Content uploaded by Colm M. Casserly
Author content
All content in this area was uploaded by Colm M. Casserly on Oct 17, 2020
Content may be subject to copyright.
Content uploaded by Colm M. Casserly
Author content
All content in this area was uploaded by Colm M. Casserly on Mar 22, 2020
Content may be subject to copyright.
Impact of low-head dams on bedload transport rates in
coarse-bedded streams
Colm M. Casserly
a,b,
⁎, Jonathan N. Turner
b
,JohnJ.O'Sullivan
a
, Michael Bruen
a
, Craig Bullock
c
,
Siobhán Atkinson
d
, Mary Kelly-Quinn
d
a
School of Civil Engineering, UCD Dooge Centre for Water Resources Research and UCD Earth Institute,University College Dublin, Dublin 4, Ireland
b
School of Geography and UCD Earth Institute, University College Dublin, Dublin 4, Ireland
c
School of Architecture, Planning and Environmental Policy, and UCD Earth Institute, University College, Dublin, Dublin 4, Ireland
d
School of Biology and Environmental Science, and UCD Earth Institute, University College Dublin, Dublin 4, Ireland
HIGHLIGHTS
•RFID tracer data was analysed at two
sites to assess sediment conveyance.
•Tracer fractions ND
90
can be carried
through and over low-head dams.
•Fractional transport rates and particle
size distributions also suggest supply-
limited conditions persist downstream.
•We propose a conceptual model and
mechanism to explain how low-head
dams may affect downstream convey-
ance indefinitely.
GRAPHICAL ABSTRACT
abstractarticle info
Article history:
Received 27 September 2019
Received in revised form 22 January 2020
Accepted 22 January 2020
Available online 24 January 2020
Editor: José Virgílio Cruz
Keywords:
Connectivity
Hydromorphology
RFID tracers
Run-of-River
Sediment
Weir
This paper presents an empirical study that uses the movement of RFID tracers to investigate the impacts of low-
head dams on solid transport dynamics incoarse-bedded streams. Here we report on the influence of two struc-
tures located in Ireland's South-East, both of which indicate that particles greater than the reach D
90
can be car-
ried through and over low-head dams.This observation suggests that both structures may have reached a state of
‘transient storage’as hypothesized by previous research. However, when the data were reinterpreted as frac-
tional transport rates using a novel application of existing empirical relations, we observed patterns consistent
with supply-limited conditions downstream. Expanding on existing conceptual models and mechanisms, we il-
lustrate how a system may continue to exhibit supply-limited conditions downstream without the need for a net
attenuation of sediment to occur indefinitely. We propose that once a transient storage capacity has been
reached, the system then enters a state of dynamic disconnectivity where the long-term average sediment flux
equals that under reference conditions, but now with the amplitude and wavelength of these sediment fluctua-
tions having increased. We hypothesize that the time-lag associated with the reduced frequency of events com-
petent enough to move bedload over the structure accounts for the time necessary to complete the ‘fill’phase of
the transient storage dynamic; a process thatwill continue until both thefill and flow thresholds are again met to
Science of the Total Environment 716 (2020) 136908
⁎Corresponding author at: School of Civil Engineering, UCD Dooge Centre for Water Resources Research and UCD Earth Institute, University College Dublin, Dublin 4, Ireland.
E-mail address: colm.casserly@ucdconnect.ie (C.M. Casserly).
https://doi.org/10.1016/j.scitotenv.2020.136908
0048-9697/© 2020 Elsevier B.V. All rights reserved.
Contents lists available at ScienceDirect
Science of the Total Environment
journal homepage: www.elsevier.com/locate/scitotenv
allow the system to reenter the ‘scour’phase. This model reconciles how asystem may exhibit a sediment deficit
for time intervals longer than those experienced under reference conditions. As water and sediment are the
drivers of channel morphology and associated habitat units, the impact a structure has on a channel's sediment
regime should therefore form part of any assessment regarding the prioritization of barriers for removal or
remediation.
© 2020 Elsevier B.V. All rights reserved.
1. Introduction
River connectivity can be described as the degree to which water, or-
ganisms and abiotic matter can move among spatially-defined units
within a river corridor or catchment, whether laterally (floodplains),
longitudinally (upstream/downstream), vertically (hyporheic zone) or
temporally (seasonality of flows) (Ward, 1989;Wohl, 2017;Grill
et al., 2019). In this context, manyrivers throughoutthe world have en-
dured long histories of alteration due to the construction of in-channel
barriers (e.g. dams, weirs, culverts, bridge aprons) for a variety of man-
agement purposes (Graf, 2005;Csiki and Rhoads, 2010;Lehner et al.,
2011). The resulting fragmentation of the drainage network has im-
peded the downstream passage of water, sediment and organic matter,
as well as the upstream movement of diadromous species and those
such as the freshwater pearl mussel (Margaritifera margaritifera) that
rely on salmonid hosts for distribution of glochidia (Newton et al.,
2008). Even for potamodromous fish species, barriers to connectivity
can affect the range of areas available to search for food or shelter
from predation, isolating populations and habitats by creating either
physical or thermal obstructions (Bednarek, 2001;Birnie-Gauvin et al.,
2017).
In Europe, restoration of natural aquatic conditions through the re-
moval of barriers is driven by the EU Water Framework Directive
2000/60/EC (WFD), which requires member states to protect, enhance
and restore all surface water bodies to good chemical and biological sta-
tus. The effects of large dams (N15 m in height) and their propensity to
trap sediment is relatively well documented (e.g. Brandt, 2000;Doyle
et al., 2002;Graf, 2005;Toniolo et al., 2007). However, 99.5% of the
16.7 million artificial barriers estimated to be fragmenting river systems
globally (Lehner et al., 2011;Jones et al., 2019) are substantially smaller
(b5 m in height) and hydraulically distinct from larger reservoir struc-
tures. These low-head dams (also referred to as weirs, overflow dams
and run-of-the-river dams) are characterized by heights that do not ex-
ceed the elevation of the channel banks and where the structure oc-
cupies the full width of the channel without entirely inhibiting the
discharge of water (Csiki and Rhoads, 2010;Gargan et al., 2011;
Pearson and Pizzuto, 2015). As almost all major rivers (N1000 km
long) in Europe are now fragmented by such structures (Grill et al.,
2019), there is a need to better understand their impact on
hydromorphology if channel restoration advocated by the WFD is to
be successfully implemented.
Sedimentconnectivity, though typically viewed as subsidiary to con-
cerns surrounding fish movement, serves an important role in a func-
tioning riverine ecosystem because it strongly influences the habitat
template upon which all aquatic biota must live (Belletti et al., 2017).
Substrate particle size distribution and stability are key determinants
of spawning habitat and benthic community structure (Louhi et al.,
2008;Riebe et al., 2014), soany change in the frequency and magnitude
of a channel's flow and sediment regime is potentially significant. A
structure's impact on bedload continuity should therefore form part of
any assessment regarding the prioritization of barriers for removal.
Existing research on the geomorphic impacts of low-head dams,
which remains limited and largely confined to US catchments (e.g.
Skalak et al., 2009;Csiki and Rhoads, 2010, 2014;Pearson et al., 2011;
Pearson and Pizzuto, 2015) suggest local scour, bed material coarsening
and the formation of downstream riffles are potential issues (Csiki and
Rhoads, 2010). In a US study of 15 low-head dams, Skalak et al.
(2009) observed some variation in median particle sizewith marginally
lower percentages of fines (b2 mm) in the downstream reach, but no
appreciable change in morphology. Similarly, Csiki and Rhoads (2014)
found few morphological or sedimentological discontinuities down-
stream of four low-head dams in Illinois. However, none of these field
studies measured bedload transport directly or provided a detailed in-
vestigation into the degree of, or mechanism for transport over these
structures.
At present there is a notable lack of empirical data published on the
impact low-head dams have on coarse sediment conveyance, nor is
there a set of replicable quantitative methodologies for doing so. This
study's primary objective is to address these shortcomings by develop-
ing an approach that uses RFID tracers and existing empirical relations
to help understand how these structures affect transport dynamics in
the field. We then utilize the empirical data from two Irish catchments
to present a conceptual model and mechanism for complex sediment
transport processes associated with low-head dams and explore the im-
plication for long-term sediment regimes.
2. Methods and materials
2.1. Regional setting and study sites
2.1.1. River Dalligan
The Dalligan Weir is located on the River Dalligan, a coarse bed-
ded 3rd order stream that drains an area of 20.2 km
2
from the south-
ern tip of the Monavullagh Mountains before entering the sea on the
south coast of Ireland in Co. Waterford (Fig. 1,andTable 1). The
2.3 m-high weir with a contracted ogee overflow spillway (Fig. 1a)
was built in 1967/ ’68. Land-use is predominantly pastural (52%)
while the uplands contain a high proportion of peat bog (20%) and
mixed forestry/semi-natural areas (21%). Catchment geology is
dominated by Ordovician volcanic rocks, while Devonian sand-
stones, mudstones and conglomerates characterise the headwaters.
