Processing of sediment pulses following the removal of three small, gravel-filled barriers

Desiree Tullos1, Matt Cox1, Cara Walter1, Kelly Kibler1

Case studies The decommissioning of dams, as an approach to restoring longitudinal connectivity and to managing aging infrastructure, presents valuable opportunities for organized study of channel responses to sediment pulses. Experiments with physical and numerical models suggest that rivers process coarse sediment pulses through dispersion. In contrast, these experiments also found that translation appears to be a more important process when the sediment pulse consists of finer material, particularly when the grain sizes are finer than is typically present in the river. While the reported physical and numerical experiments have provided valuable insight into expectations for channel dynamics, they are largely unconfirmed by field observations. To explore whether dispersion dominates the processing of gravel pulses in natural rivers, we investigated downstream channel changes associated with three barrier removals in Oregon, ranging from very small (Oak Creek culvert, height = 1.5m), small (Brownsville Dam, height = 2.5 m), to medium (Savage Rapids Dam, height = 12m) in size. Each project trapped coarse sediment initially after construction, after which bedload passed over, or through, the barriers. Material behind the barriers was finer than the dominant grains downstream at Oak Creek and Savage Rapids, but was coarser than dominant channel grains at Brownsville. We present results from two years of post removal bathymetric and substrate surveys at each site. Net deposition and scour were calculated for cross sections and the thalweg profile, both in the reservoir and downstream of the former barriers. At our sites, sediment appears to be processed by both dispersion and translation, though dispersion appears to be the more dominant process. That is, we see evidence of the material both decaying in place, suggesting dispersion, as well as some skewing of the sediment wave in the downstream direction, indicative of translation of the material. Further, the channels processed sediments rapidly, eroding substantial portions of reservoir material within the first two years following removal of the barrier. These results suggest that, in the case of small to medium reservoirs filled with non-cohesive material that is coarser than downstream sediments, substantial aggradation will likely be limited to local areas directly downstream of the barrier. As the sediment pulse decays, the wavelength of the sediment release increases while amplitude decreases, such that effects beyond the reach immediately downstream of the dam are likely to be negligible, particularly if the barrier passed bed load before removal.

Oak Creek Culvert

River: Oak Creek Height: 1.5m

Purpose: Road culvert Replaced: 2007

D50r/D50c = 0.18 S = 0.007

Relative wave amplitude = 0.001

Brownsville Dam

River: Calapooia Height: 2.5m

Purpose: mill diversion, esthetics Removal: 2007 D50r/D50c = 3.7

S = 0.0025 Relative wave amplitude = 0.002

Savage Rapids

River: Rogue Height: 12m

Purpose: irrigation diversion Removal: 2009

D50r/D50c = 0.19 S = 0.002

Relative wave amplitude = 0.009

1. OSU-Biological and Ecological Engineering Department, 116 Gilmore Hall, Corvallis, OR; tullosd@engr.orst.edu; 541.737.2038

Discussion and Conclusions For Oak Creek and Brownsville, the relative wave amplitude is low and near a common for gravel-bedded rivers (0.001) where dispersive movement of waves dominates (Lisle et al. 2001). However, the channel slopes are also low, in a range (<1%) where Froude numbers are generally low, leading to translational movement of the sediment wave. Based on the scour/deposition figures at these sites, we see evidence that both the height of the wave decreases two years post removal and generally is decaying in place. For Savage Rapids, where the amplitude is higher and the wave material is smaller than the dominant channel particles, we anticipate stronger evidence of wave movement by advection in the second survey following removal. In this preliminary analysis, we acknowledge the subjectivity in identifying wave apex and edges, but also believe that this information, supported by further analysis, can provide some insight into the movement of sediment after barrier removal in coarse bed, low gradient streams. Ongoing work. In addition to analyzing the second field survey at Savage Rapids, ongoing analysis includes calculation of Froude numbers at the annual peak flows and establishing error estimates for scour and deposition calculations.

Background

Acknowledgements Support for this research provided by:

Figure 2, a and b. Scour and fill are calculated based on cross sections. Prior to replacement, the culvert on Oak Creek was undersized, backwatering the channel upstream and trapping some sediment while passing some through the culvert. At high flows, the roadway acted as a weir as Oak Creek flowed over it. Following replacement with an open bottom culvert, material trapped upstream of the culvert rapidly eroded a bench established by the backwatering and the channel widened to its historical width. Deposition at this site occurred primarily by building existing bars and filling pools. This sediment wave appears to primarily decay in place. However, the trailing edge of the wave does appear to migrate downstream in year two, indicative of advective processes.

Figure 3, a and b. Scour and fill are calculated based on cross sections. The Brownsville Dam filled with gravel within ~1.5 years, and passed bedload since that time. Prior to removal, the channel immediately downstream of the dam was a plane bed dominated by clay hardpan, with patches of gravel present. Erosion of the stored sediment occurred rapidly , though was limited by the presence of a crib dam underneath the removed structure. Downstream deposition occurred primarily by forming new bars and transverse riffles on top of an existing plane bed. Visually, the wave appears to disperse, generally decaying in place, with slight movement of the peak downstream and the trailing limb spreading downstream, suggesting some degree of advection occurring.

