Big Harris - Sediment History and Sourcing - Miller et al., 2019

Hydrologic, geomorphic, and stratigraphic controls on suspended
sediment transport dynamics, Big Harris Creek restoration site,
North Carolina, USA
Jerry R. Miller*, Kelly Ferri, David Grow, Lionel Villarroel
Department of Geosciences and Natural Resources, Western Carolina University, Cullowhee, NC, 28723, USA
A R T I C L E I N F O
Article history:
Received 10 May 2018
Received in revised form 6 December 2018
Accepted 23 December 2018
Available online 3 January 2019
Keywords:
Suspended sediment transport
River restoration
Legacy sediments
Geomorphic response
A B S T R A C T
Water quality monitoring records were combined with geomorphic and stratigraphic data to determine
the controls on suspended sediment transport dynamics within the headwaters of Big Harris Creek, the
site of one of the largest stream restoration projects in North Carolina, USA. Land-use change associated
with European settlement resulted in spatially variable geomorphic responses that produced reaches
possessing semi-homogeneous landforms and processes, referred to as process zones. Downstream,
process zones were dominated by entrenched channels possessing banks characterized by sandy post-
settlement deposits that overlie finer-grained pre-settlement deposits. Spatial variations in the resulting
process zones strongly influenced modern suspended sediment concentrations (SSC) and loads in the
catchment, which are among the highest reported for the southeastern United States. The source and
transport dynamics of suspended sediment differed between low magnitude floods (characterized by
minimal changes in water level elevations) and moderate to high magnitude floods. Low magnitude
floods were characterized by SSCs that varied over several orders of magnitude, and that were unrelated
to flow conditions. Precipitation, during these events, rapidly mobilization fine-grained pre-settlement
deposits associated with bank failures and cattle crossings along entrenched alluvial channels. Moderate
to high magnitude floods within larger tributary basins exhibited more systematic discharge-SSC
relationships. Suspended sediment transport was dominated by sand-sized particles derived from post-
settlement legacy sediments eroded from the channel banks and headwater gullies. The observed
temporal and spatial differences in SSC between low and moderate to high magnitude floods complicates
the quantification of water quality, and shows that comparisons of water quality before and following the
implementation of restoration projects need to differentiate between distinct populations of suspended
sediment transport.
© 2019 Elsevier Ltd. All rights reserved.
1. Introduction
During the past decade, a plethora of definitions have been put
forth for the practice of ‘stream restoration’, hindering its use as a
specific descriptor of the management strategies that are applied.
Nonetheless, inherent in most definitions is an attempt to improve
aquatic and/or riparian ecosystem function and health from an
ecological perspective (see, for example, Hobbs and Norton, 1996).
Expenditures on restoration projects in the U.S. now exceed
approximately one billion dollars per year (Bernhardt et al., 2005).
These projects are most frequently intended to improve water
quality, reduce bank erosion and channel instability, enhance
selected biotic species, and hydrologically reconnect the channel to
its floodplain (i.e., the riparian zone). With regards to water quality,
a common goal is to reduce suspended sediment loads (Miller and
Kochel, 2008, 2013), which often threaten aquatic ecosystems
(Wood and Armitage, 1997; Armstrong et al., 2003; Syvitski et al.,
2005; Kemp et al., 2011; Grabowski et al., 2014). In fact, according
to the National Water Quality Inventory, a program developed to
assess the current condition of the nation’s water resources,
sediment is second only to pathogens (e.g., E. coli) as the leading
Abbreviations: BH1, montoring station at mouth of basin on Big Harris Creek;
BH2, upstream monitoring station on Big Harris Creek; BHT1, monitoring station at
mouth of Royster Creek; BHT2, monitoring station near confluence of Fletcher and
Stick Elliott creeks; cms, cubic meters per second; DOC, dissolved organic carbon;
HrIt, hourly rainfall intensity; NC, North Carolina; NRCS, United States National
Resource Conservation Service; SSC, suspended sediment concentration; SSCmax,
maximum event suspended sediment concentration; Tprec, total precipitation of an
event; USGS, United States Geological Survey.
* Corresponding author at: Department of Geosciences and Natural Resources,
Western Carolina University, Stillwell 302, Cullowhee, NC, 28723, USA.
E-mail address: [e-mail] (J.R. Miller).
https://doi.org/10.1016/j.ancene.2018.12.002
2213-3054/© 2019 Elsevier Ltd. All rights reserved.
Anthropocene 25 (2019) 100188
Contents lists available at ScienceDirect
Anthropocene
journal homepage: www.else vie r.com/locate /ance ne
cause of river impairment (USEPA, 2013). Moreover, the annual
cost of suspended sediment influx to riverine ecosystems is
estimated to range from $20 to $50 billion in North America alone
(Pimentel et al., 1995; Mukundan et al., 2012; Bain et al., 2012).
The effectiveness of river restoration to address the “sediment
problem” is strongly correlated to the degree with which the
sources, transport and redeposition of suspended materials are
understood within the basin (Vercruysse et al., 2017). Unfortu-
nately, such information is lacking for an overwhelming majority of
sites and regions where restoration is conducted. Data pertaining
to small headwater basins are particularly lacking, even though
they account for more than 70% of total stream length in the
continental United States (Leopold et al., 1964).
The primary objective of this investigation was to combine an
extensive set of monitoring data with geomorphic and strati-
graphic data to document the sources of suspended sediment,
and the controls on suspended sediment transport, within the
headwaters of Big Harris Creek over a range of temporal and
spatial scales. Specific research questions include: (1) What were
the variations in suspended sediment concentration during and
between flood events? (2) How did suspended sediment
concentrations and loads vary spatially within the basin? (3)
What were the responses of the system to anthropogenic land-
use changes that affected the basin near the turn of 20th century?
(4) How did spatial differences in channel response, and the
produced alluvial stratigraphic units, influence the contemporary
influx of sediment to and through the river? The combined
analysis provides insights into the assessment of water quality
improvements associated with restoration projects, and the
controls on suspended sediment transport dynamics that are
not only applicable to Big Harris Creek, but other headwater
streams.
2. Methods
2.1. Study area
The study area includes the upper 9.6 km2 of the Big Harris
Creek watershed located within the piedmont physiographic
province, approximately 2 km from the town of Polkville, North
Carolina, USA (Fig. 1). The site represents one of the largest and
most expensive stream restoration projects in North Carolina.
Land-use/land-cover within the project area is currently domi-
nated by forest (~48%) and grassed/pastured areas (~44%). The
remainder of the land-use/land-cover types include turkey
farming facilities, homes, roads, and row crops (Fig. 1). Both
the axial channel of Big Harris Creek and its major tributaries
were incised (prior to restoration) into alluvial valley floor
deposits. Incision varied along the drainage network, and ranged
from approximately 1.5 to more than 4 m. Prior to restoration,
cattle were grazed in many of the areas covered in pasture, and
locally crossed stream channels where they trampled bank
vegetation and bank materials. River restoration was initiated, in
part, to reduce the rates of bank erosion, improve water quality,
and hydrologically reconnect selected channel reaches with its
floodplain.
2.2. Description of monitoring sites
Four sites within the Big Harris Creek project area were selected
and instrumented between August and November 2012, and
monitored through April 2017 (Table 1). These four sites were
selected to characterize the total sediment load and yield from the
basin (BH1, Fig. 1), as well as spatial differences in sediment
concentrations, loads, and yields between subbasins in the project
area (BH2, BHT1, BHT2). Each site was instrumented with a HOBO
levellogger to record fluctuations in water levels (and discharge),
and an ISCO automated water sampler. In addition, passive (U59B)
siphon samplers were installed and maintained by the NC Division
of Water Quality from December 2012 through December 2013 at
all four sites possessing ISCOs, and at an additional 11 sites to
increase the spatial resolution of the SSC data (Fig.1). At most sites,
three to four passive samplers were located at different heights
above the channel bed to collect water samples under varying flow
conditions. The passive samples did not record the timing of
sample collection, and only obtain waters during the rise stage of a
flood. Nonetheless, they provided a low-cost method of increasing
the spatial resolution of the SSC data (MacKay and Taylor, 2012).