The study reach containing the Dalligan Weir is characterized by a
riffle-pool morphology and is laterally confined by outcropping bed-
rock and dense ri parian vegetation. Much of the Dalligan upstream of
the weir has the potential to support Atlantic salmon (Salmo salar)
and Sea trout (Salmo trutta) and the river has been designated as
‘high’WFD status (EPA, 2018). However, the weir which is located
just 2.2 km from the sea is a significant barrier preventing migratory
fish from passing further upstream.
2.1.2. River Duag
The Shanrahan Weir is located on the River Duag, a meandering 4th
order cobble-bed stream that rises in the Kilworth Mountains and
drains an area of 65.9 km
2
north of the Knockmealdowns in Co. Tipper-
ary. Land-use is predominantly pastural (58%) while the uplands con-
tain peat-bog (17%) and mixed forestry/semi-natural areas (24%).
Catchment geology is dominated by Tournaisian limestone, while the
headwaters primarily consist of Devonian sandstones (Fig. 1,and
Table 1). The 1 m high stone crump weir was constructed pre-1840s
(Fig. 1b) without a fish pass. The small, heavily vegetated abstraction
channel shown on aerial photography immediately upstream of the
weir was observed during high flow conditions to have no velocity
2C.M. Casserly et al. / Science of the Total Environment 716 (2020) 136908
element and a negligible impact on water and sediment discharges
downstream. Upstream of the weir, the Duag is fed by a tributary (3rd
order) stream draining from the south-west (Fig. 2b), while a road
bridge upstream of the weir pond acts as a local hydraulic control. The
river has a ‘good’water quality status under the WFD and offers
spawning potential for Atlantic salmon (Salmo salar).
2.2. Experimental design
The experimental design employed for data collection at the
Dalligan (DAL) and Duag (DUG) sites is shown in Fig. 2a and b. Monitor-
ing was carried out at three distinct types of reach to allow the impacts
of low-head dam emplacement on coarse sediment transport to be iso-
lated. These were: (i) the impounded upstream reach (DAL
US
and
DUG
US
) above the structure; (ii) a downstream reach (DAL
DS
and
DUG
DS
) below the structure; and (iii) a reference reach (DAL
R
) located
in the same river but 670 m upstream of any hydraulic influence of
the weir. As pre-construction sediment transport data are unavailable
and land-use is both spatially and temporally constant between reaches,
this study makes the reasonable assumption that the reference reach on
the Dalligan is representative of pre-dam conditions regarding channel
characteristics and sediment regime (sensu Stanley et al., 2002;Skalak
et al., 2009;MacVicar et al., 2015). Due to the presence of a coarse-
bedded tributary channel immediately above the Shanrahan Weir and
the relatively finer particle distribution above the impoundment, it
was not possible to find an upstream reach on the Duag that would
allow for the seeding of tracers that were sufficiently representative of
reference conditions without introducing a confounding variable. How-
ever, for comparison, particle size distributions for the tributary channel
(DUG
Trib
) and a reach 300 m above the impoundment (DUG
AI
) were
also surveyed. Each study site was sub-divided into transects (mean
spacing ≈37 m) to capture channel geometry, bedform units and bed
surface particle size distributions. Full topographic surveys were com-
pleted forboth sites using an integrated DGPS (Trimble R6/R10) and ro-
botic total station (Trimble S7). In addition to longitudinal and cross-
sectional geometries, dimensions of channel structures (e.g. weirs and
bridges) were recorded.
2.3. Establishing stage-discharge rating and modelling bankfull parameters
Water levels at the two study sites were continuously recorded at
15-min intervals using an Impress IMSL IP68 water-level recorder
(a)
(b)
Fig. 1. Location of the Dalligan and Duag sub-basins (left). (a) weir on the River Dalligan and (b) the Shanrahan Weir on the River Duag.
Table 1
Main characteristics of the Dalligan and Duag sub-basins.
River Dalligan River Duag
Drainage area above
weir [km
2
]
17.7 59.3
Basin elevation range
above weir [m]
31.7–602.4 52.8–651.2
Stream order 3rd 4th
Mean bankfull channel
width
a
[m]
9.4 11.6
Channel slope [m/m] 0.013 0.003
Sinuosity 1.14 1.17
Dominant channel
morphology
Riffle-Pool Riffle-Pool
Petrology Igneous extrusive, tuffs,
conglomerates and sedimentary
Till derived from
sandstones
Dominant land use Pasture/Peat Bog/Forestry Pasture/Peat
Bog/Forestry
Annual recorded
precipitation [mm]
(1st Feb 2018–31st Jan
2019)
1043 1081
WFD status High Good
Structure type Contracted-ogee spillway Crump
Structure height [m] 2.3 1
Structure width [m] 6.5 (notch width), 57.7 (max.
width)
30 (max. width)
Year constructed 1967/ ‘68 Pre-1840s
a
Mean bankfull channel width exclude unrepresentative cross-sections (e.g. those
closest to structure).
3C.M. Casserly et al. / Science of the Total Environment 716 (2020) 136908
secured at a geometrically stable cross-section. Monitoring stations
were located in the downstream reaches, c. 225 m and c. 115 m below
the Dalligan and Shanrahan Weirs, respectively (Fig. 2aandb).A
stage-discharge rating curve wasgenerated at both sites through veloc-
ity measurements using a Valeport Electromagnetic Flow Meter (Model
801) for lower flows (0.5–3.5 m
3
/s) when thechannel was wadable. For
the Dalligan site, discharge at intermediate flows (0.49–4.16 m
3
/s) was
also calculated using the theoretical ogee weir equation (with contrac-
tion) in conjunction with upstream water level measurements taken
from a staff gauge installed in a still water location in the weir pond.
As the crump weir on the River Duag was unsuitable for this method,
additional measurements in this river were captured using a Teledyne
Acoustic Doppler Current Profiler (ADCP). Stage-discharge relationships
for both sites were subsequently refined and extended using a synthetic
rating curve developed in the one-dimensional hydraulic modelling
package HEC-RAS (U.S. Army Corps of Engineer's Hydrologic Engineer-
ing Centre River Analysis System, version 5.0.5; Brunner, 2016). Appro-
priate hydraulic resistances (Manning's roughness) for the hydraulic
model were estimated using a range of methods (Barnes, 1849;
Cowan, 1956;Chow, 1959;Hollinrake and Samuels, 1995). Reach-
averaged bankfull discharge (Q
bf
) was estimated through HEC-RAS sim-
ulations following the approach of Mulvihill et al. (2009).HEC-RASwas
also used to generate estimates of bed shear stress and water surface
slopes.
2.4. Bedload characterisation and RFID tracers
Bedload calibre (b-axis) was determined for both rivers using
Wolman (1954) pebble counts (DAL n= 1660, DUG n= 1919). Counts
were conducted on a homogenous section of the bed within a
(a)
(b)
Fig. 2. (a) River Dalligan and (b) River Duag study areas, designating locations of study reaches, hydraulic structures, cross-sections and geomorphic units seeded with tracers. Cross-
section lengths are exaggerated for clarity.
4C.M. Casserly et al. / Science of the Total Environment 716 (2020) 136908
rectangular swathe up to one channel-width wide at selected cross-
sections. Other distinctive geomorphic units (GMUs) such as mid-
channel and side bars (Kondolf, 1997a) not captured along the transects
were also surveyed. This allowed both reach and individual GMU parti-
cle size distributions to be established (Fig. 3a and b). Transport dis-
tances of mobilised particles were determined using radio frequency
identification (RFID) tagging of particles collected from GMUs for each
monitoring reach (Autumn 2017). Once tags were inserted, particles
were returned to their parent GMU. As channel morphology has been
reported to influence displacement patterns and travel paths of tagged
stones (Papangelakis and Hassan, 2016), qualitative efforts were made
to ensure the proportion of GMUs seeded represented the prevailing
geomorphic setting of the reach being monitored. Tagged particles
were placed in unconstrained positions in sets of three (triads), with
each triad comprising a particle representative of the GMU-specific
D
35
,D
50
and D
84
values based on particle size distribution data (Fig. 3a
and b). Each GMU contained between two to six ‘triads’,withtheseeded
particles positioned c. 0.2 m around a georeferenced center point. A rep-
resentative particle was defined as one that fell within a suite of
established particle size ranges (full phi intervals) based on the Went-
worth scale (Quinlan et al., 2015). This protocol was developed to in-
crease the accuracy of inter-reach comparisons of bed mobility as the
tracers now represented the channel bed both spatially and in distribu-
tion. During the winter of 2017 a total of 155 and 138 tracers were de-
ployed in the Dalligan and Duag rivers, respectively. Tracers were
deployed within three distinct types of reaches (Section 2.2.) in order
to isolate theimpact the structure has on bedload transport. A full resur-
vey of tracer locations was carried out during the summer of 2018 (for
seeding andrecovery dates see Table 2) using an Oregon RFID® loop an-
tenna with a maximum detection range of c. 0.5 m (dependant on tag
orientation). The new position of recovered tracers was georeferenced
using a robotic total station (Trimble S7).