Figure 4. Scour and fill are calculated based on a longitudinal profile. Sediment passed through radial gates in the dam prior to removal. A pilot channel was cut and erosion of the reservoir sediments is ongoing. Deposition at this site occurred primarily by filling pools and forming new bars, including at pump intakes (Figure 5). Gravel has twice been extracted from this area. Further investigation of the dominance of dispersion or advection will occur once summer 2010 data are available.

-400

-300

-200

-100

0

100

200

300

400

-1,355 -788 -571 -334 -177 16 475 585 1,234 1,434 1,808

chan

ge in

cro

ss-s

ect

ion

al a

rea

(ft^

2)

Distance from dam- distance US is negative, distance DS is positive

Scour and Deposition one year after Brownsville Dam removal

2007-2008 deposition

2007-2008 scour

-400

-300

-200

-100

0

100

200

300

400

-1,355 -788 -571 -334 -177 16 475 585 1,234 1,434 1,808

cha

ng

e in

cro

ss-s

ect

ion

al a

rea

(ft

^2

)

Distance from dam- distance US is negative, distance DS is positive

Scour and Deposition two years after Brownsville Dam removal

2007-2009 deposition

2007-2008 scour

References

Knighton AD. 1989. River adjustments to changes in sediment load: the effects of tin mining on the Ringarooma River, Tasmania, 1875–1984. Earth Surface Processes and Landforms 14: 333–359. Madej MA, Ozaki V. 1996. Channel response to sediment wave propagation and movement, Redwood Creek, California, USA. Earth Surface Processes and Landforms 21: 911–927. Lisle TE, Pizzuto JE, Ikeda H, Iseya F, Kodama Y. 1997. Evolution of a sediment wave in an experimental channel. Water Resources Research 33: 1971–1981. Lisle TE, Cui Y, Parker G, Pizzuto JE, Dodd AM. 2001. The dominance of dispersion in the evolution of bed material waves in gravel-bed rivers. Earth Surface Processes and Landforms 26: 1409– 1420. Cao Z, Carling PA. 2003. On evolution of bed material waves in alluvial rivers. Earth Surface Processes and Landforms 28: 437–441. Cui Y, Parker G, Lisle T, Pizzuto J, and Dodd A. 2005. More on the evolution of bed material waves in alluvial rivers Earth Surface Processes and Landforms 30, 107–114 USBR. 2010. Memorandum: Travel Report to investigate reservoir sediment erosion and deposition at the former site of Savage Rapids Dam.

Figure 1: Idealized bed material waves (from Lisle et al. 2001)

Figure 5: Implications of bed wave movement. Deposition of gravel on pump intakes (Image: USBR 2010).

A number of experimental, numerical, and field studies and communications (Knighton, 1989; Madej and Ozaki, 1996, Lisle et al. 1997, Lisle et al. 2001, Cao and Carling 2003, Cui et al. 2005) have recently tried to characterize the conditions leading a pulse or wave of sediment to move by dispersion and/or advection (“translation”). This question is of relevance because, as Lisle et al. 2001 point out, dispersion of a sediment pulse will create an attenuating but temporally prolonged disturbance (e.g. aggradation) at a single location, whereas translation establishes a more prominent disturbance over a broader spatial scale as it moves downstream, but with a shorter timeframe for the impact and recovery. The implications for this debate are embedded in the design of dam decommissioning and removal (e.g. Figure 5), sediment retention structures (e.g. Toutle River), and landslide management, among others.

Roughly defined, a dispersive wave is characterized by the spreading of leading and trailing edges in place, both upstream and downstream, while the center of mass and apex either remain at the point of origin or move downstream as a function of the wave characteristics (Figure 1a). In contrast, all features (leading and trailing edges, apex, center of mass) of an advective, or translational, wave move downstream together and without diminishing substantially (Figure 1b). In reality, the movement of sediment waves is likely driven to some degree by both processes. Generally, it has been reported (Lisle et al. 2001) that two factors exert the greatest influence on the relative importance of dispersion and translation:

(1) interactions between the sediment wave shape and characteristics, hydraulics and bed load transport (2) differences in grain sizes between the wave material and bed-load material.

We present results from three small to medium barrier removals to explore whether longitudinal patterns of scour and deposition suggest dominance of dispersion or advection in transporting sediment. To characterize the hydraulic and sediment conditions at the site that may influence the dominant transport process, we include three metrics:

•D50r/D50c = ratio of reservoir D50 to averaged surface D50 in downstream reach •S: channel gradient •Relative wave amplitude = ratio of mean height above the original streambed to wave length

Key considerations regarding our sites:

•Very low percentages of fines present in reservoirs •Sediment in the reservoir was finer than what was present below the dam (D50 ratio <1) for two of our sites. •Systems were not supply limited prior to removal because the barriers were passing sediment

Abstract