Upon collection, water levels were converted into discharge
by (1) modeling the stage-discharge relationship at the site, (2)
fitting a statistical model to the generated stage-discharge data,
and (3) using the statistical regression model to convert water
levels to discharge. Stage-discharge relationships were modeled
using a Cross Section Hydraulic Analyzer developed by the NRCS
(Moore, 2009), and verified by comparing estimated discharge
values to discharge measurements collected during storm
events.
Water samples were collected in pre-cleaned 1000 ml bottles by
the ISCO automated samplers. The ISCOs were programmed to
start sampling at the beginning of a flood event (as determined by a
change in water level). Samples were then obtained at 30 min
intervals, regardless of flood magnitude. On occasion, samplers
failed during the sampling of a flood event. As a result, samples
were not always obtained during the entire event, or were
collected at intervals other than every 30-minutes. All ISCO and
U59B samples were analyzed for SSC using a slightly modified
version of the procedure developed by the USGS (Guy, 1977). The
detection limit was determined to be 2.5 mg/L; precision and
accuracy of the method, determined using an in-house standard,
were within +/- 5% of known SSC values.
Following the preliminary analysis of the sediment data in 2015,
it was decided that SSC concentrations were adequately charac-
terized for low magnitude floods. Thus, samples collected during
low magnitude events were not analyzed for SSC after 2015; only
samples obtained during moderate to high magnitude floods were
analyzed. Site BH2 was an exception in that a total of 135 samples
were collected to characterize changes in SSC following the influx
of sediment to the channel from an upstream area of clearcut
logging that occurred in 2016.
2.3. Collection and analysis of precipitation data
Precipitation data were collected using a tipping bucket rain
gauge located near the center of the basin (Fig. 1) between
February 27, 2012 and August 5, 2014. Failure of the rain gauge
prohibited the collection of data from more recent events.
Precipitation events for which SSC data were also available were
characterized in terms of their total rainfall, duration, average
intensity, and maximum 1-hr intensity.
2.4. Calculation of sediment loads and basin sediment yields
A wide range of approaches have been put forth to calculate
sediment loads, all of which possess advantages and dis-
advantages (see Letcher et al., 1999; Richards, 2001; Degens
and Donohue, 2002, for reviews). When the sampling program
was established we had intended to use extrapolation methods
to calculate load because it was expected that SSC data would
not be available for rare, high magnitude floods. However, the
lack of strong, statistically significant relationships between
discharge and SSC over the sampling period led to the use of a
flow-weighted, averaging method patterned after Walling and
2 J.R. Miller et al. / Anthropocene 25 (2019) 100188
Webb (1981). Mathematically, it is expressed as:
Load ¼ K X
ðqr Xi
i¼1
ðCri
n ÞÞ ð1Þ
where K is a conversion factor to account for units and the period of
reporting, Cri is the median concentration of the individual
samples within a specified interval of flow duration (r), qr is the
centroid discharge of the specified interval of flow, and n is the
number of samples within the specified interval of flow. The
median, rather than the mean, SSC value was used because outliers
significantly altered mean SSC values at the sites. As is common,
the flow interval used in the analysis varied such that it was
minimized at high flow and increased for low flows.
2.5. Mapping of process zones
Recent studies have shown that the magnitude, rate, and
dominant type(s) of geomorphic process(es) operating within and
adjacent to a river channel may not systematically vary along the
drainage network, but change abruptly downstream between
stream segments (Montgomery and Buffington, 1993, 1997; Grant
and Swanson, 1995; Miller et al., 2012). In order to capture abrupt
spatial variations in channel and valley form, processes, and
behavior, many investigators have applied a hierarchical analysis of
drainage networks (Montgomery and Buffington, 1993; Grant and
Swanson, 1995; Fryirs and Brierley, 2001; Miller et al., 2012, 2013).
Reach-scale units, referred to as process domains (Montgomery
and Buffington, 1993, 1997) or process zones (Brierley and Fryirs,
2005) are particularly well-suited for management purposes as
they include segments of the stream channel and adjacent valley
floor that are characterized by homogenous landforms and
processes. In this study, process zones were defined, characterized,
and mapped on the basis of channel slope, depth of channel
incision, degree of bank erosion, presence of bedrock exposures,
occurrence of knickpoints and headcuts, width and sediment
storage potential of the valley and floodplain, and the existence of
wetlands.
Fig. 1. Location of the Big Harris Creek basin in North Carolina (A), and location of the rain gauge, the four water quality monitoring stations, and the passive samplers within
the project area (B). Image shows land-cover in March 19, 2017. Spot Image obtained from Google Earth, April 2018.
J.R. Miller et al. / Anthropocene 25 (2019) 100188 3
2.6. Stratigraphic and dendrochronologic analyses
The alluvial (and colluvial) stratigraphy exposed in the channel
banks was described at 21 locations within the basin (Fig. 2).
Deposits along some reaches could not be described because of an
inability to gain property access from landowners. Delineation of
stratigraphic units and facies followed the procedures outlined in
Kottlowski (1965) and Bridge (2006). Contemporary and historic
deposits were separated on the basis of historic artifacts buried in
the deposits, and the degree of sediment weathering (soils).
Correlations were performed, where possible, on the basis of
topographic and stratigraphic position, unit sedimentology,
weathering characteristics, cross-cutting relationships, and
dendrochronological dating methods. Absolute deposit age was
also documented using radiocarbon dating methods. Radiocarbon
analyses were performed by International Chemical Inc., located in
Sunrise, Florida, USA.
The drainage network within the headwaters of Big Harris
Creek was lined by a nearly continuous cover of trees (Fig. 1),
primarily consisting of tulip poplars (genus – Liriodendron). The
channel banks could not be defined through the tree cover,
inhibiting the use of sequential aerial photographs to document
bank erosion rates. However, most trees, particularly along process
zones dominated by wide alluvial valleys, exhibited an extensive
network of roots that had been exposed by corrasion and slab
failures. Decadal-scale rates of bank erosion were estimated using
dendrogeomorphic methods (patterned after Hupp and Bornette,
2016) by determining the amount of erosion that had occurred
around the roots of dendrochronologically dated trees growing
along the upper edge of the channel bank. More specifically, bank
erosion was estimated as the distance from the bank to the most
distant root exposed within the channel (i.e., the amount of
erosion), divided by the determined age of the tree. Given that time
since tree germination was used in the calculation, the method
provides a conservative (minimum) estimate of erosion at the site.
Additional trees were also sampled (cored) to determine the
approximate time of valley floor stabilization as indicated by tree
germination. In total, 35 trees were sampled using a Swedish
increment borer. At least two cores were taken from each tree at
approximately 1 m above the ground. Samples were mounted and
prepared according to procedures described by Stokes and Smiley
(1996). Samples were sanded with up to 400 grit sandpaper,
examined under a microscope, and cross-dated using skeleton
plots.
2.7. Statistical analyses
Time series and other plot types were created using Origin 9.1.
All other statistical analyses were conducted using SYSTAT 9.0.
3. Results
3.1. Monitored hydrologic events
Water level fluctuations were recorded at four monitoring sites
(BH1, BH1, BHT1, BHT2) at 5-minute intervals (Table 2). Significant
differences exist in the number, timing, and magnitude of flood
events recorded at the monitoring stations (data not shown).
While some events were observed at all sites, others were only
observed at a few. Noted variations presumably reflect spatial
differences in rainfall and upstream basin characteristics (e.g.,
basin area, relief, land-use, soil type, etc.)(Table 3). The range of
monitored flows was impacted by a significant drought in 2015 and
2016. The USGS National Streamflow Statistics (NSS) package (Ries,
2006) was used to estimate the recurrence interval of the largest
recorded flood. The largest flood at BH1, the downstream most
monitoring station, was estimated to be a 25- to 30-year event,
whereas the largest monitored flood at BHT2 was estimated to be
about a 5-year flood (Table 2). NSS is not recommended for
application to small basins; thus, accurate estimates could not be
generated for BH2 and BHT1. The generated estimates suggest that
SSC data were obtained over a moderate range of discharge
conditions.