2.5. Data analysis
2.5.1. Workflow for estimating sediment conveyance
Treating the system as a sediment ‘conveyor belt’(Ferguson, 1981;
Kondolf, 1997b) the procedure adopted in this paper aims to quantify
sedimentstorage through an analysis of the observed differences in sed-
iment flux entering and leaving the system. Flood event bedload flux is
frequently assessed in coarse bedded streams using portable samplers
or pit-style sediment traps (e.g. Sterling and Church, 2002). However,
these methods typically sample sediment flux at fixed cross-sections
and do not capture the temporal and spatial variability that
characterises bedload transport (Hoey, 1992;Haschenburger and
Church, 1998). The workflow deployed in this study (Eqs. (1)–(9))
overcomes these limitations by using information on the virtualvelocity
of particle movement, porosity and density of bed material, and the
width, depth and portion of the active-channel bed. Virtual velocity re-
fers to the totaldistance travelled by individual particles over a monitor-
ingperiod,dividedbytheestimatedtimedurationofcompetentflows;
a period that in this study incorporates multiple flood events (Hassan
et al., 1992;Haschenburger and Church, 1998). The mass rate of bed
material transport (Q
b
) is expressed in this study by Eq. (1). This repre-
sents the ratio between the mass of sediment mobilised during a trans-
port episode and the time duration of the transport episode (Hassan
et al., 1991;Haschenburger and Church, 1998;Schneider et al., 2014;
Vázquez-Tarrío and Menéndez-Duarte, 2014)
Qbi ¼Li
twsds1−pðÞρsPmð1Þ
where Q
bi
is the bed material transport rate [kg/s] for a specific particle
size class range, L
i
is the mean travel distance of bed particles [m] in that
size range, tis the time duration of competent flows over the whole
monitoring period [s], w
s
is the reach mean active-channel width [m],
d
s
is the reach-averaged active-channel depth [m] calculated for
bankfull flows, ρ
s
is the mass density of mineral particles [taken as
2650 kg/m
3
] and pis sediment porosity. Based on bedload petrology
we took a value of 0.2 as a reasonable estimate of sediment porosity
(Vázquez-Tarrío and Menéndez-Duarte, 2014). This study defines
mean reach active-channel width (w
s
) as the break in bank slope that
typically marks the edge of permanent vegetation (Lawlor, 2004).
Values were determined using cross-sectional data and onsite evalua-
tion. As Haschenburger and Church's (1998) formulae assumes no spa-
tial variability in the active portion of the bed, this study has included
the additional variable of reach-average particle mobility (P
m
) as pro-
posed by Wilcock (1997). This is to account for partial transport condi-
tions where a portion of the particles on the stream bed remain
immobile over the duration of the study period; the veracity of which
is heavily dependent on the representativeness of tracers both in size
and spatial distribution. Particle mobility is expressed as
Pm¼Nm
Nf
ð2Þ
where N
m
is the number of mobile tracer particles, and N
f
is the total
number of recovered tracer particles (MacVicar and Roy, 2011). As
mean active-channel depth (d
s
) was not measured in the field (using
scour chains etc.), it was estimated indirectly using the empirically de-
rived scour and fill depth model (Haschenburger, 1999;sensuSloat
(a)
(b)
Fig. 3. Cumulative particle size distribution curves for surface material at the (a) Dalligan
and (b) Duag. Particle size fractions represented by tracers are also shown for each study
reach (markers). The total quantity of tracers within each fraction is; 32–45 mm (DAL =
14, DUG = 15), 45–64 mm (DAL = 61, DUG = 53), 64–90 mm (DAL = 22, DUG = 41),
90–128 mm (DAL = 30, DUG = 17), 128–181 mm (DAL = 27, DUG = 12),
181–256 mm (DAL = 1, DUG = 0).
5C.M. Casserly et al. / Science of the Total Environment 716 (2020) 136908
et al., 2017). Developed for single-thread gravel-bed channels (6 to
20 m wide), Haschenburger (1999) defines scour depth [cm] as a func-
tion of reach-average excess Shields stress ðτ=τcÞ
ds¼3:33 e−1:52τ=τc
−1ð3Þ
As we are considering the reach-averaged scour potential for
bankfull flows, Shields stress can be defined as
τ¼τbf
ρs−ρw
ðÞgD50
½ ð4Þ
where gis the acceleration due to gravity [9.81 m/s
2
], ρ
w
is the density of
water [1000 kg/m
3
], D
50
is the median particle size for the reach [m],
and τ
bf
is the reach-average bed shear stress [N/m
2
] at bankfull (esti-
mated usingHEC-RAS). Critical Shields stress ðτcÞis predicted as an em-
pirical function of reach-average slope following the equation of Lamb
et al. (2008)
τc¼0:15 S0:25 ð5Þ
where we take Sas the water surface slope at bankfull [m/m].
2.5.2. Calculating time duration of competent flows
In order to determine the time duration of competent flows (t), it
was necessary to establish a reasonable threshold discharge under
which bed mobility could be expected to occur. Once determined the
cumulative duration of the transport period could be taken as the time
interval between the first and last moments on the hydrograph for
when these values were exceeded. Bedload transport in gravel-
bedded rivers is thought to only occur at a very low rate up until a cer-
tain critical flow condition has beenreached (Ferguson, 2012). Once this
critical threshold has been exceeded, transport increases at a greater
than-linear rate with flow (Ferguson, 2005). As bedload transport dis-
tances and rates are functions of hydraulic forcing, either shear stress
(τ) (which depends on a channel's depth to slope product, Meyer-
Peter and Muller, 1948) or unit stream power (ω)(Bagnold, 1980)
can be used to predict bedload transport and the critical threshold for
when incipient motion occurs.Stream power is typically easier to calcu-
late when direct flow depth data required to calculate averaged critical
shear stresses is unavailable (Ferguson, 2005). For this reason, the unit
stream power [W/m
2
] which quantifies the rate of loss of potential en-
ergy available to perform geomorphic work per unit area is utilised
here. As the smallest fraction consistently represented by tracer parti-
cles, this study has estimated the critical unit stream power (ω
c
)neces-
sary to mobilise each of the reach specificD
35
particle sizes. As ω
c
is a
reach aggregated value that cannot fully account for the local variability
in shear stress, the returned value should only be considered an indica-
tor of the relative differences in critical flow conditions between reaches
rather than an absolute value below whichmobility never occurs. There-
fore, we have taken this threshold to represent a lower bound from
which a single reach-specific value of tcould be derived for in inclusion
in Eq. (1) and the estimation of Q
bi
for all size fractions.
By taking into account the variability of critical shear stress (τ
ci
)
Parker et al. (2011) expressed Ferguson's (2005) formulae for estimat-
ing ω
c
as
ωc¼τci 8:2τci=ρwgS
ðÞðÞ
Db
1
6ffiffiffiffiffiffi
τci
ρw
r
2
6
6
43
7
7
5
ð6Þ
taking D
b
[m] as a particle representative of the bed material (D
84
).
Parker et al. (2011) also expressed average critical shear stress (τ
ci
)
for the particle size of interest (D
i
)as
τci ¼ 0:19 S0:28! ρs−ρw!gDið7Þ
where the particle size of interest D
i
=D
35
[m].
Once the reach-averaged critical unit stream power (ω
c
) had been
determined, Eq. (8) was used to solve for the critical discharge (Q
c
)
ωc¼ρwgQcS
wbf
ð8Þ
where w
bf
is defined as the reach-averaged bankfull width (as distinct
Table 2
Summary statistics on particle size, tracer movement and flow properties for each monitoring reach
a
.
Study reach No. of
X-sections
Geomorphic units
seeded
Date
seeded
Date
recovered
No. of
tracers
Tracer size range
[mm]
D
35
[mm]
D
50
[mm]
D
84
[mm]
No.
recovered
Detection rate
[%]
River Dalligan
Reference 4 2 02/12/17 15/07/18 24 36–164 36 59 147 23 96
Upstream 8 5 03/12/17
b
16/07/18 74 34–128 33 44 86 72 97
Downstream 9 5 04/12/17 17/07/18 57 34–190 35 68 164 57 100
River Duag
Upstream 7 4 03/11/17 04/07/18 63 40–165 41 63 115 62 98
Downstream 9 9 04/11/17 05/07/18 75 33–158 46 61 105 74 99
Study reach P
m
L
D50
[m] ± Standard
Error
L
Max
[m]
S[m/m] d
s
[cm]
W
s
[m] ±Standard
Error
W
bf
[m] ±Standard
Error
Q
Max
[m
3
/s] ω
c
[W/m
2
]Q
c
[m
3
/s] t[h, min]
River Dalligan
Reference 0.61 28.7 ± 12.2 75.3 0.01328 4.16 7.7 ± 0.6 9.3 ± 0.6 8.8 53.1 3.8 18 h,
15 min
Upstream 0.68 17.2 ± 7.5 60.5 0.00641 1.85 6.8 ± 0.4 9.8 ± 0.8 8.8 39.9 6.2 8 h, 15 min
Downstream 0.42 9.5 ± 6.1 17 0.01086 2.54 7.1 ± 0.4 9.2 ± 0.3 8.8 47.1 4.1 14 h,
45 min
River Duag
Upstream 0.73 18.4 ± 6.6 96.7 0.00357 1.17 11.3 ± 1.6 12.9 ± 1.7 34.8 47.4 17.5 66 h,
Downstream 0.55 13.7 ± 6.1 109.9 0.00402 1.37 8.9 ± 0.4 10.7 ± 0.6 34.8 60.1 16.3 69 h,
15 min
a
Particle mobility (P
m
); Mean distance travelled by particles in the median surface particle size class range (L
D50
); maximum travel distance (L
Max
); bankfull water surface slope (S);
activedepth (d
s
); active width (w
s
) and bankfullwidth (w
bf
) excludeunrepresentativecross-sections; maximumestimated discharge (Q
Max
); estimatedcritical unitstream power of reach
D
35
(ω
c
); estimated critical discharge for reach D
35
(Q
c
); time duration that critical discharge was exceeded (t).
b
Eight of these tracers were seeded in the DAL
US
on 30-03-2018 (b efore ω
c
was exceeded) and were therefore included in analysis.
