evolving expectations of dam removal outcomes: …

EVOLVING EXPECTATIONS OF DAM REMOVAL OUTCOMES: DOWNSTREAMGEOMORPHIC EFFECTS FOLLOWING REMOVAL OF A SMALL, GRAVEL-FILLED DAM1

Kelly Kibler, Desiree Tullos, and Mathias Kondolf 2

ABSTRACT: Dam removal is a promising river restoration technique, particularly for the vast number of riversimpounded by small dams that no longer fulfill their intended function. As the decommissioning of small damsbecomes increasingly commonplace in the future, it is essential that decisions regarding how and when toremove these structures are informed by appropriate conceptual ideas outlining potential outcomes. To refinepredictions, it is necessary to utilize information from ongoing dam removal monitoring to evolve predictivetools, including conceptual models. Following removal of the Brownsville Dam from the Calapooia River, Oregon,aquatic habitats directly below the dam became more heterogeneous over the short term, whereas changes fur-ther downstream were virtually undetectable. One year after dam removal, substrates of bars and riffles within400 m downstream of the dam coarsened and a dominance of gravel and cobble sediments replaced previouslyhardpan substrate. New bars formed and existing bars grew such that bar area and volume increased substan-tially, and a pool-riffle structure formed where plane-bed glide formations had previously dominated. As theBrownsville Dam stored coarse rather than fine sediments, outcomes following removal differ from resultsof many prior dam removal studies. Therefore, we propose a refined conceptual model describing downstreamgeomorphic processes following small dam removal when upstream fill is dominated by coarse sediments.

(KEY TERMS: river restoration; dam removal; gravel-bed rivers; geomorphology; sediment transport; conceptualmodels.)

Kibler, Kelly, Desiree Tullos, and Mathias Kondolf, 2011. Evolving Expectations of Dam Removal Outcomes:Downstream Geomorphic Effects Following Removal of a Small, Gravel-Filled Dam. Journal of the AmericanWater Resources Association (JAWRA) 47(2):408-423. DOI: 10.1111/j.1752-1688.2011.00523.x

INTRODUCTION

In the United States (U.S.), dam removal isincreasingly implemented as a river restoration tech-nique (Hart et al., 2002; Gleick et al., 2009), reflectinga growing concern over the adverse ecological andsocial impacts of dams (Pejchar and Warner, 2001).Safety concerns and impediments to fish passage

associated with aging structures create the need fornew policies and funding sources to support removalprojects (Doyle et al., 2003b). However, to date, scien-tific investigation of dam removal outcomes is limited.Although nearly 700 dams have been removed in thelast 100 years (American Rivers et al., 1999; Gleicket al., 2009), outcomes from <5% of these removalshave been documented as published ecologicalresearch (Hart et al., 2002). Absence of monitoring at

1Paper No. JAWRA-09-0199-P of the Journal of the American Water Resources Association (JAWRA). Received January 4, 2010; acceptedDecember 22, 2010. ª 2011 American Water Resources Association. Discussions are open until six months from print publication.

2Respectively, Ph.D. Student, Water Resources Engineering, Oregon State University, Corvallis, Oregon 97330; Assistant Professor,Biological and Ecological Engineering, Oregon State University, Corvallis, Oregon; Professor, Department of Landscape Architecture andEnvironmental Planning, University of California, Berkeley, California 94720 (E-Mail ⁄ Kibler: [email protected]).

JAWRA 408 JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION

JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION

Vol. 47, No. 2 AMERICAN WATER RESOURCES ASSOCIATION April 2011

removal sites, inconsistent and incomplete experi-mental designs, and deficient baseline data have hin-dered the advancement of dam removal science(Heinz Center, 2002).

From a broader perspective, the scant quantity ofpublished research is further limited by differencesin the quality and type of information collectedamong sites. Moreover, physical settings and ecologi-cal constraints differ widely among rivers and dams,such that no single suite of dam removal effects canbe universally projected. Lacking an extensive bodyof knowledge and experience, managers, stewards,and researchers often struggle to predict the magni-tude, timing, and spatial extent of physical and eco-logical outcomes of dam removal (Hart et al., 2002;Heinz Center, 2002). As tradeoffs exist between therestoration benefit and the potential for disturbanceposed by dam removal (Stanley and Doyle, 2003),uncertainty about short- and long-term consequencescomplicates decision-making processes regardingwhether and how to remove dams (Aspen Institute,2002).

Existing research on the outcomes of dam removalhas primarily occurred on low-gradient rivers thattransport sand or silt (Stewart and Grant, 2005), doc-umenting deposition of fine sediments downstream ofdam failures and removals (Kanehl et al., 1997; Evanet al., 2000; Wohl and Cenderelli, 2000; Bushaw-Newton et al., 2002; Stanley et al., 2002; Doyle et al.,2003a; Ahearn and Dahlgren, 2005; Cheng andGranata, 2007; Burroughs et al., 2009). Past studiesmonitoring releases of stored reservoir sedimentshave reported a variety of morphological changesrelated to sediment routing and deposition throughdownstream reaches, including fining of channel sub-strates (Cheng and Granata, 2007), intrusion of finesediments into channel substrates (Stanley et al.,2002), filling of pools (Wohl and Cenderelli, 2000),decreased channel depth and increased width (Doyleet al., 2003a; Burroughs et al., 2009), and transitionto simplified bed forms (Kanehl et al., 1997; Bushaw-Newton et al., 2002; Stanley et al., 2002). Suchreported geomorphic alterations imply direct andindirect effects to downstream aquatic biological com-munities. For instance, the fine sediments evacuatedfrom Halligan Reservoir into the North Fork PoudreRiver filled pools and interstitial pore spaces withinthe channel’s cobble and boulder bed, impairingspawning and holding habitats for trout (Wohl andCenderelli, 2000).

