close

Вход

Забыли?

вход по аккаунту

?

Fine grain sediment transport over a coarse grain gravel bed

код для вставкиСкачать
FINE GRAIN SEDIMENT TRANSPORT OVER A
COARSE GRAIN GRAVEL BED
by
ANTHONY JOHN LAGRECA
B.A., University of Colorado, 2009
A thesis submitted to the
Faculty of the Graduate School of the
University of Colorado in partial fulfillment
of the requirement for the degree of
Master of Arts
Department of Geography
2009
UMI Number: 1476956
All rights reserved
INFORMATION TO ALL USERS
The quality of this reproduction is dependent upon the quality of the copy submitted.
In the unlikely event that the author did not send a complete manuscript
and there are missing pages, these will be noted. Also, if material had to be removed,
a note will indicate the deletion.
UMI 1476956
Copyright 2010 by ProQuest LLC.
All rights reserved. This edition of the work is protected against
unauthorized copying under Title 17, United States Code.
ProQuest LLC
789 East Eisenhower Parkway
P.O. Box 1346
Ann Arbor, MI 48106-1346
This thesis entitled:
Fine Grain Sediment Transport Over a
Coarse Grain Gravel Bed
written by Anthony John LaGreca
has been approved for the Department of Geography
John Pitlick
Suzanne Anderson
Jonathan Nelson
Date
The final copy of thesis has been examined by the signatories, and we
Find that both the content and the form meet acceptable presentation
Standards of scholarly work in the above mentioned discipline.
ABSTRACT
Anthony John Lagreca (M.A., Geography)
FINE GRAIN SEDIMENT TRANSPORT OVER A COARSE GRAIN GRAVEL BED
Thesis directed by Professor John Pitlick
Flume experiments were conducted on a sand and gravel bed to understand the effects of surface
sand concentrations on flow properties and examine the effectiveness of bed-load transport for
flushing flows. Fine sediment infiltration is a common occurrence in dammed rivers, or rivers
with an altered hydrograph, intrusion of fine sediments can be harmful to benthic invertebrates
and degrade fish spawning habitat. Experiments were conducted using a 6m flume with a
working width of 0.3m. Velocity and turbulence profiles were collected with a Laser Doppler
Velocimeter, sediment transport was measured with a trap, the bed was photographed and the
photos analyzed to yield the surface sand cover. Results show that near-bed shear stress
decreases by half as the open work gravel bed was filled with sand. Two flushing runs were
designed with flow levels ranging from those capable of sand transport to those capable of gravel
transport. Results show the fully sand covered bed persisted in a steady state until the sediment
supply was depleted, after which the bed flushed the surface sand. Flows capable of sand
transport flushed 30-40% of the surface sand, flows capable of gravel transport flushed up to
80% of the sand. In all cases bed load transport was not sufficient to fully clean the bed.
Additional analysis suggests that flushing flows require stress levels 10-30% higher than τcg to
flush the bed to values of less than 30% sand cover. The data also show that higher-discharge
flows require significantly less water, than low discharge flows to achieve similar flushing
results.
iii
DEDICATION
This thesis is for my parents Robert and Cindy LaGreca. To my father for
taking me on all of those wonderful hikes in the Rocky Mountains and talking with
me about everything around us, from the glacial moraines to life on other planets.
Those hikes and conversations did more to excite me about science than any class I
have ever taken. To my mother for constantly encouraging, prodding and
sometimes nagging me to work hard and live up to my potential ( I am not sure that
I have done it yet). Mom, the total support and love you have given me throughout
my life is nothing short of extraordinary. I love you both very much and couldn’t
have done it without you.
ACKNOWLEDGMENTS
I would like to thank my advisor John Pitlick for somehow wanting me to come back to
school and for helping me develop the idea, design and analysis for this thesis. Much thanks for
the many challenging hours trying to read my early drafts of this thesis and your helpful
comments and criticism. Most importantly, thank you for teaching me about the river.
Thanks to Jon Nelson for letting me use the flume, teaching me how to use it and not
getting really mad when I broke it. Thanks also for your valuable insights on the design of the
experiments and the analysis of the data.
Thanks to everyone at the Sediment Transport Laboratory for making me feel at home
and letting me keep my lunch in the fridge. In particular, I would like to thank Paul Kinzel for
programing the Lab View software and general help with little things. I would also like to thank
Brandy Logan for her enthusiasm, interest and helpful insight into my work.
Big thanks to Erich Mueller for listening to me ramble on about blue sand and gravel for
more hours than I would like to admit. Those brainstorming sessions provided invaluable help to
this process.
Huge thanks to Anna LaGreca for teaching me how to become a computer programmer,
reading the many horrible drafts of the thesis and for being the love of my life.
Last I would like to thank Ween for providing the soundtrack and to the movie and song
“Weird Science”.
v
CONTENTS
CHAPTER
I.
II.
PAGE
INTRODUCTION
1
Background
1
Previous Work
2
Objectives
4
METHODS AND
EXPERIMENTAL DESIGN
5
Experimental Facilities
III.
IV.
and Instrumentation
5
Procedures
8
Data Analysis
10
ANALYSIS AND RESULTS
17
Filling Experiments
17
First Flushing Experiment
20
Second Flushing Experiment
27
Management Implications
32
SUMMARY AND CONCLUSIONS
REFERENCES
36
39
vi
Chapter 1
Introduction
1.1 Background
The world’s rivers have been significantly altered by human activity. A
primary consequence of anthropogenic river alteration is the intrusion and
accumulation of fine grained sediments on the river bed. Fine grained sediment
(fines) in a river system is often defined as particles <2 mm in diameter. Fines are a
significant portion of the sediment load for gravel bed rivers; however, regulated
rivers are often plagued with an excess of fines.
There are two main processes that cause a river bed to clog with fine
sediment. First, disturbances in the watershed caused by logging, mining, agriculture,
overgrazing and urban development can lead to an increase in the amount of fine
sediment delivered to a river system (Owens et al., 2005). Second, flow regulation
can reduce the transport capacity of a river leading to deposition of fine sediment on
the bed and banks and, potentially, changes in channel morphology (Karlinger et al.,
1983; Andrews, 1986; Van Steeter and Pitlick, 1998; Allred and Schmidt, 1999).
Excess fine sediment has adverse effects on river ecosystems. Fine
sediment in the water column can block sunlight, and sediment on the bed can reduce
attachment area for algae, limiting primary production (Van Nieuwenhuyse and
LaPerriere, 1986; Osmundson and Scheer, 1998; Osmundson et al., 2002). Benthic
invertebrate populations suffer if fine grained deposits create a seal in the gravel bed
and diminish hyporheic exchange (Wood and Armitage, 1997; Baxter and Hauer,
2000). A fine sediment seal can also reduce oxygen supplies necessary for salmonid
eggs and alevin to develop properly (Kondolf et al., 1993; Sear, 1993; Kondolf,
2000). Additionally sediment intrusion into the gravel matrix can also block alevin
emergence (Hawke 1978). Finally, loss of complexity due to sedimentation can lead
1
to a loss of backwater habitat important to river fish (Tyus and Karp, 1989; Gurtin et
al., 2003).
One common solution to the ecological problems introduced by excessive
fine sediment is to provide controlled releases to mobilize the fines and flush them
from the river system (Reiser et al., 1989; Kondolf and Wilcock, 1996; Whiting,
2002). Often the water available for environmental maintenance flows is limited, and
reservoir operators may be reluctant to release water for this purpose unless they can
be assured that releases will have the desired ecological or geomorphological affects.