6C.M. Casserly et al. / Science of the Total Environment 716 (2020) 136908
from w
s
). We use w
bf
as it is our closest approximation of the wetted-
channel width at Q
c
. In line with the estimation of reach-averaged
active-channel width (w
s
), unrepresentative cross-sections such as
those immediately adjacent to the dam structure were omitted in the
estimation of w
bf
. In this study it is assumed Q
c
is constant for a given
reach over the entire study period. However, in practice this will vary
slightly over time depending on local variations in particle size, addi-
tional inputs of sediment and changes in bed configuration
(Rickenmann, 2001). Once Q
c
has been determined, the total time (t)
over the monitoring period where Q≥Q
c
can be calculated.
2.5.3. Calculating scaled transport distances for individual particle size
fractions
Particle travel distances L
i
(Eq. (1)), were calculated from the mean
transport distances of individual displaced tracer particles within a
given particle size range from a known starting position to their recov-
ery location on the re-survey. Following MacVicar and Roy (2011),
Bradley and Tucker (2012), and MacVicar et al. (2015), movement in
excess of 1 m defined particle mobility. This threshold was defined as
twice the antenna's detection radius (c. 0.5 m). However, as the tracer
population only reflects the movement of some size classes, it was nec-
essary to infer the distances travelled by the other size classes (sensu
Vázquez-Tarrío and Menéndez-Duarte, 2014). Travel distances for par-
ticle sizes not included in the RFID tagging were estimated using a fitted
relation based on, and then compared to the original non-dimensional
relationship of Church and Hassan (1992) for riffle-pool streams,
where scaled transport (L
∗
) distances are expressed as a function of
scaled particle sizes (D
∗
)
L¼Li
LD50Sur f
¼1:77 1−log10
Di
D50Sur f
1:35
ð9Þ
where D
50Surf
is the median particle size of the streambed surface mate-
rial, L
D50Surf
is the mean distance travelled by particles in the median
surface particle size class range, and L
i
is the mean recorded transport
distance of individual particles in the class size range of diameter D
i
(i.e. the mean diameter of mobile tracers within a narrow size range).
In the original formulae Church and Hassan (1992) used the median
particle size of the subsurface material instead of the surface (D
50Surf
),
as we do in this study. However, they implied D
50
surface particle size
could be used in its place as seen in the work of other authors
(Ferguson and Wathen, 1998;Vázquez-Tarrío and Menéndez-Duarte,
2014;MacVicar et al., 2015). To preserve the original form of the rela-
tion found by Church and Hassan (1992) and to account for the propor-
tionally finer particles that characterise subsurface material, scaled
particle size (D
∗
=D
i
/D
50Surf
) in our dataset was multiplied by a factor
of 2.2 (sensu Wilcock, 1997;Vázquez-Tarrío and Menendez-Duarte,
2014). As the original Church and Hassan (1992) relation cannot be as-
sumed to represent all rivers or datasets exactly (Vázquez-Tarrío et al.,
2019), we fitted a separate curve to each of our monitoring reaches.
This allowed an idealised projected transport distance for each half-
phi interval (from D
16
to maximum displaced tracer particle) to be de-
termined for inclusion in Eq. (1) following the approach of Vázquez-
Tarrío and Menendez-Duarte (2014). One potential methodological
constraint is that the use of unconstrained tracers may overestimate
particle mobility and travel lengths compared to in-situ bed material.
However, as the original relation developed by Church and Hassan
(1992) was restricted to unconstrained tracers we feel the approach is
valid here. In addition, a recent analysis of 33 tracer studies by
Vázquez-Tarrío et al., (2019) found that unconstrained data tends to
closely overlap with data for constrained studies where the tracers
have been seeded for less than three years.
2.5.4. Time-sensitivity, uncertainty and sources of error
Estimates of fractional transport rates derived from Eqs. (1)-(9) in this
study are weighted on the estimated duration of competent flows, so an
appreciable source of uncertainty is contained in our approximation of
flow magnitude and the threshold determined for particle motion. How-
ever, because this study is interested in the relative differences in transport
rates rather than absolute values, it was only important to ensure that the
thresholdsestimatedforeachreachwereproportionaltoeachother(e.g.
using reach D
35
). As we are also treating transport rates as a proxy indica-
tor of sediment supply into and out of the system, the use of a reach-
specificvalueofQ
c
may disguise the actual differences in total load mov-
ing through each reach. During our analysis, to test the time-sensitivity of
these results and to compare inter-reach rates for the monitorng period as
a whole, we also ran calculations where we assumed parity for the time
variable (t)inEq.(1), taking the lowest estimated Q
c
value as a common
threshold for the whole river. As many of the variables in Eq. (1) are either
constrained by scientific understanding or field observation (e.g. seasonal
variation in Manning's n), we also performed a sensitivity analysis to illus-
trate a degree of uncertainty (sensu Pizzuto et al., 2007;Pearson and
Pizzuto, 2015). This was achieved by conservatively increasing and de-
creasing each variable by 25% to account for potential error in both the
discharge rating curve and the variables used in Eq. (1).
3. Results
3.1. Bedload characterisation and tracer mobility
Cumulative particle size distribution curves and the size ranges rep-
resented by tracer particles are shown in Fig. 3a and b. Median particle
sizes (D
50
) ranged from 59 to 68 mm at the Dalligan and 61–63 mm at
(a)
(b)
Fig. 4. Plots showing ti me series of discha rge and reach-averaged critical discharge
thresholds for particle mobility estimated for the (a) River Dalligan and (b) River Duag
over the monitoring period.
7C.M. Casserly et al. / Science of the Total Environment 716 (2020) 136908
the Duag (DUG
Trib
= 50 mm and DUG
AI
= 42 mm). The median particle
size was shown to be 15% coarser in the DAL
DS
relative to DAL
R
,and45%
and 22% coarser in the DUG
DS
relative to DUG
AI
and DUG
Trib
, respec-
tively. In order to determine if the presence of a low-head dam was hav-
ing an impact on bed surface composition we proposed the following
null hypothesis (H
0
): that particle size distributions between the down-
stream reach and the reach(es) supplying the impoundment were sim-
ilar. At both sites differences in particle size distributions were observed
to be statistically significant, therefore allowing the null hypothesis to
be rejected (Two-Sample Kolmogorov-Smirnov Test; DAL
R
vDAL
DS
(D = 0.11, p= .013), DUG
Trib
vDUG
DS
(D = 0.15, pb.01), DUG
AI
v
DUG
DS
(D = 0.24, pb.01). Size fractions where particle sizes deviated
most notably were found in the coarser (ND
50
) size fractions at the
Dalligan (Fig. 3a). Conversely, the greatest differences at the Duag
were observed among the sub-D
35
size fractions (Fig. 3b). Particle size
distributions were also found to be statistically significant between
the DAL
US
vDAL
DS
,(D=0.29,pb.01), but not significant between
the DUG
US
vDUG
DS
reaches, (D = 0.07, pN.05).
The size-range of tracer particles that were observed to have moved
(N1 m) over the monitoring period were 34–190 mm and 33–142 mm
at the Dalligan and Duag, respectively. Particle detection rates
were N95% at both sites while mobility (P
m
) rates were identified as
ranging from 0.42 to 0.68 at the Dalligan and 0.55 to 0.73 at the Duag.
The lowest mobility rates at both sites were found in DAL
DS
and
DUG
DS
. Mean travel distances of the reach median particle size (L
D50
)
ranged between 28.7 m and 9.5 m at the Dalligan and 18.4 m and
13.7 m at the Duag. Maximum travel distances (L
Max
) ranged from
75.3 m (DAL
R
) and 109.9 m (DUG
DS
). A full set of summary statistics
on particle size and tracer movement is presented in Table 2.
3.2. Hydrology and threshold for incipient motion
From Eqs. (6)–(8), estimates for reach-averaged critical discharge
(Q
c
) for incipient motion ranged from 3.8 m
3
/s to 6.2 m
3
/s at the
Dalligan, and 16.3 m
3
/s to 17.5 m
3
/s at the Duag. Three flood events
(DL1 - DL3) were recorded at the Dalligan and six at the Duag (SH1 -
(a)
(b)
Fig. 5. Spatial distribution of tracer movement over the structures on the (a) Dalligan and (b) Duag. Initial triad seeding positions (yellow triangles) and final recovery positions (red
circles) are indicated. Travel distances were calculated along the channel centre line for tracers that moved N1 m, while arrows for illustration purposes only indicate straight line
travel of tracers that moved N2m.