3.2. Suspended sediment concentrations and loads
3.2.1. Magnitude and spatial differences in SSC and loads
Mean SSC values at the four sites ranged from 632 mg/L at BH1
to 1109 mg/L at BH2, the latter following the localized clearcutting
of trees in the subbasin’s headwaters (Table 4). Median values
tended to be lower, particularly for BH2, as the mean was strongly
influenced by a few high SSC measurements (Table 4). Maximum
values exceeded 7165 mg/L at the four sites, and was greater than
9830 mg/L at the three upstream monitoring sites.
SSC values documented using passive (U59B) samplers were
often higher than those measured using ISCO automated samplers
(Table 4), including at sites where sampler types were co-located.
The ISCO samplers collected water and sediment from a single
height above the channel bed, but at multiple (up to 24) times
during the entire flood. In contrast, passive samplers collected
water immediately below the water surface during rising flow
conditions, but at 3 to 4 heights above the channel bed. As shown
Table 1
Location and general characteristics of suspended sediment sampling sites. See
Fig. 1 for site locations.
Site
Number
Basin
Area
(km2)
Instrumentation Monitoring
Period
Additional
Comments
BH1 9.61 ISCO
Levellogger
Turbidity Probe
U59B samplers
11/1/2012
to 04/30/
2017
Channel is incised
~3 m below the
valley floor, and is
characterized by
easily eroded,
vertical or near
vertical banks, and
a mobile channel
bed
BH2 2.12 ISCO
Levellogger
U59B samplers
8/22/2012
to 04/30/
2017
Reach
characterized by
relatively high
gradient, large
channel bed
material and local
exposures of
bedrock
BHT1 0.64 ISCO
Levellogger
U59B samplers
8/21/2012
to 04/30/
2017
Site is close to BH2;
channel
characteristics
differ; slope at
BHT2 is lower and
bed material is
significantly
smaller
BHT2 3.29 ISCO
Levellogger
U59B samplers
8/21/2012
to 04/30/
2017
Site 11 is located on
a tributary to Big
Harris Creek;
channel is incised
several meters
below the valley
floor; channel bed
dominated by
sand-sized
sediments. Reach
characterized by
periodic passage of
sand sheets
4 J.R. Miller et al. / Anthropocene 25 (2019) 100188
Table 2
Descriptive statistics of discharge measured at the four monitoring sites.
Site na Meanb
(cms)
Maximum
(cms)
RI Maxc
(years)
BH1 368,716 0.1258 29.478 25-30
BH2 377,406 0.020 0.825 —d
BHT1 407,927 0.0835 1.487 —d
BHT2 282,538 0.1352 9.199 ~5
a Number of measurements made at 5-minute intervals at the site.
b Based on mean of all measurements taken.
c Estimated recurrence interval of maximum event recorded during monitoring
period; determined using rural parametric equation for North Carolina as contained
in the USGS National Streamflow Statistics program (NSS).
d Application of NSS to basins less than 2.59 km2 (1 mi2) is not recommended;
thus, values not provided.
Table 3
Summary of selected basin morphometric and land-use characteristics within the
basins upstream of the monitoring sites. Land-use mapping based on March 19,
2017 Google Earth Imagery.
BH1
(Total Basin)
BHT2 BH2 BHT1
Basin Area (km2) 9.61 3.29 2.12 0.64
Basin Relief (m) 67.06 42.67 48.77 36.58
Basin Length (km) 3.61 2.01 1.92 1.21
Land-Use (% of total area)
Forest 47.6 56.7 61.6 6.6
Pasture, Lawns, etc. 43.5 32.3 28.7 61.0
Row Crops & Barren Ground 7.3 9.5 8.1 29.7
Impervious Cover 1.5 1.5 1.6 2.7
Basin Area (km2) 9.61 3.29 2.12 0.64
Fig. 2. Map of the Big Harris Basin. Drainage divides of subbasins are delineated by solid black lines. Colored lines show the distribution of process zone types within the basin.
Dashed lines indicate type of process zone has not been field verified. See Table 6 for process zone characteristics.
J.R. Miller et al. / Anthropocene 25 (2019) 100188 5
below, concentrations of suspended sediments were often higher
during the onset of a flood than during the falling stage of the flood.
Thus, differences in measured SSC between samplers reflect, in
part, differences in the timing and position of sample collection
with the water column. Other factors, such as the active versus
passive collection of river water, may also have contributed to the
observed differences in SSC values.
Spatially, SSC measured by the passive samplers varied in a
manner that is consistent with the data collected at the ISCO
automated sites (Fig. 3). SSC was relatively low in BH2 and along
the lower reaches of Big Harris Creek (immediately upstream of
BH1) in comparison to BHT1 and BHT2 (prior to clearcutting in the
BH2 subbasin).
Sediment loads for the four monitoring sites were calculated
using Eq. (1). Catchment wide, sediment loads varied from
529 tonnes/yr at BH2 to 3895 tonnes/yr at BH1 (Table 5). In
contrast to total suspended sediment loads, the highest sediment
yields, 1649 tonnes/km2/yr were measured in BHT1 (the smallest
monitored subbasin) (Table 5).
3.2.2. Temporal variations in SSC over the monitoring period
During low magnitude flood events, large increases in SSC
occurred at all four sites before significant increases in discharge.
Thus, discharge-SSC relationships were either poorly-defined, or
exhibited nearly vertical trends, indicating that SSC is semi-
independent of discharge (Fig. 4). In contrast, during moderate to
high magnitude floods a statistically significant trend exists
(p < 0.05) between discharge and SSC at BH1, although the trend
is characterized by considerable variability (Fig. 4a). At a discharge
of 1 cubic meter per second (cms), for example, SSC varied from a
few hundred to a few thousand mg/L. Discharge-SSC relationships
were more strongly defined for individual floods, as is evident on
Fig. 4a as linear data point ‘trails’ shown by the blue circles.
Data from BH2 was also characterized by two distinct
populations of flood events (Fig. 4d), although the statistically
significant relationship between discharge and SSC is less
pronounced. At the two upstream most sites (BH2, BHT1),
statistically significant relationships between discharge and SSC
existed during some individual moderate to high magnitude
floods. However, when data from multiple events was combined,
no relationship was present (Fig. 4b, c).
On March 2, 2016, a large amount of fine-grained (< 63 mm)
sediment was observed on the bed of the channel at and upstream
of BH2. The sediment was derived from the clearcutting of trees
from a 0.20 km2 area in the headwaters of the subbasin. The influx
of sediment was accompanied by a temporary increase in SSC at
BH2 (Table 4). A t-test demonstrated that there was a statistically
significant (p < 0.05) difference in SSC between pre- and post-
cutting periods. The increase in SSC did not influence concen-
trations further downstream at BH1.
Table 4
Statistical summary of collected SSC data; concentrations presented in mg/L.
Site n Mean Minimum Median Maximum
BH1 406 632 2.5 385 7165
BH2 (total) 287 1109 2.5 221 10,342
BHT1 139 713 24 216 9,832
BHT2 154 803 43 415 11,081
Passive 153 2,322 80 1,446 13,961
BH2 (total) 287 1109 2.5 221 10,342
BH2
(Pre-Cutting)
152 182 8.8 92 9,586
BH2
(post-Cutting)
135 1,477 2.5 848 10,342
Fig. 3. SSC data collected using U59B passive samples located at 15 sites in the basin. With one exception in each area, mean SSC is lower along the channel drained by BH2 and
the lower reach of Big Harris Creek (labeled BH1). See Fig. 1 for sampler locations.
Table 5
Summary of suspended sediment loads and basin sediment yields.
Statistic BH1 BH2 BHT1 BHT2
Load
(tonnes/year)
3895 529 1,056 2,256
Sediment yield
(tonnes/km2/year)
425 249 1649 687
SSC values used in calculation based on median value for specific range of discharge
exceedance.
6 J.R. Miller et al. / Anthropocene 25 (2019) 100188
3.2.3. Variations in SSC during individual flood events
As noted above, semi-systematic direct relationships between
discharge and SSC were common during moderate to high
magnitude floods. Nonetheless, variations in SSC were far from
simple. The most systematic variations in SSC and discharge
occurred at BH1 during moderate to large floods (Fig. 5). In the
simplest case, SSC peaked before peak discharge (Fig. 5a) creating a
phenomenon known as hysteresis in which SSC for a given
discharge exhibited two (or more) distinct values during a single
event depending on whether the sample was collected during the
rising or falling phase of the flood. The most commonly observed
form of hysteresis was a clockwise loop, such as that which
occurred on July 4, 2013 at site BH1 (Fig. 5b). In this case, SSC is
higher during the rising limb of the hydrograph than the falling
limb. At a discharge of 0.6 cms, for example, SSC was more than 3
times higher (~1300 mg/L) during the rising limb than for the same
discharge during the falling limb of the hydrograph (when it as
about 400 mg/L)(Fig. 5b).