Large magnitude releases that suspend the fine sediments and mobilize the gravel
matrix are often recommended to flush fines from the system; however, these high
stage releases can come at a high financial cost due to losses of storage, potential
power production and irrigation water. These releases may also flush large amounts
of gravel from the system, causing habitat degradation. Given these costs, reservoir
operators may choose to use smaller magnitude releases where fine sediment is
partially or wholly transported as bed load. It is important to understand how
effective this mode of transport is for flushing flows. Near-bed flow properties, the
sediment supply and the speed of sediment transport are factors likely to influence the
effectiveness of these flushing flows.
1.2 Previous Work
1.2.1 Infiltration and Flushing Studies
Earlier flume studies have shown that sediment transport mode and fine
sediment supply are important factors controlling the fine sediment infiltration and
flushing. Several flume studies have shown that fines will infiltrate a clean gravel
bed regardless of the transport mode (Beschta and Jackson, 1979; Diplas and Parker,
1992). Fine sediment infiltration and deposition increases with supply until the bed is
fully sand covered (Diplas and Parker, 1992). Flushing flow studies have shown that
flows capable of suspending the fines but incapable of transporting the gravel matrix
can only partially flush the bed surface and cannot clean fines from the gravel
2
interstices or the bed subsurface (Beschta and Jackson, 1979; O’Brien, 1987; Diplas
and Parker, 1992; Diplas, 1994). Only active transport of the gravel matrix has been
shown to remove fines from the interstices and the bed subsurface (Diplas and Parker,
1992; Wu and Chou, 2003). It is important to note that in the flushing experiments
outlined above, no additional fine sediment was supplied and the fines were primarily
transported in suspension.
1.2.2 Near-Bed Flow Studies
It is important to understand how changes in bed sediment properties
affect the near-bed flow and potentially the flushing process. As a gravel bed
transitions from a smooth (sand covered) to a rough (clean gravel) state, near-bed
velocities decrease slightly while near-bed shear stress increases substantially
(Sambrook Smith and Nicholas, 2005). Flow over a rough bed experiences a much
higher range in turbulent velocity fluctuations, drag forces and lift forces, which
increases the likelihood of sediment transport (Schmeeckle et al., 2007; Nelson et al.,
1995). Particle exposure can also affect the efficacy of flushing flows. Particles
hidden in the gravel interstices experience less than one quarter of the drag forces
applied to an exposed particle on the bed surface and also have increased friction
angles, decreasing the likelihood of transport (Schmeeckle and Nelson, 2003;
Schmeeckle et al., 2007).
1.2.3 Effects of Near-Bed Flow Properties on Sediment Transport
Changes in bed composition and the associated changes found in
turbulence, force and stress are important to sediment transport and ultimately
flushing flows. Nickling and McKena Neuman (1995) covered marbles with a layer
of sand in a wind tunnel to simulate transport over an immobile rough bed. The
authors found that as the roughness elements were exposed, the sediment flux was
elevated above that of the fully sand covered bed. Similarly, Grams and Wilcock
(2007) investigated the effects of bed sand cover on fine sediment entrainment and
transport rates using a flume with an artificially roughened bed. They demonstrated
that transport rates exceeded a rate predicted by the amount of sand present on the
3
bed surface, and that less sand cover on the bed resulted in higher near-bed suspended
sediment concentrations than a bed with more sand cover. This tubulence/transport
response indicates that as roughness increases, fine sediment flushing may be
enhanced; however, hiding effects and increased friction angles may provide a limit
to flushing. In their wind tunnel experiments Nickling and McKena Neuman (1995)
found that the increased sediment fluxes quickly decreased to nearly zero as the
roughness elements were further exposed. These studies suggest that flow over a
rough bed can both enhance and retard sediment transport.
1.3 Objectives
The primary objectives of the flume experiments described in this paper,
were 1) to establish the effects fine sediment bed cover has on near-bed flow
properties and sediment transport and 2) examine the efficacy of bed load dominated
flushing flows. Measurements of bed cover, near-bed flow properties and sediment
transport were taken during the filling and flushing of an experimental bed. To isolate
the effects of sand cover on the near-bed flow, a clean gravel bed was incrementally
filled while taking detailed velocity measurements. The flushing experiments
conducted for this study differed from previous experiments in two important ways.
First, fine sediment was transported exclusively as bed load. Second, an ample
supply of fine sediment was supplied to better represent natural rivers where many
kilometers of channel may be affected by fine sediment.
4
Chapter 2
Methods and Experimental Design
2.1 Introduction
The experimental work described in this paper was designed to collect
continuous coupled measurements of bed sand cover, near-bed flow properties and
sand transport in flows with varying discharge and sediment supply. Photographs of
the bed surface, taken throughout the experiments, were analyzed to estimate bed
surface sand cover. A laser Doppler velocimeter (LDV) was used to collect velocity
and turbulence data over changing bed conditions. A sediment trap at the tail of the
flume allowed for measurement of sand transport.
Two different experiments were conducted to examine relationship
between the three variables, bed cover, near-bed flow, and sand transport. The first
experiment was designed to isolate the effects of bed sand cover on near-bed flow
properties. During this experiment an open framework gravel bed was filled with
sand in small increments. Bed photos and velocity measurements were taken at each
increment. The second experiment was designed to examine the efficacy of flushing
flows when the fines are primarily transported as bed load. This experiment consisted
of three runs, each with different flow levels. The first flow was capable of consistent
sand transport, the second was capable of limited gravel transport and the last was
able to fully mobilize the gravel bed. Bed photos, velocity measurements and
sediment transport data were collected continuously during these runs. Both
experiments were repeated for data validation.
2.2 Experimental Facilities and Instrumentation
The experiments were conducted at the USGS Sediment Transport
Laboratory in Golden, CO, using a clear walled tilting flume with a 6 m working
length and 0.3 m working width (Figure 2.1). The flume was equipped with two 60
Hz pumps which were capable of producing discharges over 63 l/s and were
5
adjustable in 0.063 l/s increments. Discharge was measured with a venturi meter
mounted on the outflow pipe of the pumps. The flume tailgate was mounted on a
motorized screw elevator, allowing for millimeter-scale adjustments of flow depth.
The flume slope was set to 0.001 to achieve uniform flow throughout the working
area.
Figure 2.1. Picture of the full flume length (left) with the LDV and the rolling tray
used to mount the video camera. Close up of the LDV (right) and the 3 axis traverse
system.
The flume was equipped with a laser Doppler velocimeter (LDV) to
measure vertical velocity profiles. The LDV was located 4.2 m from the head of the
flume. This distance was chosen to allow sufficient time for the flow to develop over
the experimental bed and diminish entry effects found at the head box. The LDV was
mounted on a 3-axis traverse system that can position the laser vertically or laterally
along preprogrammed increments of distance. In these experiments, only vertical
measurements were taken. The laser beams were aligned to sample in the center of
the flume. Profiles of the downstream and vertical components of velocity, u and w,
were measured by taking individual readings at a series of points extending from the
bed surface to a depth of 12 cm for the filling experiments and 7.2 cm for the flushing
experiments. During the filling experiments individual readings were taken for 60 s
at 25 points. During the flushing experiments the sampling interval was decreased to
30 s and the number of points was reduced to 10. This truncated velocity profile
6
reduced the measurement time and allowed the LDV to capture changes in flow due
to evolving bed cover conditions.
Photographs of the bed were taken, continuously with a video camera and
periodically with a digital still camera, to record and analyze changing bed
conditions. The video camera was placed above the flume on a rolling tray and
positioned above the location of the LDV. The camera was adjusted so the laser
beams were in the center of the frame and only the experimental bed was visible. A
USB cable transmitted the video stream to a computer where signal processing
software was programed to capture and save one frame every 30 s. Along with the
video images, high resolution still photographs were taken of the entire bed in 30 cm
increments when the flow was stopped. These sets of photographs are referred to as
full bed photos.