8C.M. Casserly et al. / Science of the Total Environment 716 (2020) 136908
SH6) which were estimated to have the ability to disturb the reach D
35
value (Fig. 4a and b). Maximum estimated discharge (Q
Max
)was
8.8 m
3
/s and 34.8 m
3
/s at the Dalligan and Duag, respectively. The
time durations (t) when discharge either equalled or exceeded Q
c
are
shown in Table 2. Rainfall data collected at both sites (Table 1) suggest
that the net rainfall experienced was comparable with local and na-
tional mean annual rainfall records (Noone et al., 2016).
3.3. Transport over low-head dams
Eleven tracer particles passed over the Dalligan Weir (Fig. 5a) with a
size range of 39–110 mm, constituting 29% of all the tracers seeded
within the first 50 m upstream of the structure. Between initial place-
ment and recovery, eight clasts passed over the weir on the River
Duag (Fig. 5b) with a size range of 42–142 mm, with six of the eight
clasts remaining in the plunge pool beneath the structure. The eight
tracers that passed over the structure constitute 24% of all the tracers
seeded within the first 50 m upstream of the dam. Two of the eight
tracers travelled 96.7 m and 88.7 m downstream of the structure and
became deposited on an unvegetated sidebar. Despite the presence of
a substantial mid-channel bar located 8 m below the structure, no
tracers were recovered there at the time of resurvey.
3.4. Scaled transport distances and fractional transport rates
Scaled transport distances for the Dalligan and Duag are shown in
Fig. 6a and b, respectively. A single clast in the finest fraction from
DAL
DS
(shown in Fig. 6a) was not included in the calculations because
this clearly did not conform to the pattern displayed by the data. The re-
sults thereafter reflect the classic convex shape reported by Church and
Hassan (1992) that suggests movement of smaller particles are primar-
ily dependent on the relativetrapping efficiency in the bed, while move-
ment of larger particles is limited by size. The R
2
values for all five
reaches showed moderate to very good fits (Dalligan: R
2
=0.58–0.88,
Duag: R
2
=0.77–0.93) allowing for the estimation of transport dis-
tances for unrepresented particle size fractions. Fractional transport
rates for all size ranges from D
16
to the maximum tracer mobilised are
presented in Fig. 7a-d. The results show that transport rates decrease
within all study reaches with increasing particle size, indicating size se-
lective and partial transport. Transport rates across all but the coarsest
size fractions differ between monitoring reaches, with lowest transport
rates observed in the downstream reachesat both sites. However, as we
have used a common threshold for incipient motion in the calculation of
tfor each particle size, the relative difference in transport rates within
study reaches may be overstated. A comparison of fractional transport
rates for the median particle sizes (Fig. 7a and b) indicate that the
DAL
DS
experienced 85% and 76% less transport relative to both the
DAL
R
and DAL
US
respectively. Rates experienced in the DUG
DS
were
48% less than those observed in DUG
US
(this difference increases to
53% if the tracers seeded just upstream of the tributary confluence are
excluded). When calculations were performed assuming parity for the
time variable (t)inEq.(1), fractional transport rates between monitor-
ing reaches (Fig. 7c and d) show patterns identical to those in Fig. 7aand
b. Rates for the median particle sizes at the DAL
DS
experienced 88% and
57% less transport relative to both the DAL
R
and DAL
US
respectively.
Similarly, rates experienced in the DUG
DS
were 45% less than those ob-
served in DUG
US
(a difference that increases to 51% if the tracers up-
stream of the tributary are likewise excluded). These observations
indicate that the inter-reach variability in the Q
c
necessary to entrain
the D
35
appears to have a minor influence on the relative transport
rates when compared to the other variables used in Eq. (1).
3.5. Simulated cross-sectional bed shear stress
Results generated from HEC-RAS model simulations run at steady
flow rates from 1 m
3
/s up to the estimated Q
Max
indicate that the lowest
bed shear stress zones during low flows in the upstream (impounded
reach) are predominantly situated closest to the weir structure at both
sites (Fig. 8a and b). Results at the Dalligan (Fig. 8a) show eight
(a)
(b)
Fig. 6. Scaled total tracer transport di stance as a function of scaled particle size at the
(a) River Dalligan and (b) River Duag. Each point represents the mean tr ansport
distance of all displaced tracers that have fallen within a given half-phi interval (defined
as the mean diameter of these tracers). Curves represent the logarithmic fitforeach
individual monitoring reach. The original fit found by Church and Hassan (1992) is also
shown. Thesingle tracer that was removedfrom analysis is indicated by an Xin panel (a).
9C.M. Casserly et al. / Science of the Total Environment 716 (2020) 136908
occasions where a reversal of shear stress at cross-sections closer to the
weir is observed to overtake cross-sections further upstream as dis-
charge increases. Only on two occasions did the opposite occur before
terminal discharge was reached. Results at the Duag (Fig. 8b) exhibit
eight instances of this same pattern. Once the weir on the Duag becomes
drowned out (c. 16 m
3
/s) the overall trend in increasing shear stress
shows evidence of either plateau or decline, indicating a gradual return
to relative shear stress in four instances.
4. Discussion
4.1. Interpretation of field observations and fractional transport rates
This paper presents an empirical study that uses the movement of
RFID tracers to investigate the impact low-head dams have on coarse
sediment conveyance in gravel-cobble streams. Here we report on the
influence of two structures, both of which indicate that particles greater
than the reach D
90
can be carried through and over low-head dams.
These field observations suggest that both systems may have reached
a state of ‘transient storage’as hypothesized by other authors
(Pearson and Pizzuto, 2015). However, when particle size distribution
data were examined and tracer movement reinterpreted as fractional
transport rates, we observed patterns consistent with supply-limited
conditions downstream, demonstrating conflicting lines of evidence be-
tween the event-scale tracer movement and long-term sediment re-
gime. Expanding on existing conceptual models and mechanisms, we
discuss here how a system may continue to exhibit supply-limited
conditions downstream without the need for a net attenuation of sedi-
ment to occur indefinitely.
At present conceptual models of unfilled reservoir dams typically
suggest c. 100% trapping efficiency regarding bedload material
(Toniolo et al., 2007;Csiki and Rhoads, 2010). This reduction in supply
to the downstream reach can result in often pronounced bed and
bank scour as the sediment starved river tries to entrain material
equal to its carrying capacity (Csiki and Rhoads, 2010). In contrast, re-
cent studies have reported little evidence that low-head dams have
any long-term impact on downstream morphology (Skalak et al.,
2009;Csiki and Rhoads, 2014). Pearson and Pizzuto (2015) in their
modelling of a 200-year-old low-head dam on the Red Clay Creek ob-
served that only 25% of the structure's accommodation space was filled
with sediment. They proposed that the impoundment eventually
adopted a morphology that promotes a pattern of ‘transient storage’
and export of coarse material. Relying on numerical models and the
similarity of particle size distributions found in the mid-channel bars
commonly found beneath such structures, they hypothesized that
these reservoirs may actually function like long pools that adjust their
bed texture and elevation to periodically move bed material over the
structure. Although no evidence for deposition was observed in the
mid-channel bar located beneath the Shanrahan Weir (no bar was pres-
ent at the Dalligan), results show that tracer particles with a greater di-
ameter than the upstream D
90
can pass over both weirs. These
observations provide evidence in support of their argument and appear
to complement an unpublished study recently conducted in the United
States (Magilligan and O'Brien, 2018, AGU Abstract).
Fig. 7. Fractional bedload transport ratesfor each monitoringreach at the (a) River Dalligan and (b)River Duag when takingaccount of the inter-reach variability in Q
c
, and at the (c) River
Dalliganand (d) River Duag when a common Q
c
is assumedfor each river. Errorbars represent the maximum increase and decrease in transport ratesresulting from a sensitivityanalysis
(+/- 25%) of any of the variables used in Eq. (1).