Clockwise-hysteresis was also observed at the other three
monitoring sites. However, hysteresis was often more complicated
at these sites, particularly when the flood was associated with
multiple periods of relatively intense precipitation. More impor-
tantly, the peak in SSC often occurred very early in the flood; in
some instances, SSC rose rapidly with the onset of precipitation
and peaked before significant increases in flow (Fig. 5c). Turbidity
data collected at BH1 exhibited similar trends (see Fig. S1 in
Supplementary Materials).
Counter-clockwise hysteresis was occasionally observed at BH2,
BHT1, and BHT2; it was particularly common at BH2 (Fig. 5f).
During some discharge events that exhibited multiple periods of
high flow, SSC rose during the initial phase of the flood (creating
clockwise hysteresis). SSC then rose again following subsequent
peaks in discharge, creating a counter-clockwise hysteretic loop
(Fig. 5e,f).
3.3. Analysis of storm related controls on SSC
To gain additional insights into the controls on SSC, precipita-
tion data collected near the center of the basin (Fig. 1) was related
to maximum event SSC. A total of 27 storms for which both
precipitation and SSC data were collected were available between
February 27, 2012 and August 5, 2014. Relationships between SSC
and total rainfall, duration, average intensity, and maximum
intensity were determined for each of the four monitoring sites
using stepwise multiple regression. The analysis revealed that
variations in SSC at BH1 can be explained, in part, by total event
precipitation (Tprec) and average hourly rainfall intensity (HrIt),
such that:
SSCmax = -61.41 + 2705.1HrIt +491.4Tprec (2)
The relationship is statistically significant (p < 0.05), and
explained 47% of the total variability in SSC values.
Stepwise regression conducted on data from BH2 produced a
model in which total event rainfall explained about 33% of the
variation in maximum flood SSC. Statistically significant regression
equations could not be developed for BHT1 or BHT2, in part
because data for all parameters were available for only 11 and 12
events, respectively.
3.4. Type and distribution of process zones
Process zones represent channel/valley reaches characterized
by semi-homogeneous landforms and geomorphic processes. Eight
different types of process zones were defined within the Big Harris
Fig. 4. Relations between discharge and suspended sediment concentration (SSC) at (A) BH1, (B) BHT1, (C) BH2, and (D) BHT2. Note large variations (exceeding 1–2 orders of
magnitude) in SSC over a narrow range of discharge during low flow conditions. Dashed vertical line defines discharge threshold between low and moderate/high magnitude
floods. Red circles – data from low magnitude floods; Blue circles – data from moderate to high magnitude floods. Statistically significant relationships between SSC and
discharge (dashed orange line) occurred during moderate and larger events within the larger basins represented by BH1 (A) and BHT2 (D). Inverse relationship observed in
plot A (dark blue circles) is primarily associated the April 27, 2013 flood (For interpretation of the references to colour in this figure legend, the reader is referred to the web
version of this article).
J.R. Miller et al. / Anthropocene 25 (2019) 100188 7
Creek project area (Fig. 6). The spatial distribution of the process
zones is shown in Fig. 2 and quantified in Figure S1; their
characteristics are provided in Table 6. A more detailed description
of the similarities and differences of the defined process zones is
provided with the Supplementary Materials that accompany the
paper.
3.5. Alluvial stratigraphy and sedimentology of recent valley fill
The recent alluvial stratigraphy of the upper Big Harris Creek
basin was described in the channel banks at 21 sites shown on
Fig. 2. Process zones characterized by wide valleys and entrenched
alluvial channels (Groups 2–4, Table 6) possessed banks consisting
of two primary sedimentary units (Figs. 7a and 8). Where both
were present, the upper unit ranged from 0 to more than 150 cm in
thickness (Fig. 2); it consisted of dark reddish brown to buff colored
layers of massive sandy loam sediments and/or cross- or
horizontally-bedded, loose sands. Its sediments contained historic
artifacts, such as metal straps, indicating that the unit post-dates
European settlement in the area. These sediments are therefore
referred to here as legacy sediments (Fig. 7).
The upper sedimentary unit (i.e., the legacy sediment) abruptly
overlies a finer-grained (clayey), dark grey to black unit along an
abrupt boundary (Fig. 7a,b). The grain size of these sediments
systematically decreased downward into grey (gleyed) fine-
grained, silt- to clay-loamy sands, and eventually to loamy sands
to gravels. Radiocarbon dates obtained on organic sediments (2
samples) and wood (1 sample) collected from along Upper Stick
Elliott Creek constrained the age of the unit to between 290 +/- 30
YBP and 3760 +/- 30 YBP. It appears, then, to predate European
settlement in the area, as was expected.
Channel banks along ephemeral and perennial headwater
channels as well as deeply incised channels were primarily
composed of relatively old, highly weathered alluvial and colluvial
sediment of varying texture (Figs. 6, 7c). Although grain size
variations were visible, exact depositional boundaries were
typically obscured by thick soil development. An exception
occurred in areas characterized by topographic depressions,
particularly within the headwaters of Stick Elliott and Fletcher
Creeks, where the dark grey to black, pre-settlement deposits
occurred at the surface of the valley fill. These deposits varied in
thickness from about 1 to 2 m.
Fig. 5. Variations in discharge and SSC at BH1 during flood events on July 4, 2013 (A) and the associated hysteresis loops (B). (C) Variations in discharge and SSC at BH2 during
an event on October 7, 2013, and the hysteresis loop observed during the event (D). Note that the initial increase in suspended sediment concentrations occurred at the start of
rising water levels and create a clock-wise hysteretic loop. In contrast, variations in discharge and SSC during an event on July 27, 2013 flood at BHT1 (E) created a counter-
clockwise hysteresis loop at BHT1 if data collected during the second rise in discharge are considered (F).
8 J.R. Miller et al. / Anthropocene 25 (2019) 100188
3.6. Bank erosion and erosion rates
Channel banks along process zones consisting of entrenched
alluvial valleys (with or without bedrock) were extensively eroded,
creating long reaches characterized by nearly vertical banks devoid
of vegetation. The primary processes of erosion were corrasion (the
grain by grain removal of sediment) and slab failures, resulting in the
exposure of tree roots along the bank (Fig. 9). Field observations
revealed that slab failures primarily occurred during the later
(falling) stages of a flood, or immediately after a flood event. Thus,
failed debris locally occurred between events along the base of the
banks and the edge of the water during base flow conditions.
Decadal-scale bank erosion rates estimated at 19 sites using
dendrogeomorphic methods ranged from 1.02 to 5.81 cm/yr; the
median was 2.30 cm/yr (Fig. 10b). As noted earlier, these rates are
based on the age of the tree since germination and, therefore,
represent conservative (minimum) estimates of the rate of bank
erosion.
4. Discussion
4.1. Comparison of suspended sediment production and transport
within the region
Suspended sediment concentrations, loads, and yields have
been examined within a wide range of basins in the piedmont of
North Carolina by Simmons (1993) and, more recently, Simon and
Klimetz (2008). Suspended sediment production and transport
Fig. 6. Examples of selected process zone by group within the project area.
J.R. Miller et al. / Anthropocene 25 (2019) 100188 9
within Big Harris Creek are substantially higher than in other
catchments characterized in the region by these two investiga-
tions. Simmons (1993), for example, analyzed suspended sediment
data collected at 152 gaging stations in North Carolina between
1970 and 1979. He found that the highest recorded SSC values were
in agricultural and urban impacted basins, and reached a
maximum of 5600 mg/L. This maximum value is well below the
maximum SSC value measured for Big Harris Creek (11,081 mg/L,
Table 4).
