender, E. 2004. Heavy Metals in the Environment--Historical Trends, pp. 67-105. 
In Environmental Geochemistry (ed. B.S. Lollar) Vol. 9 Treatise on 
Geochemistry (eds. H.D. Holland and K.K. Turekian), Elsevier-Pergamon, 
Oxford. 
Carter, J., P.N. Owens, D.E. Walling, and G.J.L. Leeks. 2003. Fingerprinting 
suspended sediment sources in a large urban river system. Sci. Total Environ. 
314:513-534.  
Cleaves, E.T., J. Edwards, Jr., and J.D. Glaser. 1968. Geologic map of Maryland: 
Maryland Geological Survey, 1 sheet, scale 1:250,000.  
Collins, A.L., D.E. Walling, and G.J.L. Leeks. 1997a. Sediment sources in the upper 
Severn catchment: A fingerprinting approach. Hydrology and Earth Systems 
1(3):509-521. 
Collins, A.L., D.E. Walling, and G.J.L. Leeks. 1997b. Use of the geochemical record 
preserved in floodplain deposits to reconstruct recent changes in river basin 
sediment sources. Geomorphology 19:151-167. 
Csato, I., C. Plank, M.O. Hayes, J. Michel, and Z. Nixon. 2005. Characterization and 
Ground Water Flow in the Poplar Point Area, Anacostia River Basin, 
Washington, D.C. (online). Available at 
http://strata.geol.sc.edu/Istvan/Anacostia/Anacostia.html Research Planning, 
Inc., Columbia, South Carolina. University of South Carolina - Geology 
Department. 
Deboer, D.H., and G. Crosby. 1995. Evaluating the potential of sem/eds analysis for 
fingerprinting suspended sediment derived from 2 contrasting topsoils. Catena 
24:243-258. 
Duan, F.K., K.B. He, Y.L. Ma, F.M. Yang, X.C. Yu, S.H. Cadle, T. Chan, and P.A. 
Mulawa. 2006. Concentration and chemical characteristics of pm2.5 in 
Beijing, china: 2001-2002. Sci. Total Environ. 355:264-275. 
 91 
 
 
Dunne, T., L.B. Leopold. 1978. Water in Environmental Planning. W.H. Freeman and 
Company, New York. 
Environmental Protection Agency. 2006. Lead Gasoline (on-line). Available at 
http://www.epa.gov/history/topics/lead/03.htm 
Environmental Protection Agency. 2006. Lead Poisoning (on-line). Available at 
http://www.epa.gov/region02/health/leadpoisoning.htm. 
Environmental Protection Agency. 2006. Chesapeake 2000 (on-line). Available at 
http://www.chesapeakebay.net/agreement.htm 
Gao, B., M.T. Walter, T.S. Steenhuis, J.Y. Parlange, B.K. Richards, W.L. Hogarth, 
and C.W. Rose. 2005. Investigating raindrop effects on transport of sediment 
and non-sorbed chemicals from soil to surface runoff. J. Hydrol. 308:313-320. 
Gao, Y., E.D. Nelson, M.P. Field, Q. Ding, H. Li, R.M. Sherrell, C.L. Gigliotti, D.A. 
Van Ry, T.R. Glenn, and S.J. Eisenreich. 2002. Characterization of 
atmospheric trace elements on pm2.5 particulate matter over the New York-
new jersey harbor estuary. Atmospheric Environment 36:1077-1086. 
Gellis, A.C. and Landwehr, J.M.. 2006. Identifying sources of fine-grained 
suspended-sediment in the Pocomoke River, an Eastern Shore tributary to the 
Chesapeake Bay: Proceedings of the Joint Federal Interagency Conference 
2006, 8th Federal Interagency Sedimentation Conference, April 2-6, 2006, 
Reno, NV, Paper 5C-1 in CD_ROM file ISBN 0-9779007-1-1. 
Glaser, John. 1996. Prince George's County Geologic Map With Distribution of Silt 
Loam Soils in Upland Deposits by Hack (1977) 
Gruszowski, K.E., I.D.L. Foster, J.A. Lees, and S.M. Charlesworth. 2003. Sediment 
sources and transport pathways in a rural catchment, Herefordshire, UK. 
Hydrological Processes 17:2665-2681. 
Hack, J. T.. 1977. Geologic map for land-use planning, Prince Georges County, 
Maryland: U.S. Geological Survey. 
Impellitteri, C.A., Y. Lu, J.K. Saxe, H.E. Allen, W.J.G.M. Peijnenburg. 2002. 
Correlation of the partitioning of dissolved organic matter fractions with the 
desorption of Cd, Cu, Ni, Pb and Zn from 18 Dutch soils. Environment 
International 28:401-410. 
He, Q., P.O. Owens. 1995. Determination of suspended sediment provenance using 
caesium-137, unsupported lead-210 and radium-226: A numerical mixing 
model approach. In: Foster I.D.L., et al., ed. Sediment and water quality in 
river catchments. Chichester, UK: Wiley, 1995:207-228.  