10 C.M. Casserly et al. / Science of the Total Environment 716 (2020) 136908
Evidence that particles are passing over both structures may be con-
sidered indicative that a degree of transient storage has been already
achieved at both sites, suggesting a lack of any meaningful
disconnectivity. However, a comparison of fractional transport rates
suggests a marked reduction in observed sediment flux in both down-
stream reaches relative to rates either entering or moving within their
respective upstream reaches (Fig. 7a-d). Furthermore, coarser particle
size distributions recorded downstream relative to those supplying
the impoundment were observed to be statistically significant at both
sites. We argue that fractional transport rates and particle sizedistribu-
tions are indicators of the prevailing long-term transport regime,
whereas tracers passing over the dam tell us about event-scale dynam-
ics. Therefore, when viewed within the context of a sediment ‘conveyor
belt’(Ferguson, 1981) these data suggest supply-limited conditions re-
main at least locally, downstream. Our interpretation of this pattern
assumes conditions above the impoundment can be considered a func-
tion of the reference sediment supply (S
ref
), water discharge (Q
ref
),
transport capacity (T
ref
) and particle size distribution (PSD
ref
). Condi-
tions in the impounded reach are therefore a function of this identical
water discharge and sediment supply, but also a reduced transport ca-
pacity due to the presence of the dam's backwater effect (bT
ref
). This re-
duction in transport capacity translates into a reduced sediment supply
(bS
ref
) to the downstream reach as sediment is deposited behind the
dam. Below the structure, this reduction in sediment supply results in
an initial increase in local transport capacity (NT
ref
), which leads to a
progressive winnowing of finer material causing a net coarsening of
the bed relative to those seen under reference conditions (NPSD
ref
)
(sensu Kondolf, 1997b). Once bed surface composition in the down-
stream reach has adjusted to this new transport regime, it will thereaf-
ter take proportionally higher water discharge values relative to our
reference rate (NQ
ref
) to transport the same portion of the bed. This
model would therefore predict that a structure still acting as an imped-
iment to sediment conveyance would exhibit a pattern where (i) there
is proportionally less bedload transport in the downstream reach rela-
tive to the rates either entering or moving within the impounded
reach for the same water discharge over a given monitoring period,
and (ii) the downstream reach would exhibita coarser particle sizedis-
tribution relative to reaches supplying the system.Both study sites meet
these two conditions.
4.2. Mechanism for transport into and out of the impoundment
In order to obtain a degree of transient storage that is punctuated by
extended periods of ‘fill’a unique mechanism is required that allows
more material to be passed over the structure than is being supplied
into the impoundment under the same flows. Otherwise any flows com-
petent enough to pass material over the dam would also be competent
enough to transport equal or more material into the impoundment. Re-
sults generated from HEC-RAS model simulations for both upstream
reaches (Fig. 8a and b) indicate that the lowest bed shear stress zones
during low flows are predominantly situated closest to the weir struc-
tures. As discharge increases mean bed shear stresses at both sites are
observed to increase at a greater rate among the cross-sections located
closest to the weir relative to those further upstream on eight occasions.
These datasuggest that a mechanism analogous to the ‘velocity reversal
hypothesis’developed by Keller (1971) and attributed to the mainte-
nance of riffle-pool sequences (Keller, 1971;Thompson et al., 1999;
Milan et al., 2001) may be occurring here. As river stage increases, the
tractive forces nearest the low-head dam periodically increase at a
rate greater than thoseexperienced above the impounded area. We hy-
pothesisthat this mechanismmay explain how bedload can periodically
pass over the structure while also leaving enough space available for a
subsequent period of ‘fill’to occur.
4.3. Long-term impact of low-head dams on sediment regime
Though not hydraulically identical, Piton and Recking (2017) pro-
posed a model that described the long-term impact check dams have
on a channel's sediment regime. Using small scale flume experiments,
they proposed that an initial period of near total bed-load trapping
would occurafter initial dam construction,followed by a pattern of tem-
porary storage and later release of sediment over the structure. They
concluded that instantaneous transport intensity would not be altered,
but that peak sediment discharge would occur more frequently and
for shorter time durations. This they argued, would result in a decreased
volume for single transport events yet still maintain the same total
transport over the long-term. This model may explain the pattern of
transient storage proposed by Pearson and Pizzuto (2015) as it main-
tains the same long-term average sediment flux to the downstream
reach. However, this mechanism may not be applicable to low gradient
alluvial channels that do not experience the flash floods and intense
(a)
(b)
Fig. 8. Cross-sectional bed shear stress plotted against increasing discharge for th e
(a) River Dalligan and (b) River Duag upstream reaches. Each cross-section is numbered
(U1 to U6), with the distance upstream of the weir shown in the legend [m]. White Xs
show reversal in relative shear stress between cross-sections and the vertical dashed
line denotes upstream reach-averaged bankfull discharge (Q
bf
).
11C.M. Casserly et al. / Science of the TotalEnvironment 716 (2020) 136908
solid transport that characterise steep mountain streams (SN2%). Fur-
thermore Piton and Recking's (2017) model if applied to low-land sys-
tems would also suggest that the processes of export (scour) and
subsequent attenuation (fill) of material should occur within a time
window shorter than the system would be able to transport the same
material under reference conditions, nor does it offer an explanation
for the persistent supply-limited conditions indicated by our results.
Here we propose an alternate model that explains how our results
could simultaneously exhibit both a pattern of transient storage at the
event scale, but prevailing supply-limited conditions downstream. For
context, the well-documented impact that large reservoir dams have
on net sediment transport through time is shown in Fig. 9a. Here the
pre-reservoir dam sediment regime (reference condition) is repre-
sented as a steady-state equilibrium over the relevant temporal scales
of human concern (decades-centuries). This steady-state represents
the mean net sediment supply around which seasonal transport rates
fluctuate. This mean net supply is permanently reduced downstream
post-structure construction as c. 100% of sediment is trapped in the res-
ervoir. A return to the reference sedimentregime seen prior to dam con-
struction will only occur far enough downstream at the “point of
concentration recovery”(Chien, 1985;Csiki and Rhoads, 2010)or
when the structure is removed. Our model also represents a similar re-
duction in net supply downstream immediately after low-head dam
construction (Fig. 9b) as the backwater effect causes a reduction in ve-
locity and a subsequent loss of transport capacity that promotes deposi-
tion in the upstream end of the backwater zone. In agreement with
Pearson and Pizzuto (2015) we also propose that a sediment ramp is
progressively formed as coarse material builds up behind the structure
during a recovery period until a transient storage capacity is reached. In-
tuitively, it would be assumed that once this transient storage capacity
has been attained, then all incoming coarse material would make its
way over the structure at a rate consistent with that under reference
conditions, as logically aggradation cannot continue indefinitely. Evi-
dence presented in this study suggest that this may not be the case, as
supply-limited conditions have been observed downstream, at least at
the temporospatial scale examined in this study. Here we propose that
once a transient storage capacity has been reached the system enters a
state of dynamic disconnectivity (Wohl et al., 2019) where the long-
term average sediment flux equals that under reference conditions sim-
ilar to that suggested by Piton and Recking (2017), only now amplitude
and wavelength of these fluctuationshave increased. We proposed that
a reduction in the frequency of events competent enough to export ma-
terial over the dam accounts for the time duration necessary to com-
plete the ‘fill’phase of the transient storage dynamic, a process that
will continue until both the fill and flow thresholds are again met to
allow the system to reenter the ‘scour’phase. The time-lag associated
with the reduced frequency of events, but increased sediment discharge
during these events, reconciles how a system may exhibit a supply-
limited regime downstream for time durations longer than those expe-
rienced under reference conditions. We also hypothesize that the rela-
tive change in wavelength resulting from dam emplacement (i.e. time
lag between scour events) is a function of dam height and its potential
to become drowned out. We believe this may explain why the magni-
tude of supply-limitation seen at the Dalligan is considerably greater
than that observed at the Duag.
4.4. Implications and conclusion
Using field data derived from the movement of RFID tracers, com-
bined with a novel application of established empirical relations, this
Fig. 9. Conceptual model illustrating sediment transport regime before and after dam construction downstream of a (a) large reservoir dam and (b) low-head dam. Data presented is
hypothetical.
12 C.M. Casserly et al. / Science of the Total Environment 716 (2020) 136908
paper investigates the transport dynamics of two gravel-cobble streams
and the implications of low-head dam emplacement on longitudinal
bedload flux. Despite evidence that particles of all seeded size fractions
(up to D
90
) can pass over these structures, a comparison of particle size
distributions and fractional bedload transport rates reveal that supply-
limited transport conditions remain downstream at both sites.This indi-
cates that low-head dams may continue to alter the hydrosedimentary
processes of fluvial systems long after dam construction. We hypothe-
size that the magnitude of supply-limitation is predominantly a func-
tion of dam height and the structure's propensity to become drowned
out under high flows. These empirical field observations form the
basis of a conceptual model of transient storage that builds on earlier
theoretical (Pearson and Pizzuto, 2015)andflume (Piton and Recking,
2017) studies. This model is consistent with emerging ideas that view
river connectivity as a continuum of states ranging from fully connected
to complete disconnectivity (Wohl et al., 2019), where disconnectivity
may only be defined as such over short timescales.
As a channel's prevailing flow and sediment regime is a determinant
of physical habitat and benthic community structure, a reduction in the
temporal frequency of sediment transport may prove to be ecologically
significant. For example, spawning habitat which is sensitive to changes
in sediment loading and flow regime, needs to be coarse enough to en-
courage the through flow of oxygen-rich water and be capable of
resisting fluvial scour, but also fine enough for fish to move sediment
and build redds (Kondolf and Wolman, 1993;Louhi et al., 2008;Riebe
et al., 2014). The impact low-head structures have on sediment
disconnectivity to the downstream reach is likely to be relatively local-
ized, with any habitat deficit confined to the reach immediately below
the structure. What has yet to be adequately quantified, however, is
the accumulative impact these structures may have in catchments
where barriers remain prolific. Lehner et al. (2011) reported that
there may be N16.6 million low-head s tructures worldwide impounding
an estimated 2.11 × 10
4
km
2
of channel. If the extent of downstream
habitat loss is even a fraction of that observed as a result of the imposi-
tion of lentic conditions above low-head dams (Santucci et al., 2005),
then the accumulative loss in high density catchments is likely to be
substantial and warrants further study.