A comparison of sediment loads within the region to Big Harris
Creek is complicated by differences in basin area. As a headwater
catchment, the Big Harris Creek project area is substantially
smaller than other monitored catchments, and would be expected
on the basis of catchment size to exhibit lower sediment loads.
However, sediment loads measured within the Big Harris Creek
project area are some of the highest reported for the piedmont of
North Carolina by both Simmons (1993) and Simon and Klimetz
(2008). The maximum sediment load reported by Simmons (1993)
for urban basins <25 km2 in size (about twice the size of Big Harris
Creek above BH1, 9.61 km2) was 399 tonnes/yr within McMullen
Creek, near Charlotte, North Carolina. The maximum sediment
load observed for a rural basin affected by agriculture was
2813 tonnes/yr within Double Creek near Roseville (Simmons,
1993). The estimated load at BH1 in Big Harris Creek was
3895 tonnes/yr (Table 5).
Sediment yield describes the amount of sediment discharged
from the catchment per unit basin area. Simon and Klimetz (2008)
found that within the piedmont of the southeastern U.S., the
median sediment yield was 39 tonnes/km2/yr; the 90th percentile
of unstable basins studied exhibited a mean sediment yield of
72 tonnes/km2/yr, well below the 425 tonnes/km2/yr estimated for
BH1 within Big Harris Creek. The values provided by Simon and
Klimetz (2008) are lower than those provided earlier by Simmons
(1993). Simmons found that the maximum sediment yield
estimated for an agriculturally affected basin within the piedmont
was 154 tonnes/km2/yr. Only Irwin Creek, a clear outlier affected by
urban activities near Charlotte exhibited a sediment yield higher
than estimated for Big Harris Creek (525 versus 425 tonnes/km2/yr,
respectively) (Table 5). Most urban affected watersheds examined
by Simmons (1993) exhibited sediment yields of <~230 tonnes/km2/
yr.
4.2. Geomorphic evolution of the drainage network
Significant differences in suspended sediment concentrations,
loads, and yields were observed between the four drainages that
were monitored at BH1, BH2, BHT1, and BHT2 as well as with the
passive samplers (Tables 3–5). In the following sections, it is
argued that these spatial differences, and the relatively high values
of sediment production and transport within the Big Harris Creek
basin, are related to the types and distribution of process zones
within the catchment, which, in turn, reflect the basin’s geomor-
phic responses to land-use change.
As noted earlier, process zones characterized by entrenched
alluvial channels (Table 6, Fig. 7) consistently contain alluvial
valley fill deposits composed to two primary units. The lower unit,
Table 6
Summary of process zone types and characteristics. See Fig. 6 for photographic examples of process zones.
Reach Valley Characteristics Slope Description
Ephemeral, Headwater
Reaches
Relatively steep, narrow valleys;
colluvium encroaches on valley floor
Relatively
high (~0.02)
Group 1 - Low-order, ephemeral channels typically found at upstream most section of
the drainage network; gullies extend from channels into colluvial hillslope;
entrenchment is relatively high (>2.5 m); channels may terminate in headcut; reach
dominated by ongoing incision, headcut migration, and/or bank erosion; no legacy
sediments present; limited sediment storage in bards or other channel features
Perennial, Headwater
Reaches
Narrow, but well defined alluvial
surface, with entrenched channel
Low to
Moderate
(~0.01-.015)
Group 1- Low-order, perennial reach; channels entrenched several meters below
valley floor; inset floodplain locally present in trench; thin layers of legacy sediments
locally present on terrace surface; sediment derived predominantly from bank
erosion and minor channel incision; limited sediment storage in bars and other
channel features
Entrenched Alluvial
Channel with Inset
Terraces
Wide, well-defined alluvial valley
floor
Low to
Moderate (~.01
- .015)
Group 2 - Channel entrenched 1-3 m below valley floor; well-defined inset floodplain
within trench; bank often semi-vertical on side opposite of inset and composed of
both pre- and post-settlement (legacy) deposits; reach dominated by bank erosion
and minor channel incision; sediment storage occurs on inset terrace, within channel
bars, and on valley floor
Entrenched Alluvial
Channel without Inset
Terraces
Wide, well-defined alluvial valley
floor
Low to
Moderate (~.01
- .015)
Group 2 - Channel entrenched 1-3 m below valley floor; process zone generally found
along downstream sections of a subbasin; banks are near vertical and commonly
composed of both legacy and pre-settlement deposits; reach dominated by bank
erosion; sediment storage occurs in relatively small bars and on valley floor
Alluvial Channel, with
Bedrock Exposures
Wide to moderately wide valley
floor
Moderate to
high (~0.15-
0.2)
Group 2 – Entrenched channel (1-3 m) with local bedrock outcrops along channel
bed; may form knickpoints that limited channel incision; bedrock areas surrounded
by loose, sand and gravel; bank erosion is common, but not as extensive as along
entrenched alluvial reaches; reach dominated by sediment transport with sediment
storage on valley floor
Entrenched Channel with
Mobile Bed
Wide, well-defined alluvial valley Low (~0.01-
0.008)
Group 2 - Channel entrenched 3-4+m below valley floor; channel bed prone to large
changes during flood, but composed of well-defined morphologic units; banks are
typically vertical, and subject to significant erosion by corrasion and with frequent
slab failures; in-channel storage is high in bars, but depth of incision inhibits valley
floor sediment storage; bank materials consist of both legacy and pre-settlement
sediments
Highly Entrenched
Channel
Variable valley width composed of
both colluvial and alluvial sediments
High to
moderate
(~0.15-0.2)
Group 3 - Channel entrenched 2-5 m below valley floor; trench walls, composed of
relatively old alluvial deposits and saprolite, confine flow and allow for efficient
sediment transport; erosion of steep trench walls is significant; post-settlement
deposits limited to local terraces inset in trench, particularly along downstream
reaches.
Alluvial Channel with
Riparian Wetlands
Wide, alluvial valley Low to
Moderate
(~.01 - .015)
Group 4 - Channel entrenchment is limit and cut into alluvial fill; riparian wetlands
with small springs border channel; bedrock close to surface, and locally outcrops in
bed and banks; reaches are currently stable; moderate sediment storage in channel
bars and on floodplain/valley floor.
10 J.R. Miller et al. / Anthropocene 25 (2019) 100188
consisting of dark colored, finer-grained, organic rich sediments,
dates back to approximately 4000 YBP. The grain size and
composition of this unit suggest that the hydrologic and
geomorphic conditions at the time of deposition were fundamen-
tally different from those that existed during the deposition of the
overlying coarser-grained, buff colored sediments. The transition
from lower to upper basin conditions was presumably related to
land-use changes that occurred at the end of the 1800s and early
1900s. Historical data and landowner accounts indicate that most
the previously forested basin was clearcut to grow cotton. In fact,
from the late 1800s until about the late 1940s the area was one of
the largest cotton producers in the U.S. The change in land-use led
to several decades of severe upland erosion and valley bottom
sedimentation that has been well documented throughout the
southeastern U.S. (e.g., Trimble, 1974; Leigh and Price, 2006; Leigh,
2010), and that was responsible for producing what is referred to in
the literature as post-settlement alluvium (or legacy sediments)
(Waters et al., 2007).
The thickest and most extensive legacy deposits in Big Harris
Creek were associated with process zones characterized by
entrenched alluvial channels (with or without inset terraces)
and entrenched channels with mobile beds (Fig. 2). These process
zones possess wide valley floors that allowed for the accumulation
of eroded upland sediments. In contrast, along upstream reaches,
dominated by ephemeral and perennial headwater channels,
legacy sediments were generally absent. However, an inset terrace
composed of pre-settlement deposits locally occurred within
deeply incised reaches. Large trees, dated to between 1944 and
1953, occurred on the surface of the terrace, indicating that
incision in headwater areas begun during the period of extreme
upland erosion (late 1800s to early 1900s), and allowed the dark-
colored, pre-settlement deposits to remain at the surface of the
entrenched valley floor (e.g., along Stick Elliott and Fletcher
Creeks). Upland gullies that formed during the late 1800s and early
1900s, and which headed on hillslopes, were also associated with
ephemeral or perennial headwater channels.