Jacobson, R.B., and D.J. Coleman. 1986. Stratigraphy and recent evolution of 
Maryland piedmont flood plains. American Journal of Science 286:617-637. 
Jansson, M.B. 2002. Determining sediment source areas in a tropical river basin, 
Costa Rica. Catena 47:63-84. 
Jantz, P., S. Goetz, C. Jantz. 2005. Urbanization and the loss of resource lands in the 
Chesapeake Bay Watershed. Environ. Management. 36:6:808-825. 
Jeelani, G., and A.Q. Shah. 2006. Geochemical characteristics of water and sediment 
from the Dal lake, Kashmir Himalaya: Constraints on weathering and 
anthropogenic activity. Environmental Geology 50:12-23. 
 92 
 
 
Jenkins, P.A., R.W. Duck, J.S. Rowan, and J. Walden. 2002. Fingerprinting of bed 
sediment in the Tay estuary, Scotland: An environmental magnetism 
approach. Hydrology and Earth System Sciences 6:1007-1016. 
Juracek, K.E., A.C. Ziegler. 2006. The legacy of leaded gasoline in bottom sediment 
of small rural reservoirs. J. Environ. Qual. 35:2092-2102.  
Konrad, C.P., D.B. Booth, and S.J. Burges. 2005. Effects of urban development in the 
Puget lowland, Washington, on interannual stream flow patterns: 
Consequences for channel form and streambed disturbance. Water Resour. 
Res. 41. 
Leung, C.M., and J.J. Jiao. 2006. Heavy metal and trace element distributions in 
groundwater in natural slopes and highly urbanized spaces in mid-levels area, 
Hong Kong. Water Res. 40:753-767. 
Markewich, H. W., M.J. Pavich, M.J. Mausbach, R.L. Hall, R.G. Johnson, and P.P. 
Hearn. 1987. Age relations between soils and geology in the coastal plain of 
Maryland and Virginia. U.S. Geological Survey Bulletin 159-A. 
Maryland Department of Natural Resources. 2005. Surf your watershed-watershed 
profile Anacostia River. 
www.dnr.state.md.us/watersheds/surf/prof/pdf.02140205-wp.pdf.  
Marvin, C.H., G.G. Leppard, M.M. West, G.A. Stern, A.R. Boden, and B.E. 
McCarry. 2004. Refined tunable methodology for characterization of 
contaminant-particle relationships in surface water. J. Environ. Qual. 33:2132-
2140. 
Matisoff, G., E.C. Bonniwell, and P.J. Whiting. 2002. Soil erosion and sediment 
sources in an Ohio watershed using beryllium-7, cesium-137, and lead-210. J. 
Environ. Qual. 31:54-61. 
Matisoff, G., C.G. Wilson, and P.J. Whiting. 2005. The be-7/pb-210(xs) ratio as an 
indicator of suspended sediment age or fraction new sediment in suspension. 
Earth Surf. Processes Landforms 30:1191-1201. 
Moglen, G.E. and R.E. Beighley. 2002. Spatially Explicit Hydrologic Modeling of 
Land Use Change. JAWRA, 38(1): 241-253. 
Motha, J.A., P.J. Wallbrink, P.B. Hairsine, and R.B. Grayson. 2002. Tracer properties 
of eroded sediment and source material. Hydrological Processes 16:1983-
2000. 
Motha, J.A., P.J. Wallbrink, P.B. Hairsine, and R.B. Grayson. 2003. Determining the 
sources of suspended sediment in a forested catchment in southeastern 
Australia. Water Resour. Res. 39. 
Nagle, G.N., and J.C. Ritchie. 2004. Wheat field erosion rates and channel bottom 
sediment sources in an intensively cropped northeastern Oregon drainage 
basin. Land Degradation & Development 15:15-26. 
National Oceanic and Atmospheric Administration. 2005. Weather forecasts (on-line) 
Available at: http://www.erh.noaa.gov. 
Nicholas, A.P., P.J. Ashworth, M.J. Kirkby, M.G. Macklin, and T. Murray. 1995. 
Sediment slugs: Large-scale fluctuations in fluvial sediment transport rates 
and storage volumes. Progress in Physical Geography 19:500-519. 
Owens, P.N.; D.E. Walling, G.J.L. Leeks. Tracing fluvial suspended sediment sources 
in the catchment of the River Tweed, Scotland, using composite fingerprints 
 93 
 
 
and a numerical mixing model In Tracers in geomorphology. (Ed) I. D. L. 
Foster. John Wiley & Sons. 291-308. 
Papanicolaou, A.N., J.F. Fox, and J. Marshall. 2003. Soil fingerprinting in the Palouse 
basin, USA using stable carbon and nitrogen isotopes. International Journal of 
Sediment Research 18:278-284. 
Phillips, J.M., M.A. Russell, and D.E. Walling. 2000. Time-integrated sampling of 
fluvial suspended sediment: A simple methodology for small catchments. 
Hydrological Processes 14:2589-2602. 