Declaration of competing interest
The authors declare that they have no known competing financial
interests or personal relationships that could have appeared to influ-
ence the work reported in this paper.
Acknowledgements
This research was funded by the Irish Environmental Protection
Agency as part of the Reconnect project (Grant Number: 2015-W-LS-
8). We would like to acknowledge the kind permission granted to us
by the landowners whose property the study sites run through and to
Inland Fisheries Ireland for facilitating access. We are also indebted to
Seosamh O'Coileir and to a great number of lab technicians,research as-
sistants and students over the course of this study for their assistance in
both the field and workshop. Finally, we would like to thank two anon-
ymous peer reviewers for their constructive feedback and valuable
comments on the paper.
References
Bagnold,R.A., 1980. An empirical correlation of bedload transportrates in flumes and nat-
ural rivers. Proceedings of the Royal Society of London. Series A, Mathematical and
Physical Sciences 372 (1751), 453–473.
Barnes, H.H.,1849. “Roughness Characteristics of Natural Channels”U.S Geological Survey
Water Supply Paper. U.S Geological Survey, Washington, DC, p. 1967.
Bednarek, A.T., 2001. Undamming rivers: a review of the ecological impacts of dam re-
moval. Environ. Manag. 27, 803–814. https://doi.org/10.1007/s002670010189.
Belletti, B., Rinaldi, M., Bussettini, M., Comiti, F., Gurnell, A.M., Mao, L., Nardi, L., Vezza, P.,
2017. Characterising physical habitats and fluvial hydromorphology: a new system
for the survey and classification of river geom orphic units. Geomor phology 283,
143–157. https://doi.org/10.1016/j.geomorph.2017.01.032.
Birnie-Gauvin, K., Aarestrup, K., Riis, T.M.O., Jepsen, N., Koed, A., 2017. Shining a light on
the loss of rheophilic fish habitat in lowland rivers as a forgotten consequenceof bar-
riers, and its implications for management. Aquat. Conserv. Mar. Freshwat. Ecosyst.
27, 1345–1349.
Bradley,D.N., Tucker, G.E., 2012.Measuring gravel transport and dispersion in a mountain
river using passive radio tracers: tracking gravel with PIT tags. Earth Surf. Process.
Landf. 37, 1034–1045. https://doi.org/10.1002/esp.3223.
Brandt, S.A., 2000. Classification of geomorphological effects downstream of dams. Catena
40, 375–401. https://doi.org/10.1016/S0341-8162(00)00093-X.
Brunner, G., 2016. HEC-RAS, River Analysis System User’s Manual, Version 5.0, February
2016. U.S. Army Corps of Engineers Hydrologic Engineering Center (HEC) (CDP-69).
Chien, N., 1985. Changes in river regime after the construction of upstream reservoirs.
Earth Surf. Process. Landf. 10, 143–159.
Chow, V.T., 1959. Open Channel Hydraulics. McGraw-Hill, New York, p. 1959.
Church, M., Hassan, M.A., 1992. Size and distance of travel of unconstrained clasts on a
streambed. Water Resour. Res. 28, 299–303.
Cowan, W.L., 1956. Estimating hydraulic roughness coefficients. Agric. Eng. 37, 473–475.
Csiki, S., Rhoads, B.L., 2010. Hydraulic and geomorphological effects of run-of-river dams.
Progressin Physical Geography: Earth and Environment34, 755–780. https://doi.org/
10.1177/0309133310369435.
Csiki, S.J., Rhoads, B.L., 2014. Influence of four run-of-river dams on channel morphology
and sediment characteristics in Illinois, USA. Geomorphology 206, 215–229. https://
doi.org/10.1016/j.geomorph.2013.10.009.
Doyle, M.W., Stanley, E.H., Harbor, J.M., 2002. Geomorphic analogies for assessing proba-
ble channel response to dam removal. JAWRA Journal of the American Water Re-
sources Association 3 8, 1567–1579.
EPA (Environmental Protection Agency), 2018. Catchment science and management unit.
Colligan Mahon Catchment Assessment 2010–2015 (HA 17), Version No. 3.
Ferguson,R.I., 1981. Channel forms and channel changes. In: Lewin, J. (Ed.), British Rivers.
Allen and Unwin, London, pp. 90–125.
Ferguson, R.I., 2005. Estimating critical stream power for bedload transport calculations in gravel-
bed rivers. Geomorphology 70, 33–41. https://doi.org/10.1016/j.geomorph.2005.03.009.
Ferguson, R.I., 2012. River channel slope, flow resistance, and gravel entrainment thresh-
olds. Water Resour. Res. 48. https://doi.org/10.1029/2011WR010850.
Ferguson, R.I., Wathen, S.J., 1998. Tracer-pebble movement along a concave river profile:
virtual velocity in relation to grain size and shear stress. Water Reso ur. Res. 34,
2031–2038. https://doi.org/10.1029/98WR01283.
Gargan, P.G., Roche, W.K., Keane, S., King, J.J., Cullagh, A., Mills, P., O’Keeffe, J., 2011. Com-
parison of field- and GIS-based assessments of barriers to Atlantic salmon migration:
a case study in the Nore catchment, Republic of Ireland. J. Appl. Ichthyol. 27, 66–72.
https://doi.org/10.1111/j.1439-0426.2011.01855.x.
Graf, W.L., 2005. Geomorphology and American dams: the scientific, social,and economic
context. Geomorphology 71, 3–26. https://doi.org/10.1016/j.geomorph.2004.05.005.
Grill, G., Lehner, B., Thieme, M., Geenen, B., Tickner, D., Antonelli, F., Babu, S., Borrelli, P.,
Cheng, L., Crochetiere, H., Ehalt Macedo, H., Filgueiras, R., Goichot, M., Higgins, J.,
Hogan, Z., Lip, B., McClain, M.E., Meng, J., Mulligan, M., Ni lsson, C., Olden, J.D.,
Opperman, J.J., Petry, P., Reidy Liermann, C., Sáenz, L., Salinas-Rodríguez, S., Schelle,
P., Schmitt, R.J.P., Snider, J., Tan, F., Tockner, K., Valdujo, P.H., van Soesbergen, A.,
Zarfl, C., 2019. Map ping the world’sfree-flowing rivers. Na ture 569, 215–221.
https://doi.org/10.1038/s41586-019-1111-9.
Haschenburger, J.K., 1999. A probability modelof scour and fill depths in gravel-bed chan-
nels. Water Resour. Res. 35, 2857–2869.
Haschenburger, J.K., Church, M., 1998. Bed material transport estimated from the virtual
velocity of sediment. Earth Surf. Process. Landf. 23, 791–808.
Hassan, M.A., Church, M., Schick, A.P., 1991. Distance of movement of coarse particles in
gravel bed streams. Water Resour. Res. 27, 503–511.
Hassan, M.A., Church, M., Ashworth, P.J., 1992. Virtual rate and mean distance of travel of
individual clasts in gravel-bed channels. Earth Surf. Process. Landf. 1, 617–627.
Hoey, T., 1992. Temporal variations in bedload transport rates and sediment storage in
gravel-bed rivers. Progress in Physical Geography: Earth and Environment 16,
319–338. https://doi.org/10.1177/030913339201600303.
Hollinrake, P., Samuels, P.G., 1995. Flood discharge assessment: an engineering guide. HR
Wallingford Ltd, Wallingford, UK, Report SR 379.
Jones, J., Börger,L., Tummers, J., Jones, P.,Lucas, M., Kerr, J.,Kemp, P., Bizzi,S., Consuegra, S.,
Marcello, L., Vowles, A., Belletti, B., Verspoor, E., Van de Bund, W., Gough, P., Garcia de
Leaniz, C., 2019. A comprehensive assessment of stream fragmentation in Great Brit-
ain. Sci. Total Environ. 673, 756–762. https://doi.org/10.1016/j.scitotenv.2019.04.125.
Keller, E.A., 1971. Areal sorting of bed-load material: the hypothesis of velocity reversal.
Bull. Geol. Soc. Am. 82, 753–756. https://doi.org/10.1130/0016-7606(1971)82[753:
ASOBMT]2.0.CO;2.
Kondolf, G.M., 1997a. Application of the pebble count: notes on purpose, method, and
variants. J. Am. Water Resour. Assoc. 33, 79–87.
Kondolf, G.M., 1997b. Profile: hungry water: effects of dams and gravel mining on river
channels. Environ. Manag. 21, 533–551. https://doi.org/10.1007/s002679900048.
Kondolf, G.M., Wolman, M.G., 1993. The sizes of salmonid spawning gravels. Water
Resour. Res. 29, 2275–2285.
Lamb, M.P.,Dietrich, W.E., Venditti, J.G.,2008. Is the critical shieldsstress for incipient sed-
iment motion dependent on channel-bed slope? Journal of Geophysical Research-
Earth Surface 113 (F2) (pp. F02008-n/a).