Due in part to the preponderance of evidence fromdam removals in which the reservoir has releasedfine materials, dominant conceptual representationsof channel change downstream of dam removalsoften infer that sediments released from the reser-voir homogenize downstream habitat structure, bed

armoring and bar formations, and decrease substrategrain size, simplifying downstream habitats for manyyears after dam removal (Pizzuto, 2002). However,the number of studies documenting outcomes that falloutside of this expectation of habitat simplification isgrowing. Following the removal of two small Oregondams that stored coarse material, Dinner Creek andMaple Gulch, Stewart (2005) observed that coarsesediments from the reservoir were deposited close tothe dam, creating riffle-pool complexes in DinnerCreek where plane-bed morphology had existed priorto dam removal. In a physical model of a gravel pulsemoving through a sediment-starved riffle-pool reachwith alternating bars, Downs et al. (2009) observedthat the gravel pulse increased the complexity ofexisting habitat, maintaining the riffle-pool morphol-ogy as the pulse moved through the reach and leav-ing a legacy of bar deposits and related areas of scournear bars that persisted after the pulse had passed.Though gravel deposited into pools as well as on rif-fles and bars, most deposition occurred in areas oflow shear stress at the tail of the pool and pool depthwas not significantly altered.

A considerable number of the 2 million-plus damsestimated to exist in the U.S. are less than 2 m inheight (Graf, 1993; Shuman, 1995; Poff and Hart,2002). Many of these structures have passed theirdesign lifetimes, and in some cases, the reservoirshave completely filled with sediment. Due to thesheer number of such aging structures, decommis-sioning of small dams with full reservoirs is likely tobecome common in the near future (Doyle et al.,2002, 2003b). The ecological effects of removing smalldams are likely to differ from those associated withlarger dam removal, just as effects of coarse sedimentrelease may diverge from outcomes observed afterreleases of fine sediments. Thus, establishing a set oflikely consequences and predictive tools targeted spe-cifically to the various conditions in which damremoval may occur is an essential area of research tosupport decision making for future dam removals. Insupport of this research direction, we present down-stream channel changes observed following removalof the Brownsville Dam from the Calapooia River inWestern Oregon. A small structure (1.8-3.4 m inheight, depending on season), impounding a reservoirfilled with sediment, Brownsville Dam in many waystypifies the multitudes of defunct small dams thatmay be removed from American rivers in the future.However, unlike many reservoirs monitored in paststudies of reservoir releases, the Brownsville Damreservoir was filled with primarily coarse sediments.Adding to the breadth of knowledge regarding damremoval from rivers that transport gravels, resultsof the Brownsville Dam removal may be useful inpredicting downstream channel changes following

EVOLVING EXPECTATIONS OF DAM REMOVAL OUTCOMES: DOWNSTREAM GEOMORPHIC EFFECTS FOLLOWING REMOVAL OF A SMALL, GRAVEL-FILLED DAM

JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION 409 JAWRA

removal of other small dams that store coarse sedi-ments.

METHODS

Brownsville Dam Background

A tributary of the Willamette River, the CalapooiaRiver drains the Western Cascades of Oregon.Brownsville Dam was located in the mid portion ofthe 950 km2 Calapooia River watershed, upstream ofthe town of Brownsville (Figure 1). The upper Cala-pooia basin is a moderately steep catchment (channelgradient ranges from 0.44 to 1.94%), dominated byprivate forest land, which transitions to a wider val-ley of low to modest grade (0.10 to 0.44%), comprisedof primarily agricultural and low-density urban landuses. All peak flows and 90% of the Calapooia River’sannual runoff occur between November and May(Runyon et al., 2004).

Brownsville Dam was a hollow concrete dam, ini-tially constructed in the 1880s as a summer diversionstructure for a local woolen mill and rebuilt in the1960s to maintain aesthetics in the diversion canal.The 33.5 m wide dam raised the level of the Cala-pooia River by 1.8 m during the high-flow season(October-May). During summer months, removableflashboards were installed, impounding the riverlevel to 3.4 m above its historic elevation. When the

flashboards were removed for the high-water season,channel-forming flows passed over the dam with littleregulation.

After the reservoir filled with sediment, water andgravel-sized bed load passed over the dam, eventuallyundermining the structure on the downstream side,making the aging dam a liability for the owners. Inaddition to reducing the hazard risk presented by theBrownsville Dam, the 2007 removal was also justifiedin large part as removing a barrier to fish migration.Winter steelhead trout (Oncorhynchus mykiss) andspring Chinook salmon (Oncorhynchus tshawytscha),both of which are designated as Threatened underthe Endangered Species Act (ESA), seasonallymigrate through this section of the Calapooia tospawning and rearing grounds in the upper Cala-pooia basin. Although a fish ladder was constructedalong the right abutment of the Brownsville Dam, itwas not functional at all flows and was considered abarrier to migrating salmon. Twelve kilometersdownstream of the Brownsville Dam, the CalapooiaRiver is bifurcated for another mill diversion andapproximately 60% of the flow is diverted throughthe Sodom Ditch (Figure 1), which is regulated bySodom Dam. Also identified as a migration barrierfor fish, Sodom Dam is slated for removal during thesummer of 2011.

Data Collection

We conducted a preremoval baseline survey ofchannel morphology and habitat in the CalapooiaRiver study area during the summer base-flow season(July-August) leading up to the late-summer removalof the Brownsville Dam. In the reservoir reach (0-400 m upstream of the dam), the reach downstreamof the dam (0-1,600 m), and an upstream controlreach (650 m in length, located 1,600 m upstream ofthe reservoir), we surveyed channel cross-sections(103 total), longitudinal thalweg profiles, and barmargins and cross-sections, using a Nikon DTM-352Total Station (Nikon, Tokyo, Japan).

We collected and analyzed bulk sediment samplesfrom surface and subsurface bed materials (Churchet al., 1987), and conducted pebble counts (Wolman,1954) in riffles and bars of all study reaches. We sam-pled four bars and five riffles in the reach down-stream of the dam, two bars and two riffles in thereservoir reach, and two bars and two riffles in theupstream control reach. Surface bulk samples werecollected to the depth of the intermediate axis of thelargest surface particle while subsurface sampleswere collected to twice the depth of the surface layer.

We characterized aquatic habitat using the OregonDepartment of Fish and Wildlife (ODFW) AquaticFIGURE 1. Location of the Calapooia River and Brownsville Dam.

KIBLER, TULLOS, AND KONDOLF


Habitat Inventory methodology (Anlauf and Jones,2007), which includes visual estimation of substratecomposition (McHugh and Budy, 2005; Faustini andKaufmann, 2007), and characterization of channelunit types as riffles, glides, and pools (Nielson andJohnson, 1983). We established a streamflow-gaugingstation, which recorded stage data at 15 min inter-vals, and developed a stage-discharge rating curvebased on flow measurements over a range of dis-charges.