Sediment eroded during early incision of these upstream areas
was transported downstream where it was delivered to the
floodplain/valley floor surface, thereby burying the dark, organic
rich pre-settlement deposits. The obtained dendrochronological
data show that most tulip poplar trees located along Big Harris
Creek germinated in the 1950s and 1960s (Fig. 10), the time that
conservation practices were implemented in the basin. Given that
significant deposition and reworking of sediment on the valley
floor would have either killed recently planted trees, or inhibited
tree germination, channel incision was likely to have begun along
downstream reaches of the drainage network around the mid-
1900s. Incision exposed both the legacy sediments and the older
pre-settlement deposits within process zones now characterized
by entrenched alluvial channels. Locally, incision was inhibited by
resistant bedrock units that outcrop along the channel bed. These
bedrock units, and their influence on incision, led to the
development of two types of process zones: alluvial channels
with bedrock exposure and alluvial channels with riparian
wetlands. Legacy sediments were limited along both types of
process zones.
The influence of bedrock on channel development is well
illustrated by comparing channel form, and process zone type,
upstream of BHT1 and BH2 (Fig. 2). The channel bed within the
BH2 drainage is periodically characterized by bedrock exposures
that restricted channel incision, and which lead to elevated water
tables. The result was the development of a relatively wide valley
floor locally characterized by alluvial channels with riparian
wetlands (Figs. 2, S1). In contrast, bedrock exposures are limited
along the bed of Royster Creek upstream of BHT1. Moreover, the
bedrock differs from the strata that underlies BH2 in that it is
comprised of more easily weathered and eroded rocks. As a result,
the depth of channel incision was significant, exceeding 4 m below
the valley floor, and increases upstream with distance from the
axial drainage. Incision resulted in the development of a deeply
incised channel (Fig. 2) characterized by a v-shaped valley that
possesses steep, locally barren slopes that sluff sediment from
upland soils and, at depth, saprolite, into the channel along the
mid- and upstream reaches of the BHT1 drainage (Figs. 6, S1).
Within these areas there is little room for sediment re-deposition
and storage.
4.3. Controls on suspended sediment concentrations and loads
The recognition that legacy deposits may be an important
source of sediments to a stream or river is not new. An increasing
number of studies have shown that the erosion of legacy sediments
can degrade water quality, aquatic habitat, and potentially, riparian
biotic assemblages (Allan, 2004; Jackson et al., 2005; Bain et al.,
2012; Niemitz et al., 2013; Surasinghe and Baldwin, 2014). What is
significant from this study is that the geomorphic history of the
basin, and the resulting legacy sediments, influenced the source
and transport of SSC within a small, headwater stream.
Field observations have shown that bank erosion associated
with channel incision was a significant problem within Big Harris
Creek, especially along process zones associated with entrenched
Fig. 7. Bank exposure located along entrenched alluvial channels (A) near the
confluence of Stick Elliot and Fletcher Creek, and (B) upstream, along Stick Elliot
Creek. Reddish brown to buff colored sediments represent legacy sediments
enriched in sand-sized sediments. Dark sediments represent fine-grained, pre-
settlement deposits. (C) “old” alluvial deposits that comprise upland deposits at the
confluence of Stick Elliott and Fletcher Creeks.
J.R. Miller et al. / Anthropocene 25 (2019) 100188 11
Fig. 8. Stratigraphic section measured near BH1, near the mouth of the study basin. Graph shows the finer-grained nature of the pre-settlement sediments, and the sandy
nature of the legacy sediments.
Fig. 9. Exposure of tree roots caused by bank erosion. Rates of erosion were determined by (a) dendrochronologically determining the age of the trees and (b) using the roots
as a guide to the amount of erosion that has occurred since germination of the tree (illustrated by the red arrow). Bank erosion locally exceeded 3 m (For interpretation of the
references to colour in this figure legend, the reader is referred to the web version of this article).
12 J.R. Miller et al. / Anthropocene 25 (2019) 100188
alluvial channels (Table 6). Erosion primarily occurred by means of
corrasion (the grain by grain removal of particles) from sandier
bank units, and slab failures where finer, more cohesive bank units
occurred (Fig. 9). If it is assumed, for illustrative purposes, that the
median rate of bank erosion (determined using dendrochronolog-
ical methods, Fig. 10) was 2.30 cm/yr, the average bank height was
2.5 m and the bulk density of the bank material was 1.65 g/cm3,
94.9 tonnes of sediment would have been supplied to the channel
per linear kilometer of bank per year. Of course, bank heights,
erosion rates, etc. are not consistent along the channel, and more
detailed analyses are needed to refine such sediment budgeting.
Nonetheless, when compared to measured sediment loads, this
rough calculation demonstrates that bank erosion was a significant
source of sediment to the channel. For example, if sediment was
derived from both sides of the channel, the median calculated
erosion rate suggests that about half of the sediment load
measured in the larger drainages (BHT2 and BH1) was derived
from bank erosion; within the BH2 drainage, nearly all of the
measured suspended sediment load can be accounted for by bank
erosion.
Given the importance of bank materials as a sediment source,
the basin’s geomorphic history and its resulting alluvial stratigra-
phy appears to be responsible for the duality in suspended
sediment transport observed within the monitoring data. Bank
erosion by slab failure often resulted in the accumulation of loose
sediment at the base of the banks and along the edge of the water
during base flow. Raindrop impacts and runoff from these deposits
during precipitation was able to have delivered fine-sediments to
the water column during low magnitude floods (those
characterized by minimal changes in water levels). During larger
events, these fine sediments could be flushed into the water before
significant increases in discharge occurred. Thus, the input of bank-
side sediment accounts for both the observed, highly elevated SSCs
during low magnitude floods, and the noted hysteresis effects in
which rapid rises in SSC (Fig. 5) occurred before significant
increases in discharge. Most of this sediment was likely to have
been derived from the pre-settlement deposits because (1) their
finer-grained and more cohesive nature allowed these deposits to
more frequently fail by slab failure, and (2) more runoff cold be
generated from these finer-grained sediments during low intensity
precipitation than from failed materials derived from the sandier,
more permeable legacy sediments. Some of the sediment was also
derived from cattle crossings where cows loosened the bank
sediments and moved it to the edge of the water. Importantly,
because discharge in the channel was limited during these low
magnitude floods, significant increases in SSC did not require large
inputs of fine sediment.
During higher magnitude floods, these fine-sediments would
also be delivered to the channel during the very early stages of the
flood. However, as these events progressed, and stream discharge
increased, additional sediments were required to increase SSC.
Inspection of the samples collected during these floods showed
that not only was SSC higher, but the sediments in transport were
dominated by larger, sand-sized particles. These sediments were
presumably derived from the erosion of loose, non-cohesive legacy
deposits exposed in the channel banks, and from upstream gullies
associated with ephemeral and perennial headwater reaches.
Much of the variability in SSC during these large floods can be
attributed to clockwise hysteretic loops. Clockwise hysteresis is
indicative of the “first-flush phenomena” in which easily eroded
sediments result in rapid increases in SSC during the early phases
of a flood. SSC then decreases before the peak in discharge is
reached as the availability of loose, easily mobilized sediment
declines (Williams, 1989). In Big Harris Creek, clockwise hysteresis
during the larger floods (those characterized by a direct relation-
ship between SSC and discharge, Fig. 4) supports the argument that
suspended sediment transport was influenced by the availability of
easily mobilized sediment from failed bank material as well as
coarser grained sediments from legacy deposits and gullies.
The hypothesized difference in sediment sources between
small and moderate to large floods is consistent with the
regression analysis (Equation 2) that found, using data from
BH1, that SSC varied as a function of average rainfall intensity and
total event precipitation. Events of higher intensity and more
rainfall were capable of mobilizing and delivering more sediment
to the water column than lower intensity storms characterized by
less rainfall that were unable to transport sediment from the pre-
settlement deposits to the channel, or to generate the overland
flow required for gully and/or bank erosion.