Lawlor, S.M., 2004. Determination of Channel-Morphology Characteristics, Bankfull Dis-
charge, and Various Design-Peak Discharges in Western Montana. U.S. Geolog ical
Survey Scientific Investigations Report vols. 2004–5263.
Lehner, B., Liermann, C.R., Revenga, C., Vörösmarty, C., Fekete, B., Crouzet, P., Döll, P.,
Endejan, M., Frenken, K., Magome, J., Nilsson, C., Robertson, J.C., Rödel, R., Sindorf,
13C.M. Casserly et al. / Science of the TotalEnvironment 716 (2020) 136908
N., Wisser, D., 2011. High-resolution mapping of the world’s reservoirs and dams for
sustainable river-flow management. Front. Ecol. Environ. 9, 494–502. https://doi.org/
10.1890/100125.
Louhi, P., Mäki-Petäys, A., Erkinaro, J., 2008. Spawning habitat of Atlantic salmon and
brown trout: general criteria and intragravel factors. River Res. Appl. 24, 330–339.
MacVicar, B., Chapuis, M., Buckrell, E., Roy, A., 2015. Assessing the performance of in-
stream restoration projects using Radio Frequency Identification (RFID) transpon-
ders. Water 7, 5566–5591. https://doi.org/10.3390/w7105566.
MacVicar, B.J., Roy, A.G., 2011. Sediment mobility in a forced riffle-pool. Geomorphology
125, 445–456. https://doi.org/10.1016/j.geomorph.2010.10.031.
Magilligan, F. J., O'Brien, M., Marti, M., Renshaw, C. E. (2018). “The impact of run-of-river
dams on sediment longitudinal connectivity and downstream channel equilibrium”.
American Geophysical Union, Fall Meeting 2018, Abstract #EP33B-03.
Meyer-Peter, E., Muller,R., 1948. Formulas for bed-load transport. Proceedings of the 2nd
Meeting of the International Association for Hydraulic Structures Research. Interna-
tional Association for Hydraulic Structures Research, Stockholm, pp. 39–64.
Milan, D.J.,Heritage, G.L., Large, A.R.G., Charlton, M.E.,2001. Stage dependent variability in
tractive force distr ibution thr ough a riffle–pool sequence. Catena 44, 85–109. https://
doi.org/10.1016/S0341-8162(00)00155-7.
Mulvihill, C.I., Baldigo, B.P., Miller, S.J., DeKoskie, D., DuBois, J., 2009. Bankfull Discharge
and Channel Characteristics of Streams in New York State. U.S. Geological Survey Sci-
entific Investigations Report vols. 2009–5144.
Newton, T.J., Woolnough,D.A., Strayer, D.L., 2008. Using landscape ecology to understand
and manage freshwater mussel populations. J. N. Am. Benthol. Soc. 27, 424–439.
https://doi.org/10.1899/07-076.1.
Noone, S., Murphy, C., Coll, J., Matthews, T., Mullan, D., Wilby, R.L., Walsh, S., 2016. Ho-
mogenization and analysis of an expanded long-term monthly rainfall network for
the Island of Ireland (1850–2010). Int. J. Climatol. 36, 2837–2853.
Papangelakis, E., Hassan, M.A., 2016. The role of channel morphology on the mobility and
dispersion of bed sediment in a small gravel-bed stream. Earth Surf. Process. Landf.
41, 2191–2206.
Parker, C., Clifford, N.J., Thorne, C.R., 2011. Understanding the influence of slope on the
threshold of coarse grain motion: revisiting critical stream power. Geomorphology
126, 51–65. https://doi.org/10.1016/j.geomorph.2010.10.027.
Pearson, A.J., Pizzuto, J., 2015. Bedload transport over run-of-river dams, Delaware, U.S.A.
Geomorphology 248, 382–395. https://doi.org/10.1016/j.geomorph.2015.07.025.
Pearson, A.J., Snyder, N.P., Collins, M.J., 2011. Rates and processes of channel response to
dam removal with a sand-filled impoundment. Water Resour. Res. 47.
Piton, G., Recking, A., 2017. Effects of check dams on bed-load transport and steep-slope
stream morphodynamics. Geomorphology 291, 94–105. https://doi.org/10.1016/j.
geomorph.2016.03.001.
Pizzuto, J., Moglen, G., Palmer, M., Nelson, K., 2007. Two model scenarios illustrating the
effects of land use and climate change on gravel riverbeds of suburban Maryland,
U.S.A. Developments in Earth Surface Processes, pp. 359 –381. https://doi.or g/
10.1016/S0928-2025(07)11133-0.
Quinlan, E.,Gibbins, C.N., Batalla, R.J., Vericat, D., 2015. Impacts of small scale flow regula-
tion on sediment dynamics in an ecologically important upland river. Environ.
Manag. 55, 671–686. https://doi.org/10.1007/s00267-014-0423-7.
Rickenmann, D., 2001. Comparison of bed load transport in torrents and gravel bed
streams. Water Resour. Res. 37, 3295–3305.
Riebe, C.S., Sklar, L.S., Overstreet, B.T., Wooster, J.K., 2014. Optimal reproduction in salmon
spawning substrates linked to gra in size and fish length . Water Resour. Res. 50,
898–918.
Santucci, V.J., Gephard, S.R., Pescitelli, S.M., 2005. Effects of multiple low-head dams on
fish, macroinvertebrates, habitat, and water quality in the Fox River, Illinois. N. Am.
J. Fish Manag. 25, 975–992. https://doi.org/10.1577/M03-216.1.
Schneider, J.M., Turowski, J.M., Rickenmann, D., Hegglin, R., Arrigo, S., Mao, L., Kirchner,
J.W., 2014. Scaling relationships bet ween bed load volumes, transport distances,
and stream power in steep mountain channels. Journal of Geo physical Research:
Earth Surface 119, 533–549.
Skalak, K., Pizzuto, J., Hart, D.D., 2009. Influence of small dams on downstream channel
characteristics in Pennsylvania and Maryland: implications for the long-term geo-
morphic effects of dam removal. J. Am. Water Resour. Assoc. 45, 97–109.
Sloat, M.R., Reeves, G.H., Christiansen, K.R., 2017. Stream network geomorphology medi-
ates predicted vulnerability of anadromous fish habitat to hydrol ogic change in
southeast Alaska. Glob. Chang. Biol. 23, 604–620.
Stanley, E.H., Luebke, M.A., Doyle, M.W., Marshall, D.W., 2002. Short-term changes in
channel form and macroinvertebrate communities following low-head dam removal.
J. N. Am. Benthol. Soc. 21, 172–187. https://doi.org/10.2307/1468307.
Sterling, S.M., Church, M., 2002. Sediment trapping characteristics of a pit trap and the Helley-
Smith sampler in a cobble gravel bed river. Water Resour. Res. 38, 1144–19–11.
Thompson, D.M., Wohl, E.E., Jarrett, R.D., 1999. Velocity reversals and sediment sorting in
pools and riffles controlled by channel constrictions. Geomorphology 27, 229–241.
https://doi.org/10.1016/S0169-555X(98)00082-8.
Toniolo, H., Parker, G., Voller, V., 2007. Role of ponded turbidity currents in reservoir trap
efficiency. J. Hydraul. Eng. 133, 579–595.
Vázquez-Tarrío, D., Menéndez-Duarte, R., 2014. Bedload transport rates for coarse-bed
streams in an Atlantic region (Narcea River, NW Iberian Peninsula). Geomorphology
217, 1–14. https://doi.org/10.1016/j.geomorph.2014.04.015.
Vázquez-Tarrío, D., Recking, A., Liébault, F., Tal, M., Menéndez-Duarte, R., 2019. Particle
transport in gravel-bed rivers: revisiting passive tracer data: particle transport in
gravel-bed rivers. Earth Surf. Process. Landf. 44, 112–128.
Ward, J.V., 1989. The four-dimensional nature of lotic ecosystems.J. N. Am. Benthol. Soc.8,
2–8. https://doi.org/10.2307/1467397.
Wilcock, P.R., 1997. Entrainment, displacement and transport of tracer gravels. Earth Surf.
Process. Landf. 22, 14.
Wohl, E., 2017. Connectivity in rivers. Progress in PhysicalGeography: Earth and Environ-
ment 41, 345–362. https://doi.org/10.1177/0309133317714972.
Wohl, E., Brierley, G.,Cadol, D., Coulthard, T.J., Covino, T., Fryirs, K.A., Grant, G., Hilton,R.G.,
Lane, S.N., Magilligan, F.J., Meitzen, K.M., Passalacqua, P., Poeppl, R.E., Rathburn, S.L.,
Sklar, L.S., 2019 . Connectivity a s an emergent property of geomorphic systems.
Earth Surf. Process. Landf. 44, 4–26.
Wolman, M.G., 1954. A method of sampling coarse river-bed material. Eos, Transactions
American Geophysical Union 35 (6), 951–956. https:/ /doi.org/10.10 29/
TR035i006p009 51.
14 C.M. Casserly et al. / Science of the Total Environment 716 (2020) 136908