The baseline surveys of channel cross-sections, lon-gitudinal profile, depositional features, ODFW Aqua-tic Habitat Inventory, and bed material samplingwere repeated during the base-flow season one andtwo years after removal. All results reported hereinare based on comparison of one year of preremovaldata and two years of postremoval data. Using datacollected from the upstream control reach, we imple-mented a Before-After-Control-Intervention (BACI)experimental design to detect changes to the down-stream reaches with respect to channel substrate,area and volume of depositional features, and numberand types of channel units. The control reach wasselected based upon similarity to the downstreamreach in terms of broad geomorphic characteristics(e.g., valley width, slope), yet was far enoughupstream of the dam to be morphologically unaffectedby the reservoir and dam removal.

Data Analysis

We plotted particle grain-size distributions for barand riffle sediment samples, and extracted D50

(grain size at which 50% of material is finer), andthe percentage of material finer than 4 mm. Weused published values of measurement error associ-ated with the applicable sediment sampling method-ologies to estimate uncertainty of D50 and percentfines estimates derived from Wolman pebble counts(Wolman, 1954), and bulk sediment sampling(Ferguson and Paola, 1997; Shirazi et al., 2009).Using visual estimates of benthic substrate composi-tion from ODFW Aquatic Habitat Surveys, we calcu-lated reach-level averages of substrate composition.We applied empirically derived measurement errorspublished by ODWF (Anlauf and Jones, 2007) tovisual substrate estimates as characterizations ofuncertainty associated with the visual substrateclassification method.

We used Ordinary Kriging to generate bar surfacesbased on surveyed bar perimeters and cross-sections.We processed mapped bar areas to correct for fluctua-tions in discharge between survey years, using thehighest water level at the time of surveying (whichoccurred in 2008) as a constant datum. We then

calculated three-dimensional bar area and estimatedbar volume by computing the volume of materialbetween the generated bar surface and the tempo-rally constant datum.

We performed repeat surveys of wetted bar perim-eters to determine within-year survey error associ-ated with estimated bar areas. We computeduncertainty associated with bar volumes consideringtwo potential error sources: errors in approximatingbar surfaces using Ordinary Kriging of surveyed coor-dinates, and errors in bar volume stemming from theuncertainty of true bar areas. Volumetric uncertaintyestimates associated with the Ordinary Krigingmethod were generated for each bar, and we com-bined these estimates with volumetric uncertaintyarising from calculated ambiguities in bar area topropagate the total measurement error associatedwith estimated bar volumes.

As variability of many monitored parameters washigh and sample sizes were relatively small, use ofstatistical hypothesis testing to determine signifi-cance of changes observed in monitored parameterswas not appropriate for this study (Kibler et al.,2010). In order to establish significance of results, wepresent our results with explicit articulation ofparameter uncertainty, including assessment ofmeasurement error and interannual parameter vari-ability documented within the upstream controlreach. Within this analysis, change that cannot beattributed to error in parameter estimation and thatcannot be explained within the context of variabilityexpected in the absence of disturbance is attributedto disturbance caused by the dam removal andrecovery of the channel.

RESULTS

Baseline Assessment and Prediction of PostremovalEffects

Within the reservoir, a seismic refraction surveywas used to estimate the depth of accumulated allu-vium over bedrock (Northwest Geophysical Associ-ates, Inc., 2006). Integrating cross-sectional surveyswith sediment depth derived from the seismic refrac-tion survey, and approximating cross-sections astrapezoids sitting on an average channel slope(Randle and Daraio, 2003), we estimated the volumeof sediment stored behind Brownsville Dam to be11,000 m3. Bulk samples of sediment stored behindthe dam, collected to a depth of 1.8 m, indicated thatthe stored material was primarily very coarse gravel(D50 = 59 ± 1.5 mm).



Baseline surveys indicated that the reach immedi-ately below the Brownsville Dam (0-400 m) was anincised, plane-bed channel, dominated by clay hard-pan (31 ± 2% of channel substrate), with few deposi-tional features. Further downstream of the dam, thechannel transitioned to a wide and unconfined chan-nel, which we characterized using historical aerialphotography as highly dynamic, with frequent shiftsin thalweg location and changes in channel width, aswell as sizes and positions of depositional features(Walter and Tullos, 2009). We predicted that initialdeposition of mobilized reservoir sediments wouldoccur in the incised plane-bed reach immediatelydownstream of the dam. Because the reservoir hadfilled with gravel shortly after construction and hadpassed bed load since that time, we expected that themajority of the initially deposited sediment wouldeventually transport out of the reach. Within themore dynamic reach downstream, we expected thatany effects of the sediment pulse associated with damremoval would be obscured by high natural variabil-ity, making the detection of dam removal-inducedchanges difficult.

To further refine predictions of deposition, we esti-mated channel competence with distance downstream

of the dam, using the following equation after Lorangand Hauer (2003):

s0 ¼ ch sin#; ð1Þ

where so is the applied shear stress, c is the specificweight of water, h is the water depth at the 1.2RYIflow (129 m3 ⁄ s), and sin J is bed slope.

The channel profile displayed a distinct breakapproximately 400 m downstream of the BrownsvilleDam (Figure 2). Above this point, the channel gradi-ent was 0.08%, and calculated shear stress (so = 18-25 N ⁄ m2) was lower than the critical shear stress(sc = 32 N ⁄ m2; Chang, 1988) for the dominant reser-voir grain size (D50 = 59 ± 1.5 mm), making thisreach a likely locus of initial deposition. Beyond400 m downstream, the channel gradient increased to0.33% (so = 46-126 N ⁄ m2), and we expected this reachto transport the gravel supplied with minimal stor-age. Thus, we defined a likely depositional reach,DS-A, from 0-400 m downstream of the dam, withinwhich transient storage of reservoir sediments wasexpected. Beyond this point, we defined the remain-der of our downstream monitored reach as DS-B(400-1,600 m downstream of the dam), and predicted