A false floor plate at the tail of the flume was removed, creating a 30 cm
by 30 cm sediment trap. Sediment transported into the trap fell through an opening
and was routed to a collection container. The collection container was mounted on a
digital load cell with a 45 kg capacity and 0.15% accuracy. The load cell was
connected to the computer and the signal processing software recorded mass
measurements every 30 s. Tests showed that the mass of the water in the flume varied
with the water depth, but water mass was always less than 40% of the load cell
capacity. Water flow and surface waves did not produce fluctuations in the load cell
readings. Before the experiments began, known weights of sand were added to the
container and load cell readings were taken. During these tests the sediment trap and
load cell performed well; however, during the flushing experiments, off-center
loading often caused the system to perform poorly. After the first flushing experiment
the load cell stopped working entirely and was abandoned.
The experimental bed consisted of a mixture of gravel and sand. There
were several constraints imposed on the experimental sediment. First, the gravel
needed to be sufficiently large to create a rough/turbulent near-bed flow, but small
enough to be transported at higher flows. The gravel used was a well rounded
7
aquarium gravel with a median diameter of 8 mm. Second, the sand had to have a
high visual contrast with the gravel to facilitate bed photo analysis. Blue “Grade T
Colorquartz Crystals” from 3M were chosen because the bright blue color contrasted
with the earth tones of the gravel. Colorquartz is a sub angular quartz sand coated
with a blue ceramic pigment and has a density of 2.65 g/cm3. Finally, the
experimental sand had to be sufficiently large to be consistently transported as bed
load. Grade T Colorquartz has a median diameter of 1.0 mm. This sand was much
smaller than the gravel but was not likely to go into suspension at the highest
flows.
2.3 Procedures
2.3.1 Filling Experiments
At the start of the experiments the clean gravel was screeded to a depth of
5 cm along the bottom of the flume, creating a rough bed. To fill the bed, 187 g of
sand was poured on top of 25 x 25 cm areas of the gravel bed surface throughout the
entire bed length. The mass of sand represented 5% increments of the volumetric
gravel pore space. A ruler was used to agitate the gravel surface, forcing the sand into
the bed subsurface. A total of 22.7 kg of sand was added to completely fill the
subsurface. A constant discharge of 12.6 l/s with a depth of 20 cm was applied after
each filling increment. Detailed velocity profiles were recorded to establish near-bed
flow properties. Photos of the bed at the LDV site were taken after each increment to
document the change in sand cover on the bed. No load cell measurements were
taken during these experiments because the discharge was not capable of transporting
sand.
To fill and cover the bed, additional sand was poured on the surface, and
the water was agitated to sweep the sand grains off of the tops of the gravel particles
into the interstices. An additional 22.7 kg of sand was added to fully cover the bed.
The second filling experiment followed the same procedure as the first. However,
8
since the bed surface still had partial sand cover from the first flushing experiment,
only 14 kg of sand were added to fully cover the bed.
2.3.2 Flushing Experiments
The flushing experiments began after the bed was fully covered with sand.
No sediment was fed into the flume; instead, 4.2 m of the bed upstream of the LDV
acted as the sediment supply (Hassan et.al, 2006). The first flushing experiment
consisted of three runs, termed 1-A, 1-B and 1-C. At the beginning of run 1-A,
discharge was slowly increased to 23.8 l/s which produced consistent sand transport
over the length of the flume. Discharge was held constant until a dune that had
formed at the head of the flume passed the LDV site, after which discharge was
incrementally increased to 33.3 l/s. Velocity measurements, bed photos and load cell
readings were taken continuously throughout the run. The contents of the sediment
trap were emptied, dried and weighed on three occasions during run 1-A. First, the
trap was emptied after 14 hr when the dune had traveled the length of the flume. The
trap was emptied a second time after 35 hr and a third time after 51 hr.
The second and third runs of the first flushing experiment, runs 1-B and 1C respectively, were designed to transport gravel as well as sand. Discharge was kept
at 33.3 l/s for both of these runs; however, the tailgate of the flume was lowered,
resulting in shallower flows and higher velocities. Average velocity increased from
0.45 m/s during run 1-A to 0.77 m/s during run 1-B and 0.93 m/s during run 1-C. The
water surface for the high velocity runs was too wavy for the video camera to
accurately photograph through the flow. To get accurate measurements of bed cover,
two sets of full bed photos were taken during each run. Run 1-B was concluded after
5 hr and run 1-C after 3 hr of flow.
The second flushing experiment followed the same procedure as the first
experiment with some exceptions. First, the slope of the flume was increased to
0.003 to examine the effects of a higher slope on the flushing process. Second, only
two runs, 2-A and 2-B, were conducted. Third, discharge for the 2-A run was held
9
constant at 33.3 l/s during the entire run. Last, the full bed was photographed four
times during run 2-A to augment the bed photos taken with the video camera.
2.4 Data Analysis
2.4.1 Evaluation of Flow Properties
Nearly one thousand velocity profiles were collected during the four
experiments. Velocity measurements obtained from the LDV were used to estimate
different properties of the flow including the mean near-bed shear stress, τz, the mean
flow velocity, U, and the mean drag coefficient, Cd. The near-bed shear stress was
estimated from time-averaged measurements of the turbulent velocity fluctuations,
τ z = − ρu 'w '
(1)
where τz is the shear stress at height z above the bed, ρ is the density of water, u’ and
w’ are instantaneous fluctuations in downstream and vertical velocity, respectively
and the overbar indicates a time average. Since the velocity is often highly variable
near the bed, estimates of t τz were obtained by fitting a regression line to the profiles
of − ρu 'w ' in the upper part of the flow and extrapolating that line to the bed surface
(Dey and Sarkar, 2007) (Figure 2.2).
120
Clean Gravel Bed
100
Fully Sand Covered Bed
Z mm
80
60
40
20
0
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
Reynolds Stress N/m^2
Figure 2.2. Reynolds stress profiles measured over a gravel bed (closed symbols)
versus a sand covered bed (open symbols).
10
0.4
While discharge was held steady in the filling experiments, the flushing
experiments had varying discharges and thus τz and velocity did vary appreciably
throughout the project. To effectively compare data from different experimental runs,
a dimensionless drag coefficient, Cd, was calculated from the shear stress and velocity
data. The drag coefficient is defined as,
Cd =
τz
ρU 2
(2)
where τz is determined from Equation (1) and U is the mean velocity, which is
calculated by integrating the velocity profile over the flow depth,
U =
1 h
1 n
u
dz
≈
∑u ∆ z
h ∫zo z
h i=1 i i
(3)
where uz equals the velocity at depth z and zo is the roughness length which was
assumed to be 1/10 of the gravel D84 (1.0 mm).
2.4.2 Sediment Transport Analysis
Transport data were analyzed in two ways. First, the sediment trap was
periodically emptied and the sand was dried and weighed. The mass of sand collected
was divided by the run time in hours. The resulting transport rates are referred to as
the time averaged transport rates. The second method used the load cell data to
calculate how transport rates changed with time. Although the load cell did not
function properly during much of the experiment, two long records of accumulated
mass were collected from run 1-A (Figure 2.3). The first time series ran for 4.8 hours
and the second time series lasted for 11.6 hours. The total mass of sand collected
during the 16.4 hours represented by these two time series was 3.6 kg, which is less
than the 6 kg collected during the longer time averaged period.
11
22
21
Mass (kg)
20
y = 0.1864x + 14.19
19
2
R = 0.9879
18
17
y = 0.3101x + 9.1351
2
R = 0.9926
16
15
14
16
18
20
22
24
26
28
30
32
34
36
Run Time (hr)
Figure 2.3. Time series of two different load cell records during the 1-A run. The
first time series is composed of 600 individual load cell readings and the second is
composed of 1300 readings.
The trend line fitted to the first time series represents a transport rate of
0.28 g/m/s and the trend line fitted to the second time series represents a transport rate
of 0.17 g/m/s. Since the flow remained constant during this time, the difference in the
two transport rates may be the result of changing bed cover conditions or changing
near-bed flow properties. The first transport rate was higher than the time averaged
rate of 0.2615 g/m/s and the second was lower. Overall, both the total mass collected
and the transport rates are similar to those of the time averaged period, indicating the
load cell was working properly.