Poulsen, J., and A. French. 2006. Discriminant Function Analysis (on-line). Available 
at: http://online.sfsu.edu/~efc/classes/biol710/discrim. San Francisco State 
University. 
Prepas, E.E. and Charette, T. 2004. Worldwide Eutrophication of Water Bodies: 
Causes, Concerns, Controls, pp. 311-326. In Environmental Geochemistry 
(ed. B.S. Lollar) Vol. 9 Treatise on Geochemistry (eds. H.D. Holland and 
K.K. Turekian), Elsevier-Pergamon, Oxford. 
Rauter, A., G. Weigelhofer, J. Waringer, and T.J. Battin. 2005. Transport and 
metabolic fate of sewage particles in a recipient stream. J. Environ. Qual. 
34:1591-1599. 
Rhoton, F.E., W.E. Emmerich, D.C. Goodrich, S.N. Miller, and D.S. McChesney. 
2005. Sediment source identification using soil geomorphological 
characteristics in a semiarid watershed. Submitted to Soil Science Society of 
America Journal. 
Ritchie, J.C., J.A. Spraberr, J.R. Mchenry. Estimating Soil Erosion From 
Redistribution Of Fallout Cs-137. 1974. Soil Sci. Soc. Am. J. Vl 38 Is 1 137-
139  
Ritchie, J.C., and J.R. McHenry. 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. and C.A. Ritchie. 2006. Bibliography of Publications of 137Cesium 
Studies Related to Erosion and Sediment Deposition. 
http://hydrolab.arsusda.gov/cesium 
Rowan, J.S.; P. Goodwill; S.W. Franks. Uncertainty estimation in fingerprinting 
suspended sediment sources In Tracers in geomorphology. (Ed) I. D. L. 
Foster. John Wiley & Sons. 279-290. 
Rudnick, R.L., and S. Gao. 2003. Composition of the continental crust, p. 1-64, In R. 
L. Rudnick, ed. The crust, Vol. 3. Elsevier-Pergamon, Oxford. 
SAS Institute Inc., SAS? 9.1.2, Qualification Tools User?s Guide, Cary, NC: SAS 
Institute Inc., 2004. 
Scudlark, J.R.; K.C. Rice; K.M. Conko; O.P. Bricker; T.M. Church. 2005. 
Transmission of atmospherically derived trace elements through an 
undeveloped, forested Maryland watershed. Water, Air, and Soil Pollution 
163:53-79. 
Slattery, M.C.; T.P. Burt, J. Walden. 1995. Application of mineral magnetic 
measurements to quantify within-storm variations in suspended sediment 
sources. International Association of Hydrological Sciences. 229:143-151.  
 94 
 
 
Sun, Y.; Chi, P.; Shiue, M. 2001. Comparison of different digestion methods for total 
decomposition of siliceous and organic environmental samples. Analytical 
Sciences, 17:1395-1399. 
Tuccillo, M.E. 2006. Size fractionation of metals in runoff from residential and 
highway storm sewers. Sci. Total Environ. 355:288-300. 
U.S. Department of the Interior, U.S. Geological Survey. 1999. U.S. Geological 
Circular 1225. The Quality of Our Nation's Waters--Nutrients and Pesticides. 
Reston, Virginia. 
U.S. Geological Survey. 2004. Glossary (on-line). Available at: 
http://interactive2.usgs.gov/glossary/index.asp 
University of Maryland, Department of Meteorology. 2006. MM5 forecasts (on-line). 
Available at: http://www.atmos.umd.edu/~mm5. 
Vanoni, V.A., (ed.) 1975. Sedimentation engineering. ASCE, Reston. 
Wallbrink, P.J., A.S. Murray, J.M. Olley, and L.J. Olive. 1998. Determining sources 
and transit times of suspended sediment in the Murrumbidgee river, new south 
Wales, Australia, using fallout cs-137 and pb-210. Water Resour. Res. 
34:879-887. 
Walling, D.E. 2005. Tracing suspended sediment sources in catchments and river 
systems. Sci. Total Environ. 344:159-184. 
Walling, D.E., P.N. Owens, and G.J.L. Leeks. 1999. Fingerprinting suspended 
sediment sources in the catchment of the river Ouse, Yorkshire, UK. 
Hydrological Processes 13:955-975.  
Walling, D.E., and J.C. Woodward. 1992. Use of radiometric fingerprints to derive 
information on suspended sediment sources. p. 153?164. In Erosion and 
sediment transport monitoring programmes in river basins. . Publ. 210. IAHS, 
Wallingford, UK. 
Washington Suburban Sanitary Commission. 2002. Biosolids Data (on-line). 
Available at (http://www.wsscwater.com/biosolids. 
Weather Underground. 2006. Available at http://weather.marylandweather.com/. 
Wolman, G. M. Schick, P. Asher. 1967. Effects of construction on fluvial sediment, 
urban and suburban areas of Maryland. Water Resources Res. 3:2:451-464. 
Woodward, J., and I. Foster. 1997. Erosion and suspended sediment transfer in river 
catchments: Environment controls, processes and problems. Geography 
82:353-376. 
 95