0

50

100

150

200

250

300

102

104

106

108

110

112

+0

1+00

2+00

3+00

4+00

5+00

6+00

7+00

8+00

9+00

10+0

0

11+0

0

12+0

0

13+0

0

14+0

0

15+0

0

16+0

0

Shea

r str

ess (

Nm

-2)

Elev

atio

n (m

)

109

111

113

115

117

119

+0

1+00

2+00

3+00

4+00

5+00

6+00

7+00

Elev

atio

n (m

)

DS-A gradient = 0.0008 DS-B gradient = 0.0033

US gradient = 0.0023

τc = 32 Nm-2

FIGURE 2. Longitudinal Elevation Profiles (solid lines) and Gradients of the DS-A, DS-B, and US Reaches.Shear stress (dashed line) is displayed for a 1.2 return year interval discharge in DS-A and DS-B, relative to the

critical shear stress required to transport the median grain size of reservoir sediments (dotted line). Note that shearstress in the DS-A reach is lower than that necessary to transport the median grain size of reservoir sediments.



minimal detection of sediment-related effects in thisreach.

Baseline data from surface bulk samples takenfrom DS-A bed features indicated that surface mate-rials were fine to coarse gravel (bar D50 = 6 ±1.5 mm, riffle D50 = 23 ± 1.5 mm), with median grainsize smaller than that of the very coarse gravelcomprising reservoir sediments. Furthermore, per-centages of material finer than 4 mm were greater inDS-A than in reservoir sediment samples. We thuspredicted that the very coarse gravel stored in thereservoir would deposit into DS-A, and that the chan-nel bed grain size would initially increase. As thisreach of the Calapooia River had limited access to thefloodplain and contained no pools, we projected thatbar formation along channel margins would be theprimary response for storing the released sediment.Considering the relatively small volume of stored res-ervoir sediments, and that bed load had passed overthe dam for many years, we anticipated that, outsideof grade recovery at the dam site, changes in down-stream substrate condition and gravel storage wouldprimarily be transient effects.

Postremoval Observations

Floods Following Dam Removal. In the yearfollowing dam removal, flows during the high-waterseason did not exceed the 1.2 return year intervalflow. However, several hydrologic events exceededthis approximation of the channel-forming dischargein the second year after dam removal (Figure 3).

Median Substrate Grain Sizes. In the yearfollowing dam removal, we observed an increase inmedian grain size (D50) of bars and riffles closest tothe dam (Figure 4). The increase of D50 in bars dis-played a trend that attenuated in the downstreamdirection, with D50 of the bar closest to the dam (DS-A, +150 m) increasing by 600% the year following

dam removal, D50 of the two most upstream bars ofDS-B (+550 m, +780 m) increasing slightly (13 and15%, respectively), and D50 in the bar furthest fromthe dam (+960 m) decreasing slightly (22%). Like-wise, changes to D50 of riffles were most pronouncedin sites closest to the dam and were absent in fardownstream riffles.

Two years following dam removal, D50 in bars andriffles downstream of the dam decreased relative tothe previous year, with sites closest to the damreturning to conditions similar to preremoval condi-tions. D50 in bars further downstream becameslightly smaller than preremoval D50, whereas D50 insome riffles were slightly larger than preremovalyears. D50 in two riffles far downstream of the dam(+850 m and +1,270 m) did not increase the year fol-lowing removal, but dropped perceptibly two yearsafter removal. Fluctuations in bar and riffle D50 inthe upstream control reach indicated that the naturalvariability of D50 in bars and riffles is high in wateryears that experience channel-forming flows, suchthat, with the possible exception of changes observedin DS-A at site +150 m, changes in D50 observeddownstream of the dam fall into the range of interan-nual variability characteristic of this section of theCalapooia River.

Percentages of Fine Materials. Percentages ofmaterial finer than 4 mm decreased dramatically inbars and riffles of all reaches, including the upstreamcontrol reach, the year following dam removal(Figure 5). Two years after dam removal, fine materi-als in bars recovered to percentages similar to prere-moval conditions, with the exception of the barclosest to the dam, where percentages of fine mate-rials remained much lower than the preremovalcondition. Conversely, percentages of fine materialsin riffles remained low relative to the preremovalcondition, with the exception of the most downstreamriffle, where percentages of fine materials increasedperceptibly.

0

50

100

150

200

250

10/1/07 1/9/08 4/18/08 7/27/08 11/4/08 2/12/09 5/23/09 8/31/09

disc

harg

e (m

3 s-1

)

1.2 RYI = 129 m3s-1

FIGURE 3. Calapooia River Flows. Discharge monitored downstream of the dam, for twoyears following dam removal, displayed relative to the 1.2 return year interval flow.



Substrate Size Class Composition. Substratesin DS-A transitioned from hardpan-dominated(31 ± 2%) before the dam removal to largely com-prised of gravel and cobble (48 ± 15% and 28 ± 8%

respectively) the year following removal, with gravelscontinuing to dominate substrates (70 ± 22%) twoyears after removal (Figure 6). Comparatively minorchanges in substrate composition were observed in

19

31

22

27

9

33

6

40

2124

27

13

26

30

18

26

20 20

0

10

20

30

40

50

60

70

80

-2400 -2230 +150 +550 +780 +960

D50- Bars, surface (mm)

40 38

25

22

69

28

23

66

2928

45

34

67

68

27

24

38

26

4043

16

0

10

20

30

40

50

60

70

80

-2670 -2160 +150 +550 +850 +1010 +1270

D50- Riffles, surface (mm)

US DS-BDS-A US DS-BDS-A

FIGURE 4. D50 in Bars and Riffles of the Control Reach (US) and Downstream Reaches (DS-A and DS-B) Before and After Dam Removal.The horizontal axis is position in the watershed relative to the dam. Negative numbers are distances (in meters) upstream of the dam;positive numbers are distances downstream of the dam. White squares are 2007 samples, black circles are 2008 samples, and white circlesare 2009 samples. Error bars represent measurement error, calculated according to measurement error estimations given in Wolman, 1954(pebble counts); Ferguson and Paola, 1997; and Shirazi et al., 2009 (bulk samples).