2.4.3 Image Analysis
To estimate the bed sand cover during the filling and flushing experiments,
the video and still bed photographs were processed and analyzed using Matlab image
analysis routines written for these experiments. First, individual pixels from the bed
photos were examined to determine the RGB values that differentiated the blue sand
from the gravel. Using this established value, the sand pixels were changed to black
and gravel pixels were changed to white, yielding a black and white copy of the
original picture (Figure 2.4). The number of black pixels in the resulting image were
12
counted and divided by the total number of pixels to calculate the percentage of the
bed covered by sand.
Figure 2.4. An original high resolution color photograph of the bed (left) and the
transformed image used for calculating bed cover (right). The red oval highlights a
single blue grain on top of a pebble in the color photograph and the resulting black
pixel in the transformed image.
Individual sand and gravel pixels were examined using the pixel region
tool in Matlab which displays the red (R), green (G) and blue (B) values of each
pixel. After careful examination of the RGB values, a rule was written to differentiate
the sand from the gravel, and this rule was used to transform the photograph to a
black and white image. The first rule stated that, if a pixel has a red value less than or
equal to the blue value plus ten, the pixel is identified as sand and given a value of 0,
which is black. This rule can be written R ≤ B+10. Any pixels not meeting the rule
were identified as gravel and given a value of 1 which is white. This rule was capable
of clearly defining the sand and gravel particles in the high resolution still camera
photographs (Figure 2.4).
The RGB values for the lower resolution video photographs differed from
those of the still photos. The first rule, referred to above, counted many pixels that
were clearly sand and overestimated the size of the gravel particles, resulting in
underestimates of the amount of sand present on the bed (Figure 2.5). Consequently,
the rule was adjusted to more accurately represent the bed cover. A second rule was
written as R ≤ B+30, and a third as R ≤ B+50. The third rule appeared to best capture
13
the bed cover and was used to analyze the bed images for these experiments (Figure
2.6).
a
d
b
c
Figure 2.5. Original video photo (a) and three different transformed images using
three different RGB value rules. Using the first rule (b) analysis shows the bed
covered by 50% sand. The second rule (c shows 63% sand cover and the third rule
(d) shows 72% sand cover.
14
a
46%
282 g
b
64%
376 g
c
92%
658 g
Figure 2.6. Three original bed photos (left) and the corresponding transformed
image (right). The black and white images were produced using the third rule which
was used for these experiments. The calculated bed cover percentages are indicated
on the bottom left of each black and white image and the mass of sand on the bed
surface is indicated on the bottom right.
15
A relationship between the mass of sand applied to the bed and the
estimated surface cover was developed to test the consistency of the photo analysis.
During the filling experiments a known weight of sand was added to each 25 x 25 cm
square of the bed. The bed photographs were approximately 25 x 19 cm, so the
photos represent approximately 75% of the applied masses. Using the applied mass
data and the calculated sand cover, a rating curve was developed to relate the two
variables (Figure 2.7). The rating curve for the second fill does not match the curve
for the first fill because the bed was initially 39% covered. The regression equation
fit to the data from first fill was solved for the initial bed cover. This calculation
suggested that 205 g of sand were initially present on the bed at the beginning of the
second fill. This initial mass was added to the applied mass data for the second fill.
Once adjusted, both filling tests displayed a similar relationship between the mass of
sand and the bed cover (Figure 2.7).
100
% Sand
80
60
y = -0.0002x2 + 0.3309x 40
1st Fill
20
2nd Fill
Adjusted 2nd Fill
0
0
200
400
600
800
1000
Sand Applied (g)
Fig 2.7. Relationship between the mass of sand applied to the bed and the resulting
bed cover. Data for the first filling experiment are shown by the black diamonds and
the grey squares represent the second filling experiment. The asterisks show the
second fill data after it was corrected for the initial bed cover.
16
Chapter 3
Analysis and Results
3.1 Filling Experiments
Two sets of filling experiments were conducted, both using the same
discharge and bed slope. Baseline relationships between sand cover and flow
properties were established by calculating changes in near-bed shear stress, τz, mean
velocity, U, and the drag coefficient, Cd, as the open-framework bed was filled with
sand. Flow properties were calculated from vertical velocity profiles using equations
(1), (2) and (3) from Chapter 2.
Velocity profiles measured during the filling experiments show that nearbed velocity increased slightly as the bed became covered with sand (Figure 3.1).
Most of the change in near-bed velocity occurred within the bottom 20 mm of the
flow, which is roughly double the gravel D84. Velocity near the water surface
decreased and U remained constant at 0.32 m/s. The bottom panel of Figure 3.1
shows that for flows over a fully sand covered bed, τz decreased considerably from
that of a clean gravel bed. This decrease was apparent throughout much of the flow
depth. These effects of bed composition on velocity and near-bed shear stress are
consistent with previous work by Sambrook Smith and Nicholas (2005).
17
5.0
4.5
4.0
3.5
ln z
3.0
2.5
2.0
Clean Gravel
Bed
Sand Covered
Bed
1.5
1.0
0.5
0.0
0.15
0.2
0.25
0.3
0.35
Velocity(m/s)
120
Clean Gravel Bed
100
Fully Sand Covered Bed
Z mm
80
y = -235.9x + 110.35
R2 = 0.9521
60
40
y = -354.84x + 108.21
R2 = 0.8843
20
0
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
Reynolds Stress N/m^2
Figure 3.1. Vertical velocity profiles (top) measured over a gravel bed (open
symbols) versus a sand covered bed (closed symbols). The bottom panel shows
profiles of − ρu 'w ' (Reynolds’ Stress) measured over the same bed conditions.
18
0.4
To effectively compare stresses from the filling experiments to those of the
higher discharge flushing experiments, measurements of τz were normalized by the
mean flow velocity producing a dimensionless drag coefficient, Cd. During the first
fill test, Cd decreased from 0.0055 for the clean gravel bed to 0.0025 for a fully sandcovered bed (Fig 3.2). During the second fill test the Cd decreased from 0.0045 for a
39% sand-covered bed to 0.0029 for a fully sand-covered bed. Data from the filling
experiments show that Cd has a non-linear response to the sand cover. As the bed is
filled there is little change in drag until the bed is 55-65% sand covered, after which
the drag quickly decreases. Once the bed is 85-90% covered, Cd appears to level out.
Both tests indicate a minimum Cd of approximately 0.003 for a fully sand covered
bed. Given that U did not change appreciably during the filling experiments most of
the response in Cd can be attributed to changes in near-bed shear stress.
0.007
Drag Coefficient
0.006
0.005
0.004
0.003
First Fill
2nd Fill
0.002
0.001
0
0
20
40
60
80
100
Sand Cover %
Figure 3.2 Decrease in Cd with increasing sand cover for the first filling experiment
(closed symbols) and the second filling experiment (open symbols).
The bed roughness height, z0, was calculated using,
⎡ ln h z0 − 1 ⎤
Cd = ⎢
⎥
⎣ κ
⎦
−2
(4)
19
where h is the flow depth and κ is von Karman’s constant (0.41). The equivalent
roughness, ks, was calculated assuming ks= 30 z0, where z0 is calculated from equation
(4) (Nikuradse, 1933). Figure 3.3 shows the decrease in ks with increasing sand
cover. The linear relationship is similar for both of the filling experiments. The ks for
a clean gravel bed was approximately 9 mm, which is similar to the D84 of the gravel.
The ks for a fully sand covered bed is around 1.0 to 1.5 mm which is similar to the D84
of the sand. Mixed bed composition yields a ks that is between the grain size of the
sand and gravel.