4.4. Spatial variations in suspended sediment loads
Both the ISCO (Tables 3–5) and passive U59B sampler (Fig. 3)
data revealed that sediment concentrations, sediment loads and
sediment yields varied spatially within the Big Harris Creek project
area. The highest sediment yield was observed at BHT1. It was
several times higher than the sediment yields observed in the
drainages monitored by BH1, BHT2, and BH2 (before clearcutting)
(Table 5). Differences in sediment yield may be accounted for in
part by differences in land-use/landcover. The Royster Creek
subbasin (monitored by BHT1), which exhibits the highest
sediment yields, also exhibited the highest percentage of row
crops in the project area, and the lowest percentage of forested
terrain (Table 3). However, sediment yields also vary with the
nature of the process zones created in response to region’s changes
Fig. 10. (A) Histogram of approximate time of tree germination. Graph does not
account for collection of tree cores at~1 m above ground. Exact date of germination is
likely to be 3–4 years prior to date shown. (B) rates of erosion based on tree age and
width of exposed tree roots along channel banks. Includes trees from along Stick
Elliott Creek, Upper Big Harris Creek, and the BHT1 drainage.
J.R. Miller et al. / Anthropocene 25 (2019) 100188 13
in land use. For example, in contrast to the other subbasins, BHT1 is
dominated by a highly-entrenched channel that possesses a v-
shaped valley with steep, locally barren slopes (Fig. S1). Field
observations showed that sediment was extensively eroded from
the slopes that formed the margins of the channel, and was
transported into the channel along the mid- and upstream reaches
of the BHT1 drainage network. The influx of sediment from the
trench walls was accompanied by the influx of sediment from
gullies and bank erosion associated with ephemeral headwater
channels.
While sediment yields within BHT1 were relatively high, total
sediment export (load) from the basin was relatively low because
stream flow was limited by its small catchment area (Table 5). The
BHT2 drainage area exhibited the highest sediment loads (Table 5).
While it encompasses only about 33% of the watershed, it
contributed 57% of the calculated sediment load for the entire
project area. Land use within the subbasin during monitoring was
dominated by forest (56.7%) and pasture (32.3%) lands, which were
unlikely to have generated significant quantities of upland
sediment. Rather, the relatively high rates of suspended sediment
generation and transport from BHT2 were probably related to gully
systems associated with ephemeral and perennial channels in the
subbasin’s most upstream reaches (which encompass about 40% of
the basin)(Figs. 2, S1). A particularly large gully characterized by an
~5 m high headcut and steep, eroding sides was present within Stick
Elliot Creek. High sediment loads were also related to the
widespread occurrence of entrenched alluvial channels that were
bound, in part, by legacy sediments containing loose, non-cohesive
sand-sized sediment (Fig. 7a,b). The dendrogeomorphic data, for
example, indicated that bank erosion along entrenched alluvial
channels was significant, particularly within the BHT2 subbasin.
The fact that more than 60% of the drainage network within the
BHT2 catchment was composed of entrenched alluvial channels
(Fig. S1) locally characterized by easily eroded legacy deposits
>50 cm in thickness (Fig. 2), shows that these deposits are a
significant sediment source. Unfortunately, the existing dataset
does not allow for a direct, quantitative analysis of the importance
of the thickness and extent of legacy sediments on sediment loads,
relative to other sediment sources (e.g., gullies). However,
geochemical fingerprinting studies are currently being conducted
to more fully address the source issue. The point here is that the
stratigraphy of the bank deposits, including the thickness and
extent of the basin’s legacy sediments, influenced (1) the
relationships between discharge and SSCs, (2) total sediment
loads exported from the basin/subbasins, and (3) sediment yields.
On March 2, 2016, a large amount of fine-grained (< 63 mm)
sediment was observed on the bed of the channel at and upstream
of BH2 during the collection of water samples at the site. The influx
of fine sediment was due to upstream, localized clearcutting
associated with development. The statistically significant increase
in concentrations, combined with the observed deposition of fine
sediment on the channel bed, indicates that tributary catchments
remain highly sensitive to renewed upland land-use change.
5. Conclusions
Water quality monitoring data were combined with geomorphic
and stratigraphic data to develop a conceptual model of the sources
and transport of suspended sediment within the headwaters of Big
Harris Creek. The source and transport of suspended sediment
differed between relatively low magnitude floods (characterized by
minimal changes in water levels) and moderate to high magnitude
floods. During low intensity, low volume precipitation events,
characterized by minimal overland flow, channel-edge sediment
associated with mass wasting (slab failures) and cattle crossings was
quickly deliveredtothewatercolumn, particularly from fine-grained
pre-settlement deposits located at the base of the banks. During
these low magnitude floods, variations in SSC exceeded 1 to 2-orders
of magnitude, and reflected variations in sediment supply, rather
than differences in channelized flow. During higher magnitude
floods, direct relationships between SSC and discharge existed
within the larger subbasins (BH1, BHT2, Fig. 4). Fine-sediments were
also delivered to the channel during the very early stages of these
floods, creating clockwise hysteretic loops (through a first-flush
process). However, as these events progressed, additional sediment,
particularly from the erosion of legacy sediments and gullies, was
delivered to the channel. Thus, the distinction between low and
moderate/high magnitude floods could be defined on the basis of
SSC-discharge relationships, and varied between the monitoring
sites. The observed differences in SSC between low and moderate to
high magnitude floods complicates the assessment of water quality
changes resulting from stream restoration projects, and suggests
that comparisons will need to differentiate between these two
populations of runoff events.
While the Big Harris Creek study area was a small, headwater
catchment, significant spatial difference in SSC, sediment loads,
and sediment yields were observed. These spatial variations
appear to be related, in part, to: (1) geographical differences in the
basin’s geomorphic responses to land-use change following
European settlement, and the type of process zones that were
created, (2) the distribution of legacy sediments within the process
zones, and (3) the ability of the process zones to produce, transport
and store sediment.
Acknowledgments
The collection of nearly six years of water quality data was aided
by a large number of students and research scientists. We thank all
of them for their help, particularly Ken Walton, Jacob Sinclair, and
Danvey Walsh. Funding for the project was provided in part by the
North Carolina Division of Mitigation Services (NCDMS) and the
Whitmire Endowment at Western Carolina University. We are
grateful to both for their support. We also thank two Anonomous
reviewers for their highly constructive comments.
Appendix A. Supplementary data
Supplementary material related to this article can be found, in the
online version, at doi:https://doi.org/10.1016/j.ancene.2018.12.002.
References
Allan, J.D., 2004. Landscapes and riverscapes: the influence of land use on stream
ecosystems. Annu. Rev. Ecol. Evol. Syst. 35, 257–284.
Armstrong, J., Kemp, P., Kennedy, G., Ladle, M., Milner, N., 2003. Habitat
requirements of Atlantic salmon and brown trout in rivers and streams. Fish
Research 62, 143–170.
Bain, D.J., Green, M.B., Campbell, J.L., Chamblee, J.F., Chaoka, S., Fraterrigo, J.M.,
Kaushal, S.S., Martin, S.L., Jordan, T.E., Parolari, A.J., Sobczak, W.V., Weller, D.E.,
Wolheim, W.M., Boose, E.R., Duncan, J.M., Gettel, G.M., Hall, B.R., Kumar, P.,
Thompson, J.R., Vose, J.M., Elliott, E.M., Leigh, D.S., 2012. Legacy effects in
material flux: structural catchment changes predate long-term studies.
BioScience 62, 575–584.
Bernhardt, P.M.A., Alexander, A.J.D., Barnas, G.K., Brooks, S., Carr, J., Clayton, S.,
Daham, C., Follstad-Shah, J., Galat, D., Gloss, S., Goodwin, P., Hart, D., Hassett, B.,
Jenkinson, R., Katz, S., Kondolf, G.M., Lake, P.S., Lave, R., Meyer, J.L., O’Donnell, T.
K., Pagano, L., Powell, B., Sudduth, E., 2005. Synthesizing U.S. River restoration
efforts. Science 308, 636–637.
Bridge, J.S., 2006. Rivers and floodplains. Forms, Processes and Sedimentary Record.
Blackwell Publishing 491 pp.
Brierley, G.J., Fryirs, K.A., 2005. Geomorphology and River Management:
Applications of the River Styles Framework. Blackwell, Oxford.
Degens, B.P., Donohue, R.D., 2002. Sampling mass loads in river: a review of
approaches for identifying, evaluating, and minimizing estimation error. Water
Resources Technical Series, Water Rivers Commission, Report No. WRT 24, .