29

13

29

22

3

19

52

1

27 27

16

35

26

9

28

34

3

25

0

10

20

30

40

50

60

-2400 -2230 +150 +550 +780 +960

Percent Fines- Bars, surface

18

1

4

28

0

7

22

0

4

22

0

6

10

1

30

0 1

8

0

15

0

10

20

30

40

50

60

-2670 -2160 +150 +550 +850 +1010 +1270

Percent Fines- Riffles, surface

US DS-BDS-A US DS-BDS-A

FIGURE 5. Percentage of Material Finer Than 4 mm in Bars (Surface and Subsurface) and Riffles (Surface) of the Control Reach (US) andDownstream Reaches (DS-A and DS-B) Before and After Dam Removal. The horizontal axis is position in the watershed relative to the dam.Negative numbers are distances (in meters) upstream of the dam, positive numbers are distances downstream of the dam. White squares are2007 samples, black circles are 2008 samples, and white circles are 2009 samples. Error bars represent measurement error, calculatedaccording to measurement error estimations given in Wolman, 1954 (pebble counts); Ferguson and Paola, 1997; and Shirazi et al., 2009 (bulksamples).



the DS-B reach, although sand and gravel percent-ages did increase slightly, associated with anobserved loss of cobble and hardpan. The upstreamcontrol reach displayed a measureable increase inpercentages of sand in the low-water year followingdam removal, a trend that was evident in down-stream reaches as well. However, percentages ofgravels and hardpan were relatively consistent in theupstream control reach throughout the study period.

Bar Area and Volume. Detailed bar surveystaken before and after dam removal reveal a 120 to700% increase in bar area in DS-A between the time

of dam removal and the survey one year afterremoval (Figure 7). Likewise, bar volume in DS-Aincreased 6- to 96-fold one year following damremoval. Two years following dam removal, bar areasand volumes in DS-A were similar to the year afterdam removal (changes did not exceed measurementerror), remaining high relative to the baseline condi-tion. In contrast, we measured a small (1 to 33%)increase in bar area and virtually no change to barvolumes in DS-B the year following dam removal,both within measurement error. Comparative analy-sis in the upstream control reach indicates negligiblechange in bar area and volume over the same timeperiod. The change in bar area and volume in DS-Aone year after dam removal exceeds parameter uncer-tainty, and is large relative to changes in theupstream control reach, thus we report an increase tobar area and volume in DS-A one year after damremoval that persisted at least two years after damremoval.

Geomorphic Channel Units. Prior to damremoval, DS-A displayed relatively simple channelmorphology when compared with DS-B and the UScontrol reach (Figure 8). Devoid of depositional fea-tures and riffle-pool structure, we characterized amajority of the channel (92% of channel length) asglide channel units. One year following dam removal,the number of channel units in DS-A had increasedfrom three to five, representing creation of a new rif-fle and two pools, where pools comprised 73% of thechannel length. Comparatively, in the same time per-iod, the number of channel units in the upstreamcontrol reach decreased from 11 to 10 and in DS-Bfrom 15 to 13. Two years after dam removal, thenumber of channel units in DS-A dropped fromfive back to three, though the new riffle and poolstructure created in the first year after dam removalwas partially retained. In DS-B, channel area charac-terized as glide channel units decreased in the twoyears following dam removal: two years after damremoval, glide channel units in DS-B had decreasedfrom 43 to 16% of channel length, while percentage ofthe channel length defined as riffles and poolsincreased from 20 to 28% and from 32 to 56%, respec-tively.

DISCUSSION

Interpreting the Significance of Monitoring Results

To support decision making and engineeringdesign with regard to dam removals, there is a need

9% 11%

1%4%

26%

7%

44%44%

58%

37%

19%

26%

1% 0% 0%5%2%

8%

0

0.2

0.4

0.6

0.8

1

SILT SAND GRAVEL COBBLE BOULDER HARDPAN

US

5% 3%0%

12%19%

0%

28%

48%

70%

11%

28%

18%13%

1% 2%

31%

3%10%

0

0.2

0.4

0.6

0.8

1


DS-A

9% 9%2%

6%

16% 15%

38%

49%52%

37%

17%

25%

1% 0% 1%10% 8%5%

0

0.2

0.4

0.6

0.8

1


DS-B

FIGURE 6. Substrate Size Class Composition by Percent in theControl Reach (US) and the Downstream Reaches (DS-A and DS-B),Before and After Dam Removal. White squares are 2007 samples,black circles are 2008 samples, and white circles are 2009samples. Error bars represent measurement error, calculatedaccording to measurement error estimations given in Anlauf andJones, 2007.



to predict dam removal outcomes and the temporaland spatial scales over which effects are likely tooccur. Acquiring information needed to inform suchpredictions requires monitoring of ongoing dam remo-vals using rigorous study designs, which occurs infre-quently due to factors such as unpredictabletimelines for removal, lack of funding for monitoring,and insufficient baseline data. Additionally, damremovals, by their very nature, defy some assumptionsof robust study designs and challenge traditionalmethods of evaluating the significance of results(Kibler et al., 2010). For instance, BACI experimentaldesigns implemented on longitudinal river systemsviolate the assumption that experimental sites areindependent of control sites (Hurlbert, 1984; Norrisand Hawkins, 2000). However, in addition to the diffi-culty of finding an appropriate control site outside ofthe river basin in question, choosing a reach from

another basin may introduce parameter variability, apotential impediment to detecting small to moderatechanges.

Due to the small sample sizes of our monitoredparameters, the likelihood of generating Type IIerrors in statistical hypothesis testing is high. There-fore, we do not consider statistical hypothesis testingto be appropriate for determining significance ofchanges observed downstream of the BrownsvilleDam removal. Rather, we choose to establish signifi-cance of results using a standard of practical signifi-cance based upon change exceeding parameteruncertainty, which we characterize as a combinationof measurement error and background variabilitydocumented within the control reach. We then usethese evaluations of preremoval versus postremovaldifferences to make inferences regarding theprocesses driving changes in the channel.

a. Vertical axis is area in m2.

b. Vertical axis is volume in m3.