10
9
Fill 1
ks (mm)
8
Fill 2
7
6
5
4
3
2
1
0
0
20
40
60
80
100
Sand Cover %
Figure 3.3. Decrease in equivalent roughness with increasing sand cover for the first
filling experiment (closed symbols) and the second filling experiment (open symbols).
3.2 First Flushing Experiment
Once the bed was completely sand covered, discharge was increased to
conduct the flushing experiments. The flushing experiments consisted of three runs:
the first capable of sand transport (1-A), the second capable of limited gravel
transport (1-B) and the third capable of vigorous gravel transport (1-C). Discharge,
mean velocity, τz, and Cd measurements for each run of both flushing experiments are
reported in Table 3.1. Velocity, bed cover and sediment transport data were collected
continuously to capture evolving bed and flow conditions. Analysis of bed photos
20
and sediment transport data reveal three distinct phases during the experiment; the
steady phase, the flushing phase and the supply limited phase.
Run
Discharge
(l/s)
Mean
Velocity
(m/s)
Average
Shear Stress
(N/m^2)
Cd
1-A Steady
23.8
0.45
0.6
0.003
1-A Flush
26.3
to
33.3
0.48
to
0.55
0.66
to
1.3
0.003
to
0.005
1-A Limited
33.3
0.54
1.5
0.005
1-B
33.3
0.77
4.34
0.007
1-C
33.3
0.92
5.3
0.007
2-A Steady
33.3
0.57
1.22
0.003
0.003
to
0.005
2-A Flush
33.3
0.57
1.23
to
1.5
2-A Limited
33.3
0.56
1.5
0.005
2-B
33.3
NA
4.75
NA
Table 3.1. Discharge, mean velocity, average near-bed shear stress and drag
coefficient for all runs of both flushing experiments. Due to the extremely wavy water
surface during run 2-B, accurate measurements of flow depth were not obtained, thus
calculations of U and Cd could not be made.
The first phase, the steady phase, showed no significant flushing of the
fully sand covered bed. Data from the bed photos in Figure 3.4, show the bed cover
staying consistent near 100% during the first 16 hours of run 1-A. Cd during the
steady phase was approximately 0.003 which is consistent with results from the two
filling experiments. Time averaged transport rates during the steady phase were 389
g/h indicating active transport (Table 3.2). Even with active transport there was no
21
change in bed cover because sand flushed from the bed below the camera was
100
0.007
90
0.006
80
0.005
70
0.004
60
0.003
50
0.002
Cd
Sand Cover (%)
replaced by sand supplied from upstream.
Sand
Cd
40
30
0.001
0
0
10
20
30
40
50
Run Time (h)
Figure 3.4. Changes in Cd and sand cover during the 51 hours of the 1-A run. This
figure represents over 2500 bed photos and calculations of Cd made from over 400
velocity profiles. Note the steady, flushing and supply limited phases displayed in
both data sets.
22
Run
Mass Sand
(g)
Elapsed Run
Time (h)
Transport
Rate (g/hr)
Flux (g/m/s)
1-A Steady
5524.3
14.2
389.0
0.36
1-A Flush
5977.8
21.2
282.4
0.26
1-A Limited
1298.7
15.9
81.7
0.08
1-B-1
1440.5
2.5
587.0
0.54
1-B-2
169.1
3.2
53.4
0.05
1-C-1
1803
0.9
2003.3
1.85
1-C-2
490.8
1.9
265.3
0.25
2-A Steady
5142.2
1.1
4717.6
4.37
2-A Flush
3493.6
3.2
1109.1
1.03
2-A Limited
282.3
9.5
28.9
0.03
2-B-1
2000.9
0.7
2741.0
2.54
2-B-2
240.1
0.7
348.0
0.32
Table 3.2. The mass of sand collected, run time, time averaged transport rates and
fluxes for each run of both flushing experiments.
At the beginning of 1-A a dune formed near the head box and traveled the
full length of the flume. The dune indicated elevated transport rates possibly induced
by both a clean gravel (rough) bed and a standing wave present at the head gate.
Upstream of the dune, sediment supply was effectively limited and the bed began to
flush. During run 1-A the flushing phase began at t = 16 hr when the dune passed the
video camera site and lasted until t = 40 hr (Fig. 3.4). Photo analysis shows the bed
flushing in a linear fashion from 100% cover to between 60 and 70% cover. Cd
increased from 0.003 to 0.005, which is consistent with the filling experiments.
The time averaged transport rate for the duration of the flushing phase,
was 282 g/hr, which is 100g/hr less than the rate for the steady phase. Load cell data
23
from the first five hours of the flushing phase show a transport rate of 297 g/hr which
is slightly higher than the average rate. Load cell data for the last 12 hours of the
flush show a decreased transport rate of 185 g/hr. Fifty percent (11.5 kg) of the sand
originally applied to the surface was transported from the bed during the 35 hours
encompassing the first two phases.
The supply limited phase lasted the final 11 hours of the run; this phase is
defined by no further flushing of the bed and diminished transport rates (Figure 3.4).
During run 1-A the bed cover reached approximately 65% during the flushing phase
and was not cleaned further (Figure 3.5). Cd during this phase averaged 0.005, which
is slightly higher than the Cd value of 0.004 taken from the filling experiments for a
similar sand cover. This is likely because the velocities and discharges, during the
flushing phase, were considerably higher than those used during the flushing
experiments.
Figure 3.5 High resolution still photograph with 65% bed cover, which is
representative of the maximum flush achieved during run 1-A.
24
While the discharge used was more than sufficient to transport sand grains,
flushing was virtually nonexistent during the final 11 hours because of limited
transport. Time averaged transport rates for the supply limited phase were 81 g/hr,
which is a 71% decrease from the flushing phase. Regardless of the time averaged
transport rate, virtually no sand was observed entering the sediment trap for the last
hours of the run. Only 1.3 kg of sand were collected during this phase, leaving 44%
of the original applied sand on the bed surface. This lack of transport and bed
cleaning suggest a lower limit of bed cover for this flow.
Run 1-A was followed by two runs (1-B and 1-C), capable of incipient
gravel motion and full gravel transport, respectively. Run 1-B lasted 5.6 hours and
run 1-C for 2.8 hours. Discharge was held constant at 33.3 l/s for these two runs.
The tailgate of the flume was lowered to decrease flow depth and increase velocity,
thus achieving the higher bed stresses necessary to mobilize the gravel particles.
Since the water surface was too wavy, at these flows, to accurately photograph the
bed, the flow was stopped periodically and the entire length of the bed was
photographed to determine bed cover.
Analysis of the full bed photos shows additional cleaning of the bed with
increased flow and extended time. The top panel of figure 3.6, shows distribution of
sand cover over the length of the experimental bed for run 1-B-1 taken after 2.5 hrs
and 1-B-2 taken after 5.6 hrs of flow. At the LDV site (x = 4.2 m), the bed flushed
from approximately 65% at the end of run 1-A to less than 30% after 2.5 hours. The
average bed cover decreased from 31% after run 1-B-2 to 24% after run 1-B-2
(Figure 3.7). The time averaged transport rate for the first 2.5 hours of run 1-B was
587 g/hr, which was higher than any rate found during run 1-A (Table 3.2). The time
averaged transport rate decreased by 90% to 53 g/hr for run 1-B-2 indicating
conditions similar to the supply limited phase from run 1-A. As with run 1-A, this
run was concluded after little to no sand transport was observed for a substantial
period of time. This run removed 1.6 kg of sand from the bed leaving 37% of the
25
original 22.7 kg on the surface. It is important to note that this higher flow flushed a
larger mass of sand in 2.5 hours than was flushed during the entire 11 hours of the 1A supply limited phase.