Fryirs, K., Brierley, G.J., 2001. Variability in sediment delivery and storage along river
courses in Bega catchment, NSW, Australia: implications for geomorphic river
recovery. Geomorphology 38, 237–265.
14 J.R. Miller et al. / Anthropocene 25 (2019) 100188
Grabowski, R.C., Surian, N., Grunell, A.M., 2014. Characterizing geomorphological
change to support sustainable river restoration and management. Water, Wiley
Interdisciplinary Reviews doi:http://dx.doi.org/10.1002/wat2.1037.
Grant, G.E., Swanson, F.J.,1995. Morphology and processes of valley floors in mountain
streams, western cascades, Oregon. In: Costa, J.E., Miller, A.J., Potter, K.W., Wilcock,
P.R. (Eds.), Natural and Anthropogenic Influences in Fluvial Geomorphology.
American Geophysical Union Monograph, 89, Washington, DC, pp. 83–102.
Guy, H.P., 1977. Techniques of Water-resources Investigations of the United States
Geological Survey. Chapter c1: Laboratory Theory and Methods for Sediment
Analysis, 3rd ed. U.S. Government Printing Office, Washington D.C.
Hobbs, R.J., Norton, D.A., 1996. Towards a conceptual framework for restoration
ecology. Restor. Ecol. 4, 93–110.
Hupp, C.R., Bornette, G., 2016. Vegetation as a tool in the interpretation of fluvial
geomorphic processes and landforms in humid temperate areas. In: Kondolf, G.
M., Piégay, H.H. (Eds.), Tools in Fluvial Geomorphology, , pp. 269–288.
Jackson, C.R., Martin, J.K., Leigh, D.S., West, L.T., 2005. A southeastern piedmont
watershed sediment budget: evidence for a multi-millennial agricultural
legacy. J. Soil Water Conserv. 60, 298–310.
Kemp, P., Sear, D., Collins, A., Naden, P., Jones, I., 2011. The impacts of fine sediment
on riverine fish. Hydrol. Process. 25, 1800–1821.
Kottlowski, F.E., 1965. Measuring Stratigraphic Sections. Holt, Rinehart and
Winston, New York 253 p..
Leigh, D.S., 2010. Morphology and channel evolution of small streams in the
southern blue ridge mountains of western North Carolina. Southeast. Geogr. 50,
397–421.
Leigh, D., Price, K., 2006. Morphological and sedimentological responses of streams
to human impact in the southern Blue Ridge Mountains, USA. Geomorphology
78, 142–160.
Leopold, L.B., Wolman, M.G., Miller, J.P., 1964. Fluvial Processes in Geomorphology.
WH Freeman, San Francisco.
Letcher, R.A., Jakeman, A.J., Merritt, W.S., McKee, L.D., Eyre, B.D., Baginska, B., 1999.
Review of techniques to estimate catchment exports. New South Wales
Environmental Protection Authority Report, EPA 99/73, .
MacKay, A.K., Taylor, M.P., 2012. Event-based water quality sampling method for
application in remote rivers. River Res. Appl. 28, 1105–1112.
Miller, J.R., Kochel, C.R., 2008. Characterization and evaluation of stream restoration
projects in North Carolina. Final Report to the North Carolina Clean Water
Management Trust Fund, CWMTF Project No.: 2002B-805, .
Miller, J.R., Kochel, C.R., 2013. Use and performance of in-stream structures for river
restoration: a case study from North Carolina. Environ. Earth Sci. Res. J. 68,
1563–1574.
Miller, J.R., Lord, M., Villarroel, L., Germanoski, D., Chambers, J., 2012. Structural
organization of process zones in upland watersheds of central Nevada and its
influence on basin connectivity, dynamics, and wet meadow complexes.
Geomorphology 139–140, 284–402.
Miller, J.R., Mackin, G., Lechler, P., Lord, M., Lorentz, S., 2013. Influence of basin
connectivity on sediment source, transport, and storage within the Mkabela
Basin, South Africa. Hydrol. Earth Syst. Sci. 17, 761–781.
Montgomery, D.R., Buffington, J.M., 1993. Channel Classification, Prediction of
Channel Response, and Assessment of Channel Conditions. Report TFW-SH10-
93-002. Washington State Timber, Fish, and Wildlife, Pullman, WA.
Montgomery, D.R., Buffington, J.M., 1997. Channel-reach morphology in mountain
drainage basins. Geol. Soc. Am. Bull. 109, 596–611.
Moore, D.S., 2009. “Manning’s Equation Analysis Spreadsheet” XsecAnalyzerVer9.
xlsm. USDA Natural Resources Conservation Service. .
Mukundan, R., Walling, D., Gellis, A., Slattery, M., Radcliffe, D., 2012. Sediment
source fingerprinting: transforming from a research tool to a management tool.
Journal of the American Water Works Association 48, 1241–1257.
Niemitz, J., Haynes, C., Lasher, G., 2013. Legacy sediments and historic land use:
chemostratigraphic evidence for excess nutrient and heavy metal sources and
remobilization. Geology 41, 47–50.
Pimentel, D., Harvey, C., Resosudarmo, P., Sinclair, K., Kurz, D., McNair, M., Crist, S.,
Shpritz, L., Fitton, L., Saffouri, R., Blair, R., 1995. Environmental and economic
costs of soil erosion and conservation benefits. Science 267, 1117–1123.
Richards, R.P., 2001. Estimation of Pollutant Loads in Rivers and Streams: a Guidance
Document for NPS Programs. Report to US Environmental Protection Agency,
Grant X998397-01-0, Region VIII. .
Ries III, K.G., 2006. The National Streamflow Statistics Program: A Computer
Program for Estimating Streamflow Statistics for Ungauged Sites; U.S.
Geological Survey Methods and Techniques, 4-A6. US Government Printing
Office, Washtington D.C p. 45.
Simmons, S.E., 1993. Sediment characteristics of North Carolina streams,1970-1979.
US Geological Survey Water-Supply Paper 2364. US Government Printing Office,
Washington D.C.
Simon, A., Klimetz, L., 2008. Magnitude, frequency, and duration relationships for
suspsended sediment in stable (“reference”) southeastern streams. J. Am. Water
Resour. Assoc. 44, 1270–1283.
Stokes, M.A., Smiley, T.L., 1996. An Introduction to Tree-Ring Dating. The University
of Arizona Press, Tucson, pp. 73.
Surasinghe, T., Baldwin, R.F., 2014. Ghost of land-use past in teh context of current
land cover: evidence from salamander communities in streams of Blue Ridge
and Piedmont ecoregions. Can. J. Zool. 92, 527–536.
Syvitski, J., Vörösmarty, C., Kettner, A., Green, P., 2005. Impact of humans on the flux
of terrestrial sediment to the global coastal ocean. Science 308, 376–380.
Trimble, S.W., 1974. Man-induced Soil Erosion in the Southern Piedmont 1700-1970.
Ankeny, Iowa, Soil Conservation Society of America, pp. 180.
USEPA, 2013. Watershed Assessment, Tracking, and Environmental Results:
National Summary of State Information. US Environmental Protection Agency.
Vercruysse, K., Grabowski, R.C., Rickson, R.J., 2017. Suspended sediment transport
dynamics in rivers: multi-scale drivers of termporal variation. Earth. Rev. 166,
38–52.
Waters, R., Merritts, D., Rahnis, M., 2007. Estimating Volume, Nutrient Content, and
Rates of Stream Bank Erosion of Legacy Sediment in the Piedmont and Valley
and Ridge Physiographic Provinces, Southeastern and Central PA. Report to
Pennsylvania Department of Environmental Protection, Harrisburg, PA.
Williams, G.P., 1989. Sediment concentrations versus water discharge during single
hydrologic events in rivers. J. Hydrol. (Amst) 111, 89–106.
Wood, P., Armitage, P., 1997. Biological effects of fine sediment in the lotic
environment. Environ. Manage. 21, 203–217.
Further reading
Miller, J.R., Sinclair, J.T., Walsh, D., 2015. Controls on suspended sediment
concentrations and turbidity within a reforested, southern appalachian
headwater basin. Water. 7, 3123-3148. doi:10.3390/w7063123.
J.R. Miller et al. / Anthropocene 25 (2019) 100188 15