2887±1262

2902±400

2959±840

0

1000

2000

3000

4000

5000

2007 2008 2009

US Bar Area

237±77

1206±613

1320±247

0

500

1000

1500

2000

2007 2008 2009

DS-A Bar Area

12,482±1188

14,552±1526

13,199±1386

10000

12000

14000

16000

2007 2008 2009

DS-B Bar Area

591±237

522±56

625±163

0

200

400

600

800

1000

2007 2008 2009

US Bar Volume

8±5

439±267

173±44

0

200

400

600

800

2007 2008 2009

DS-A Bar Volume

12,785±6121

12,914±1911

12,900±5340

4000

8000

12000

16000

20000

2007 2008 2009

DS-B Bar Volume

FIGURE 7. Area (a) and Volume (b) of Bar Features in the Control Reach and DownstreamReaches Before and After Dam Removal. Error bars represent measurement error.



One year after dam removal, we observed thatmedian grain sizes (D50) of bars and riffles in thereach directly below the dam (DS-A) had increased,and noted that the signal diminished with distancedownstream. However, as D50 was relatively variablein the control reach as well, it is difficult to attributehigh significance to the changes observed down-stream of the dam removal. We conclude that,although the point sediment samples provide sugges-tive evidence for a shift in grain size directly belowthe dam, the changes that we observed further down-stream are within the realm of interannual variabil-ity one may expect to see in this section of theCalapooia River. Although interannual variability ofgrain sizes is stronger than the signal of grain-sizeincrease that we observed following dam removal,there is no evidence that median grain sizes of barsand riffles downstream of the dam decreased follow-ing dam removal, as has been observed in dam remo-vals releasing fine sediment.

Likewise, our evidence does not suggest thatremoval of the Brownsville Dam caused percentages offine materials to increase in downstream reaches. Oneyear after dam removal, we observed a substantial

drop in percentages of fine sediment in bed materialsbelow the dam and in the upstream control reach.Because the upstream control reach behaved compa-rably to the reach below the dam, the changes in thedownstream reach cannot be attributed to local effectsof the dam removal. More likely, some change in finesediment supply from the catchment or sequencing ofsediment delivery and clear-water flows in the lowwater year following removal influenced both sites ina similar way. Nonetheless, it is evident that the damremoval did not contribute to a detectable increase inpercentages of fine sediment.

According to visual substrate classifications, DS-Awas dominated by clay hardpan prior to damremoval, whereas only 3% of the channel was com-posed of hardpan after removal. Given the corre-sponding increases in gravel and cobble substratesand based on field observations, coarse sediment thatevacuated from the reservoir appears to have depos-ited above the hardpan in DS-A following damremoval. Visual documentation of substrate size clas-ses is a method that encompasses much uncertainty,as empirical evidence from Oregon streams suggests,particularly with regard to discerning percentages of

FIGURE 8. Channel Units in the Control Reach (US) and Downstream Reaches (DS-A and DS-B)Before and After Dam Removal. Vertical axis is distance moving downstream, in meters.



gravel and cobble (Anlauf and Jones, 2007). We pres-ent strong observational evidence that one year afterdam removal percentages of hardpan and bouldersdecreased in DS-A, and that percentages of cobble andgravel increased. However, although percentages ofgravel substrate increased by 250% in the two yearsfollowing dam removal, the changes do not exceedpotential errors in characterizing this parameter, andtherefore lessen our certainty in concluding that thechanges observed are beyond measurement error. Wealso note an increase in percentages of sand in bothDS-A and DS-B, though given the simultaneousincrease in sand in the control reach, it is difficult toattribute this increase to the dam removal.

The shift in substrate texture from hardpan togravel and cobble immediately downstream of thedam was documented by visual estimation, yet notexplicitly detected by bulk sampling and pebblecounts, methods that preferentially sample nonaggre-gated materials. As reaches directly downstream ofdams may often scour to resistant layers comprised ofboulders, hardpan, or bedrock (Williams and Wolman,1984), dam removal monitoring should include meth-ods that allow estimation of bedrock or hardpan inanalyses of substrate change below dams. The faciesmap (Wolman and Schick, 1967; Kondolf et al., 2003)is such an approach, in that the patterns of surficialsediment deposits, exposed bedrock, and other fea-tures are mapped to scale, such that significantchanges in distribution of such features may bedetected in repeat mapping.

Strong evidence of an increase in area and volumeof depositional features downstream of the dam, com-bined with observations of changing bed-form mor-phology, lead to the conclusion that aquatic habitatsbecame more complex in DS-A following damremoval. As habitat unit classification tends to besomewhat subjective, it is possible that year-to-yearvariability between field crews may be partiallyresponsible for some of the changes observed. ODFWreports that identification of channel units using thismethodology is characterized by a moderate degree ofmeasurement uncertainty, with an error of approxi-mately 30% (Anlauf and Jones, 2007). Consideringthe subjectivity associated with this parameter, it isdifficult to interpret quantitative estimates of change(Poole et al., 1997). Acknowledging that these datashould be interpreted semiqualitatively, we suggest,based on the observed transition from a plane-bedchannel with few depositional features to more com-plex structure including riffles, pools, and multiplebars, that channel morphology in the reach directlybelow the dam became more heterogeneous followingremoval.

As preferred gravel grain sizes for some localorganisms, such as steelhead trout and Chinook

salmon, have been reported, we also may comment onthe ecological significance of some of the changesobserved. According to a meta-analysis of salmonidspawning gravel preferences undertaken by Kondolfand Wolman (1993), steelhead prefer to spawn ingravels with D50 ranging from 31 to 46 mm, whereasChinook prefer gravels with D50 between 16 and54 mm. The D50 in the riffle immediately downstreamof the dam increased to a size that is greater thanthat preferred by Chinook and steelhead, though onlyfor one year. Given that median grain sizes of allother bars and riffles remained within the preferredsize range and that we did not detect increases inpercentages of fine materials after the dam removal,we conclude that, at least with regard to grain sizesof spawning gravels, the Brownsville Dam removaldid not impair the quality of spawning gravel habi-tats. Further, the documented shift from hardpan togravel and cobble substrates, indicating creation ofnew potential spawning habitats, may also constitutean ecologically significant benefit to quantity of habi-tat in the reach downstream of the dam. Althoughwater temperature was not monitored throughout thedam removal study, we speculate that removal of theshallow impoundment, replacement of hardpan sub-strate with gravel and cobble, and creation of poolsindicate that the dam removal may have fashioned amore favorable thermal environment for cold-watersalmonids when compared with preremoval condi-tions, through facilitation of thermal exchangebetween surface water and substrates, particularlyaround the newly deposited bars (Burkholder et al.,2008) and creation of thermal refugia (Ebersole et al.,2003).