Sand Cover (%)
45
1B-1
1B-2
40
35
30
25
20
15
1
Flow
2
3
4
5
6
5
6
Flume Bed Position (m)
Sand Cover (%)
30
25
20
15
10
1C-1
5
1C-2
0
1
Flow
2
3
4
Flume Bed Position (m)
Figure 3.6 Bed sand cover constructed from the full bed photos. The head of the
flume is at 0 m and flow is from left to right. The video camera and LDV were
located at 4.2 m. The spike at 3 m in the 1-B-1 photos represents a small dune that
traveled along the bed.
26
Figure 3.7 High resolution still photograph of the bed at the LDV site after run 1B-2. The bed cover in this photograph was estimated at 29%, which is representative
of the maximum flush for this run.
For run 1-C the tailgate was further lowered to increase velocity and
produce active gravel transport. The bottom panel of figure 3.6 shows bed cover after
0.9 hr (1-C-1) and 1.85 hr (1-C-2) of flow. During run 1-C-1 the average bed cover
decreased from 24 to 20% covered. During the next run cover decreased slightly to
an average of 18% at the end of run 1-C-2, which represents the cleanest bed found
during the entire study (Figure 3.8). The lack of significant flushing is similar to bed
conditions during the supply limited phase. Similar to the other runs, transport
slowed as the bed cleaned. The first time averaged transport rate for this run was
2003 g/hr, which far exceeded any other rates found in the first flushing experiment.
The time averaged transport rate for run 1-C-2 was 265 g/hr which was a decrease of
87% from run 1-C-1. This drastic drop in sand transport is indicative of the supply
27
limited phase. An additional 2.4 kg of sand was removed during the 2.8 hr of this
run, leaving 26% of the sand originally added to the surface of the bed. Again, more
sand was transported during this short run than during the entire transport limited
phase of run 1-A.
Figure 3.8 High resolution still photograph of the bed at the LDV site after run 1C-2. The bed cover in this photograph was estimated at 18%, which is representative
of the maximum flush for the entire study.
Overall, flushing efficacy appeared to be related to the near bed shear
stress. The τz increased by an order of magnitude from a low of 0.6 N/m2 during the
1-A steady phase to 4.3 and 5.5 N/m2 during runs 1-B and 1-C, respectively.
Transport rates increased accordingly, with larger amounts of sand transported in less
time during the high stress runs. This pattern and the presence of the dunes that
traversed the bed during each run may also indicate elevated transport rates caused by
28
the rough bed. Regardless of τz , each run eventually reached a limited state where no
flushing and little to no sand transport was observed.
3.3 Second Flushing Experiment
The second filling experiment was followed by an additional set of
flushing experiments for replication purposes. The bed slope was increased from
0.001 to 0.003, to examine the flushing process on a bed with a higher slope. The
second flushing experiment consisted of two runs 2-A and 2-B which were capable of
sand transport and limited gravel transport, respectively. Run 2-A lasted for 13.7 hr
and run 2-B only lasted 1.4 hr. Discharge for both runs was held constant at 33.3 l/s
and the tailgate of the flume was lowered during run 2-B to initiate limited gravel
transport, similar to earlier high velocity runs. Data for discharge, velocity, shear
stress and the drag coefficient are reported in Table 3.1. During run 2-A bed cover
data was obtained by both the video camera and full bed photos. Only full bed photos
were used during run 2-B.
Bed photos and Cd calculations from run 2-A show the same three-phase
progression as run 1-A, however, the increased slope and consistently higher stresses
led to shorter durations for the steady and flushing phases (Figure 3.9). The steady
phase during run 2-A lasted less than two hours. Bed cover was near 100% and Cd
was slightly higher than 0.003, which is consistent with previous results. A sharp
decrease in bed cover at 0.9 hours and the corresponding increase in Cd was caused
by a dune that formed downstream of the head box and eventually passed under the
video camera. This flushed bed, caused by the first dune, was quickly re-covered by
sand transported from upstream. The time averaged transport rate during this steady
phase was 4717 g/hr, which is considerably higher than any transport rates found
during the first flushing experiment (Table 3.2).
29
0.006
90
0.005
80
0.004
70
0.003
60
0.002
Sand Cover
Cd
50
Cd
Sand Cover (%)
100
0.001
40
0
0
2
4
6
8
10
12
14
Run Time (hr)
Figure 3.9 Changes in Cd and sand cover during the 13 hours of the 2-A run. After t
= 7 hr, the program used to collect LDV data crashed repeatedly and all velocity data
was lost.
The flushing phase lasted approximately 2 hours, during which Cd
increased from approximately 0.0035 to 0.005 which is consistent with earlier results.
The time averaged transport rate dropped 76% from the steady phase rate to 1109 g/
hr. Bed photos indicate that the bed was flushed to between 65 and 75% covered and
remained in that range for the next 9 hours. Both the reduction in transport rates and
bed cover are similar to results from flushing phase of run 1-A. The time averaged
transport rate for the final 9 hours was 28.87 g/hr, which is less than 3% of the
transport rate during the flushing phase. The relatively stable bed cover and large
decrease in transport indicate that the run was in the supply limited phase.
Full bed photos taken during run 2-A corroborate the results from the
video camera. Images taken after one hour during the steady phase, 2-A-1, show the
bed under the video camera was fully sand covered (Figure 3.10). Sites upstream
show the bed progressively cleaning from upstream to downstream as a dune travels
30
the length of the flume. Full bed photos taken during the flushing phase, 2-A-3, show
continued cleaning after three hours of flow and an average bed cover of 74%. Two
sets of full bed photos were taken during the supply limited phase after 8 and 10 hr of
flow, runs 2-A-8 and 2-A-10, respectively. The bed was cleaned to an average of
57% sand covered after run 2-A-8 and an additional two hours of flow cleaned the
bed to 55% after run 2-A-10. A total of 8.9 kg of sand was removed during the 13.7
hours of run 2-A.
Sand Cover (%)
110
100
2A-1
90
2A-3
80
2A-8
70
2A-10
60
50
40
30
2
Flow
3
4
5
6
5
6
Flume Bed Position (m)
Sand Cover (%)
40
30
20
2B-1
10
2B-2
0
2
Flow
3
4
Flume Bed Position (m)
Figure 3.10 Bed sand cover constructed from the full bed photos. The head of the
flume is at 0 m and flow is from left to right. The video camera and LDV were
located at 4.2 m.
Run 2-B completed the second flushing experiment. The tailgate of the
flume was lowered to increase velocity and achieve limited gravel transport. As with
the first experiment the water surface was too wavy to photograph, thus all bed cover
31
data were collected from the full bed photos. The increased velocity and stress
flushed the bed to an average of 30% covered after the 45 minutes of run 2-B-1
(Figure 3.10). After an additional 40 minutes, run 2-B-2 flushed the bed to an
average cover of 27%. This relatively small change in bed cover indicates that the
run had entered the supply limited phase. The time averaged sand transport rates for
this run also display behavior indicative of the supply limited phase. The transport
rate during run 2-B-1 was 2741 g/hr, which then decreased 87% to 348 g/hr for the 40
minutes of run 2-B-2. A total of 2.2 kg of sand was removed from the bed during the
1.4 hours of run 2-B, which is an order of magnitude larger than the amount of sand
removed during the entire 9 hours of the 2-A supply limited phase.
The second flushing experiments showed that increased shear stress,
caused by higher discharges and the steeper slope, allowed for higher transport rates
and a shorter duration for the steady and flushing phases. The increased stresses,
however, did not result in significantly less bed cover than the first flushing
experiment. Full bed photos from the supply limited phase show a bed cover of 60%,
at the LDV site, which is similar to the 60 to 70% cover found during run 1-A. Run
2-B, capable of limited gravel transport, flushed the bed to an average bed cover of
27%, which represents less flushing than the 24% cover found after run 1-B. All runs
of the second flushing experiment show an apparent lower limit to the flushing
process, with little to no sand transport after some point, and little change in percent
sand cover thereafter.