Revising Conceptual Models of Channel ChangeDownstream of Small, Gravel-Filled Dam Removals

Conventional perceptions of geomorphic changedownstream of dam removals often forecast smother-ing of the channel bed by a large volume of fine sedi-ments released from the reservoir, leading todramatic fining of substrate texture and homogeniza-tion of habitat structures and bed forms that will per-sist for years after the dam is removed. Theprevalence of these perceptions is based on documen-tation of such outcomes after the removal of manydams that contained a large volume of fine sedi-ments, whereas observation of alternative scenarioshas been less common. However, development ofnumerical models for predicting the erosion and depo-sition of sediment released with dam removal (Cuiet al., 2006a,b; Wong et al., 2004) and introductionof new dam removal case studies reporting alterna-tive scenarios (Stewart, 2005; Downs et al., 2009;



this study), are beginning to expand the breadth ofoutcomes observed and expected after dam removal.

Refining the conceptual models that facilitate pre-diction of dam removal outcomes is a process thatis likely to direct dam removal management towardenhanced decision making with regard to whetherand how to remove a dam. Often, particularly inreference to small dam removals, limited funding isavailable for detailed study or modeling aimed atpredicting potential outcomes, and decisions thusmay be made with relatively sparse information.For this reason, the availability of well-developedconceptual models informed by past observation canbe invaluable to the decision-making process. Tothat end, oversimplified ideas of one-size-fits-alldam removal effects have the potential to dominatestakeholder perceptions, and fears of detrimentaleffects may influence decision making, even whensuch concerns are unfounded. As an example, dur-ing discourse leading up to the decision to removeBrownsville Dam, stakeholders voiced anxiety thatdownstream aquatic habitats might be negativelyimpacted by dam removal as a consideration to notremove the dam (Elston, 2009), perhaps due in partto the prevailing conception that dam removalsresult in persistent simplification of downstreamhabitats. However, given that Brownsville Dam wasa low structure that stored a small volume ofcoarse sediment relative to flow competence, theseconventional perceptions surrounding dam removaldid not create realistic expectations for outcomes ofthis particular removal. A thorough preremovalbaseline assessment predicted an alternative out-come of limited and transient impacts to down-stream habitats, which was confirmed by threeyears of monitoring. This pattern of spatially lim-ited and transient responses, described in detailbelow, may apply to other small dam removals ingravel-bed rivers with limited fines, and can be ver-ified for individual sites through baseline assess-ments and basic hydraulic calculations. Suchanalyses may serve to inform stakeholder expecta-tions, or as a screening tool for distinguishing low-from high-risk projects. For the case of small damremoval in which the reservoir is filled with gravel,we suggest that the following pattern of responses(Figure 9), based on evidence from the sparsenumber of documented small dam removals fromgravel-bed rivers and from channel effects observedfollowing sediment pulses introduced to gravel-bedrivers, may be expected after removal.

At the Time of Removal. The reach below adam may be armored or simplified, particularly if thereservoir actively trapped sediment until theremoval, but also due to scour from an artificially

steep gradient at the dam site. In the case of fullreservoirs that had passed bed load prior to damremoval, sediment starvation may not be evident inthe reaches below the dam, though scour may still beobserved near the dam. Reservoir sediments maybegin to transport into the reach below the dam dur-ing base flows and reservoir dewatering, however, inthe case of dams that store coarse sediments, the firstflows competent to transport the dominant reservoirgrain size may not occur immediately. In this case,depending on the timing of dam removal relative tothe high-flow season, a temporal lag between damremoval and movement of reservoir sediments mayoccur. For instance, though the structure of MarmotDam was removed from the Sandy River in Oregonin July, during base-flow season, breaching of the cof-ferdam and substantial evacuation of reservoir sedi-ments did not occur until the first storm of the yearon October 19 (Grant et al., 2008).

Initial Sediment Transport. As reservoir sedi-ments move into the downstream channel, substratesshift toward the distribution of grain sizes found inthe reservoir. Other immediate channel responsestypically include a decrease in depth and reduction invariability of bed topography, effects that diminishwith time and the number of channel-forming flowssince the introduction of the sediment pulse (Madej,2001). Reservoir sediments initially deposit into pools(if any exist) or form and expand depositional fea-tures along channel margins. The volume and textureof stored sediments relative to the river’s sedimenttransport capacity will dictate the rate and patternsof substrate and channel form changes and provide acontrol to the magnitude of immediate change to beexpected following dam removal. For instance, assmall reservoirs store comparatively small volumes ofsediment (e.g., the Brownville Dam stored approxi-mately 1.5 years of sediment), the magnitude ofchanges observed even immediately following removalmay be subdued when compared with immediateobservations following removal of larger dams thatstored decades of sediment.

Locations and Timing of Sediment Deposi-tion. Approximate locations, magnitudes, and tim-ing of sediment deposition in downstream reachesmay be loosely predicted by theories of sedimentpulse movement and comparative grain sizes of reser-voir sediments and existing downstream substrates.Results from flume experiments undertaken by Lisleet al. (2001) and Downs et al. (2009) suggest that sed-iment pulses comprised of coarse material evolve pri-marily by dispersion, decaying in place, rather thantranslating downstream. Thus, in the case of smallreservoirs filled with noncohesive material that is



coarser than downstream sediments, detectabledownstream responses to dam removal will likely belocalized to the reach directly downstream of thedam, and will be detectable shortly after damremoval, with the magnitude of detectable effects

diminishing through time and with distance down-stream. As the sediment pulse decays, the wave-length (spatial extent of effect) increases whereasamplitude (magnitude of effect) decreases, such thatimpacts beyond the reach immediately downstream of

FIGURE 9. Stages of Channel Change Downstream of a Small, Gravel-Filled Dam Removal. Illustrating benchmarkstages in channel evolution following removal of the Brownsville Dam. The equilibrium channel was not observed

in this study, but is a prediction of conditions likely to be observed many years after removal.



the dam are likely to be negligible, particularly if thedam had passed bed load before removal. A pulse ofsediment may progress as described above, moving asa detectable wave with a well-defined amplitude andwavelength. However, it is more frequently the casethat the movement of well-defined wave is not evi-dent, and thus the evolution of a pulse of sedimentinstead may be evaluated through changes in bedforms and channel organization (Pitlick, 1993; Madej,2001).