3.5 Management Implications
To assist water managers with flushing flow release decisions, additional
calculations were prepared to provide simple metrics for decision support. Using the
results from the preceding sections, a relation was developed between percent sand
cover and bed load transport intensity, calculated with respect to the experimental
gravel. Bed load transport intensity is defined as the ratio of the near-bed shear stress
to the critical shear stress for incipient motion of the experimental gravel (τcg).
Critical shear stress was calculated using the relation for dimensionless shear stress
32
τ cg = 0.03(s − 1)ρ gD
(5)
where, 0.03 is the assumed value of the critical dimensionless shear stress, s is the
specific gravity of the gravel, ρ is the density of water and D is the median grain size.
This equation was chosen because only simple field measurements are necessary to
complete the analysis. The bedload transport intensity was plotted against the final
bed cover percentages taken from either the video images or the full bed photos
(Figure 3.11). This plot demonstrates that bed load-dominated flushing flows require
stress levels 10-30% higher than τcg to flush the bed to values of less than 30% sand
cover. In typical gravel-bed rivers, the threshold for gravel transport occurs in the
range from about 1/2 to 2/3 of the bankfull discharge (Pitlick and Van Steeter, 1998;
Mueller et al., 2005), hence flows in this range are required to achieve significant
flushing.
80%
Final Bed Cover (%)
70%
60%
50%
1-A
2-A
1-B
1-C
2-B
40%
30%
20%
10%
0%
0
0.5
1
1.5
Bed Load Transport Intensity
Figure 3.11 Bed sand cover at the end of each run versus the bed load transport
intensity.
Often during controlled releases there is a limited amount of water
available to achieve specific restoration goals and reservoir operators must try to
33
maximize restorative affects with the resources available. To aid water managers
with these decisions the flow rates for each run were multiplied by the duration,
yielding the total volume of water used. The volume of water was then divided by
the mass of sand transported during the run to provide a metric of flushing efficiency
for each run (Table 3). The data show that the steady and flushing phases move the
most sand with the least amount of water. The limited phases of each run can require
one to two orders of magnitude more water to transport the same amount of sand as
the steady or flushing phases. This is best demonstrated by Runs 2-B-1 and 2-B-2.
Both runs used the same volume of water but run 2-B-2, which was in a limited state,
would require eight times more water than run 2-B-1 to move an equivalent mass of
sand. The data also show that higher-discharge releases require significantly less
water to achieve similar flushing results due to the reduced time necessary for
flushing. For example, run 2-A had consistently higher discharges than run 1-A;
however, this run required less than 20% of the water to remove similar amounts of
sand. In all cases the higher-stress runs flushed the bed with less water than the
lower-stress runs.
34
Run
Mass Sand (g)
Liters of H20
Liters/Mass Sand
1-A Steady
5524.3
1,216,656
220
1-A Flush
5977.8
2,284,800
382
1-A Limited
1298.7
1,906,092
1468
1-B-1
1440.5
299,700
208
1-B-2
169.1
383,616
2269
1-C-1
1803
107,892
60
1-C-2
490.8
227,772
464
2-A Steady
5142.2
239,760
47
2-A Flush
3493.6
383,616
110
2-A Limited
282.3
899,100
3185
2-B-1
2000.9
83,916
42
2-B-2
240.1
83,916
350
Table 3. The mass of sand transported, the volume of water used and the efficiency
metric for each run.
35
Chapter 4
Summary and Conclusions
Analysis of data from both the filling and flushing experiments reveals
several trends important to understanding flushing flows. The filling experiments
show that bed sand cover has an inverse relationship on near-bed shear stress and the
equivalent roughness. For an open framework gravel bed the Cd, which is largely
driven by τz, was more than double the Cd of a fully sand covered bed. The
equivalent roughness decreased linearly from the size of the gravel particles to the
size of the sand as the bed was covered.
The flushing experiments show that bed load dominated flushing flows
display three distinct phases. During the first phase, called the steady phase, no
significant bed cleaning occurred because the upstream supply was sufficient to
replace any fine sediment flushed from the bed. Once the fine sediment supply was
exhausted, the flushing phase showed a steady linear flush of the bed surface and a
corresponding increase in Cd. The final phase, called the supply limited phase, was
characterized by limited sand transport and virtually no cleaning of the bed surface.
The amount of bed flushing is likely controlled by the near-bed shear
stress. Low τz values only capable of sand transport left 60 to 75% of the surface
covered by sand. Increased τz, capable of limited gravel transport, flushed the bed to
an average cover of 25-30%. Flows capable of active gravel transport were able to
flush the bed to an average cover of approximately 20%. Regardless of the near-bed
shear stress, even the highest flows eventually reached the supply limited state, with
no additional bed flushing and minimal sand transport.
Sediment transport data gathered in these experiments best demonstrate
the effects of changing near-bed flow properties on the flushing process. The
elevated transport rates found in flows over a rough bed, shown by previous studies,
were not expressly documented in this study. However, observations may provide
anecdotal evidence for elevated transport caused by increases in τz over a rough bed.
36
First, the dunes that formed downstream of the clean gravel/sand bed interface during
runs 1-A and 2-A indicate that the rough bed produced transport rates which exceeded
the steady phase transport rates. Second, the initial transport rates for runs 1-B and 1C were well above the steady phase rates, possibly indicating roughness enhanced
sediment transport. While these runs had higher velocities, the high transport rates
were achieved on rougher gravel beds with significantly less sand cover.
As the gravel bed was further exposed transport rates began to decline and
eventually approached zero, which is consistent with previous work. Time averaged
transport rates decreased consistently from the steady phase down to the supply
limited phase. The load cell data indicate that transport rates scale with bed cover, as
transport rates were relatively high near the beginning of the flushing phase and lower
near the end when bed cover was less. This trend may be caused by a combination of
reduced drag forces and increased friction angles acting on sand grains now hidden
from the flow. During the supply limited phase transport rates dropped significantly,
indicating that the sand grains were almost entirely hidden from the flow.
Simple calculations relating near-bed stress levels and the volume of water
to sediment transport and flushing levels provide applicable metrics for water
managers contemplating bed load dominated flushing flows for habitat maintenance
on managed river systems. Flows with near-bed shear stress levels equal to or greater
than the critical shear stress for the gravel clasts are necessary to achieve bed surface
sand concentrations less than 30%. High stress, large magnitude releases are also
found to be the most efficient use of environmental maintenance flows. The
increased speed of flushing seen with high flows leads to the greatest amount of
flushing and sediment transport with the least volume of water.
Overall, these experiments show that flushing flows relying on bed load
transport of fine sediment are not very effective. First, no flushing will occur if there
is an adequate supply of fine sediment. Since a natural river may have many
kilometers of upstream supply, the duration of the release needed to flush the entire
system may be prohibitively long. Second, the low sediment fluxes found during the
37
limited phases indicate that additional flushing may take an extended period of time,
even when the upstream supply is depleted. Third, the data show that the most
effective flushing flows require near-bed shear stresses greater than τcg. Finally, bed
load dominated flushing flows may not fully remove the fines from the bed surface
layer even if the clasts within that layer are mobilized. The flushing values achieved
in these experiments are likely insufficient to promote healthy salmonid habitat
(Kondolf, 2000; Jensen et al., 2009).
38
REFERENCES
ALLRED, T. M. & SCHMIDT, J. C. 1999. Channel narrowing by vertical accretion
along the Green River near Green River, Utah. Geological Society of America
Bulletin, 111, 1757-1772.
ANDREWS, E. D. 1986. DOWNSTREAM EFFECTS OF FLAMING GORGE
RESERVOIR ON THE GREEN RIVER, COLORADO AND UTAH. Geological
Society of America Bulletin, 97, 1012-1023.
BAXTER, C. V. & HAUER, F. R. 2000. Geomorphology, hyporheic exchange, and
selection of spawning habitat by bull trout (Salvelinus confluentus). Canadian
Journal of Fisheries and Aquatic Sciences, 57, 1470-1481.
BESCHTA, R. L. & JACKSON, W. L. 1979. INTRUSION OF FINE SEDIMENTS
INTO A STABLE GRAVEL BED. Journal of the Fisheries Research Board of
Canada, 36, 204-210.
DIPLAS, P. & PARKER, G. 1992. Deposition and removal of fines in gravel-bed
streams, in Billi, P., Hey, R.D., Thorne, C.R., and P. Tacconi, eds., Dynamics of
Gravel-bed Rivers, John Wiley and Sons, p. 314-329.
GRAMS, P. E. & WILCOCK, P. R. 2007. Equilibrium entrainment of fine sediment
over a coarse immobile bed. Water Resources Research, 43.
GURTIN, S. D., SLAUGHTER, J. E., SAMPSON, S. J. & BRADFORD, R. H. 2003.
Use of existing and reconnected backwater habitats by hatchery-reared adult
razorback suckers: A predictive model for the Imperial Division, Lower Colorado
River. Transactions of the American Fisheries Society, 132, 1125-1137.
39
HASSAN, M. A., EGOZI, R. & PARKER, G. 2006. Experiments on the effect of
hydrograph characteristics on vertical grain sorting in gravel bed rivers. Water
Resources Research, 42.
HAWKE, S. P. 1978. STRANDED REDDS OF QUINNAT SALMON IN MATHIAS
RIVER, SOUTH ISLAND, NEW-ZEALAND. New Zealand Journal of Marine and
Freshwater Research, 12, 167-171.
JENSEN, D. W., STEEL, E. A., FULLERTON, A. H. & PESS, G. R. 2009. Impact of
Fine Sediment on Egg-To-Fry Survival of Pacific Salmon: A Meta-Analysis of
Published Studies. Reviews in Fisheries Science, 17, 348-359.
KARLINGER, M.R., ESCHNER, T. R., HADLEY, R. F., & KIRCHER, J. E. 1983.
"Relation of Channel-Width Maintenance to Sediment Transport and River
Morphology: Platte River, South-Central Nebraska." Geological Survey
Professional Paper 1277-E: E1-E18
KONDOLF, G. M. 2000. Assessing salmonid spawning gravel quality. Transactions
of the American Fisheries Society, 129, 262-281.
KONDOLF, G. M., SALE, M. J. & WOLMAN, M. G. 1993. MODIFICATION OF
FLUVIAL GRAVEL SIZE BY SPAWNING SALMONIDS. Water Resources
Research, 29, 2265-2274.
KONDOLF, G. M. & WILCOCK, P. R. 1996. The flushing flow problem: Defining
and evaluating objectives. Water Resources Research, 32, 2589-2599.
40
MUELLER, E. R. & PITLICK, J. 2005. Morphologically based model of bed load
transport capacity in a headwater stream. Journal of Geophysical Research-Earth
Surface, 110.
NELSON, J. M., SHREVE, R. L., MCLEAN, S. R. & DRAKE, T. G. 1995. ROLE
OF NEAR-BED TURBULENCE STRUCTURE IN BED-LOAD TRANSPORT
AND BED FORM MECHANICS. Water Resources Research, 31, 2071-2086.
NICKLING, W. G. & NEUMAN, C. M. 1995. DEVELOPMENT OF DEFLATION
LAG SURFACES. Sedimentology, 42, 403-414.
O’BRIEN, J.S., 1987. A case study of minimum streamflow for fishery habitat in the
Yampa River, in Thorne, C.R., Bathurst, J.C. and R.D. Hey, eds., Sediment Transport
in Gravel-bed Rivers, John Wiley and Sons, p. 921-946
OSMUNDSON, D. B., RYEL, R. J., LAMARRA, V. L. & PITLICK, J. 2002. Flowsediment-biota relations: Implications for river regulation effects on native fish
abundance. Ecological Applications, 12, 1719-1739.
OSMUNDSON, D. B., & SCHEER B. 1998. Monitoring cobble-gravel
embeddedness in the stream bed of the upper Colorado River . Final Report to the
Colorado River Recovery Implementation Program. U. S. Fish and Wildlife Service,
Grand Junction , Colorado .
OWENS, P. N., BATALLA, R. J., COLLINS, A. J., GOMEZ, B., HICKS, D. M.,
HOROWITZ, A. J., KONDOLF, G. M., MARDEN, M., PAGE, M. J., PEACOCK, D.
H., PETTICREW, E. L., SALOMONS, W. & TRUSTRUM, N. A. Year. Fine-grained
sediment in river systems: Environmental significance and management issues. In,
2005. John Wiley & Sons Ltd, 693-717.
41
PITLICK, J. & VAN STEETER, M. M. 1998. Geomorphology and endangered fish
habitats of the upper Colorado River 2. Linking sediment transport to habitat
maintenance. Water Resources Research, 34, 303-316.
REISER, D. W., WESCHE, T. A. & ESTES, C. 1989. STATUS OF INSTREAM
FLOW LEGISLATION AND PRACTICES IN NORTH-AMERICA. Fisheries, 14,
22-29.
SCHMEECKLE, M. W. & NELSON, J. M. 2003. Direct numerical simulation of
bedload transport using a local, dynamic boundary condition. Sedimentology, 50,
279-301.
SCHMEECKLE, M. W., NELSON, J. M. & SHREVE, R. L. 2007. Forces on
stationary particles in near-bed turbulent flows. Journal of Geophysical ResearchEarth Surface, 112.
SEAR, D. A. 1993. FINE SEDIMENT INFILTRATION INTO GRAVEL
SPAWNING BEDS WITHIN A REGULATED RIVER EXPERIENCING FLOODS ECOLOGICAL IMPLICATIONS FOR SALMONIDS. Regulated Rivers-Research &
Management, 8, 373-390.
SMITH, G. H. S. & NICHOLAS, A. P. 2005. Effect on flow structure of sand
deposition on a gravel bed: Results from a two-dimensional flume experiment. Water
Resources Research, 41, 12.
TYUS, H. M. & KARP, C. A. 1990. SPAWNING AND MOVEMENTS OF
RAZORBACK SUCKER, XYRAUCHEN-TEXANUS, IN THE GREEN RIVER
BASIN OF COLORADO AND UTAH. Southwestern Naturalist, 35, 427-433.
42
VAN STEETER, M. M. & PITLICK, J. 1998. Geomorphology and endangered fish
habitats of the upper Colorado River 1. Historic changes in streamflow, sediment
load, and channel morphology. Water Resources Research, 34, 287-302.
VANNIEUWENHUYSE, E. E. & LAPERRIERE, J. D. 1986. EFFECTS OF
PLACER GOLD MINING ON PRIMARY PRODUCTION IN SUB-ARCTIC
STREAMS OF ALASKA. Water Resources Bulletin, 22, 91-99.
WHITING, P. J. 2002. Streamflow necessary for environmental maintenance. Annual
Review of Earth and Planetary Sciences, 30, 181-206.
WILCOCK, P. R., KONDOLF, G. M., MATTHEWS, W. V. G. & BARTA, A. F. 1996.
Specification of sediment maintenance flows for a large gravel-bed river. Water
Resources Research, 32, 2911-2921.
WOOD, P. J. & ARMITAGE, P. D. 1997. Biological effects of fine sediment in the
lotic environment. Environmental Management, 21, 203-217.
WU, F. C. & CHOU, Y. J. 2003. Simulation of gravel-sand bed response to flushing
flows using a two-fraction entrainment approach: Model development and flume
experiment. Water Resources Research, 39.
43
Документ
Категория
Без категории
Просмотров
0
Размер файла
7 777 Кб
Теги
sdewsdweddes
1/--страниц
Пожаловаться на содержимое документа