Establishment of Channel Complexity. Com-plexity of physical habitats may develop soon after asmall release of coarse sediment (Madej, 2001; Stew-art, 2005; Downs et al., 2009; this study). A moderateto high amount of available sediment relative totransport capacity creates opportunity for differentialrates of deposition and scour through space and time,potentially increasing heterogeneity of hydraulic hab-itats on the subreach and reach scale (Yarnell et al.,2006). As subsequent flows rework initially depositedsediments and the sediment pulse passes through thereach, sediments deposited into pools are scoured andbars, floodplains, and channel margins become endur-ing locations of depositional storage (Downs et al.,2009; this study). For example, patterns of flow con-vergence and divergence within the Calapooia Riversculpted sediments deposited after removal of theBrownsville Dam into alternating bars and pools. Asbed forms and depositional features develop, differen-tial sorting also creates heterogeneity with respect topatches of different substrate grain sizes, as fines arewinnowed from riffles and deposited into pools, bars,and downstream of large wood or boulders.

Establishment of Channel Organization. Out-comes from the Brownsville Dam removal indicatethat complexity of habitats immediately below a smalldam removal may recover relatively quickly, in thiscase, within one year. However, long-term analysis ofsediment pulses indicates that although heteroge-neous random bed forms may develop soon after intro-duction of the sediment pulse, more time is requiredbefore the channel organizes into regularly spacedbed forms (Madej, 2001). Again, timing of sedimentevacuation from the reservoir reach and reorganiza-tion of released sediments into regularly spaced allu-vial channel structures is largely controlled by flowcompetence, as well as valley and channel-forcing fea-tures. In the case of reservoirs that store extremelycoarse fractions that are infrequently mobilized orwhen dam removal occurs before a series of low wateryears, a lag in release of reservoir sediments and therecovery of channel organization may occur.

In summary, the consequences of a small release ofgravel from a full reservoir following small dam

removal may diverge from past observations of rela-tively large releases of fine sediments that havecaused persistent downstream substrate fining andhabitat degradation. As such, conceptual modelsbased on evidence from releases of coarse sedimentsmay provide more accurate predictions of potentialdownstream effects of sediment release followingremoval of small dams from gravel-bed rivers.

CONCLUSIONS

Outcomes observed following removal of theBrownsville Dam provide valuable confirmation ofthe range of possible geomorphic responses to damremoval, and support the need for sufficient baselinedata in future work. In the case of this small damwith a gravel-filled reservoir, significant changes todownstream channel morphology were difficult todetect due to the small magnitudes of change relativeto measurement error and background variability.However, we present strong evidence that removal ofBrownsville Dam did not have a negative impact ondownstream aquatic habitats. Rather, within 400 mof the dam, we observed a coarsening of substrategrain sizes in bars and riffles, a shift in substratetype from hardpan to gravel and cobble, an increasein area and volume of bars, and creation of rifflesand pools that replaced a simplified plane-bed chan-nel. Amalgamation of detected outcomes leads us toconclude that the gravel released from the Browns-ville Dam reservoir increased habitat heterogeneityclose to the dam and had little detectable effect tochannel morphology further downstream. Becausethe relatively minute, localized, and ecologically ben-eficial changes that we observed are not welldescribed by some conceptual perceptions of damremoval outcomes, the Brownsville Dam removaloffers a case study for verifying an alternate concep-tual model for downstream channel response to theremoval of small, gravel-filled dams. Through contin-ual verification and revision of established conceptualmodels as new information becomes available, damremoval monitoring has the potential to informand influence decision making with regard to futurerestoration.

ACKNOWLEDGMENTS

This research was supported by grants from the Oregon WaterEnhancement Board (OWEB), Project Number 207-091, and fromthe National Oceanic and Atmospheric Association (NOAA),National Marine Fisheries Service, Award Number NA07NMF4630201. The authors would additionally like to thank the



Calapooia Watershed Council for implementing removal of theBrownsville Dam, Cara Walter and Trent Carmichael for data col-lection and processing, and Jennifer Natali for graphic design ofFigure 9.

LITERATURE CITED

Ahearn, D.S. and R.A. Dahlgren, 2005. Sediment and NutrientDynamics Following a Low-Head Dam Removal at MurphyCreek, California. Limnological Oceanography 50(6):1752-1762.

American Rivers, Friends of the Earth, and Trout Unlimited, 1999.Dam Removal Success Stories: Restoring Rivers Through Selec-tive Removal of Dams That Don’t Make Sense. AR ⁄ FE ⁄ TU,Washington, D.C.

Anlauf, K.J. and K.K. Jones, 2007. Stream Habitat Conditions inWestern Oregon, 2005. Monitoring Program Report NumberOPSW-ODFW-2007-05, Oregon Department of Fish and Wild-life, Salem, Oregon.

Aspen Institute, 2002. Dam Removal: A New Option for a NewCentury. Aspen Institute, Queenstown, Maryland.

Burkholder, B.K., G.E. Grant, R. Haggerty, T. Khangaonkar, andP.J. Wampler, 2008. Influence of Hyporheic Flow and Geomor-phology on Temperature of a Large, Gravel-Bedded River,Clackamas River, Oregon, USA. Hydrological Processes 22:941-953, doi: 10.1002/hyp.6984.

Burroughs, B.A., D.B. Hayes, K.D. Klomp, J.H. Hansen, and J.Mistak, 2009. Effects of Stronach Dam Removal on FluvialGeomorphology in the Pine River, Michigan, United States.Geomorphology 110:96-107, doi: 10.1016/j.geomorph.2009.03.019.

Bushaw-Newton, K.L., D.D. Hart, J.E. Pizzuto, J.R. Thomson,J. Egan, J.T. Ashley, T.E. Johnson, R.J. Horwitz, M. Keeley,J. Lawrence, D. Charles, C. Gatenby, D.A. Kreeger, T. Nighten-gale, R.L. Thomas, and D.J. Velinsky, 2002. An IntergrativeApproach Toward Understanding Ecological Responses to DamRemoval: The Manatawny Creek Study. Journal of the Ameri-can Water Resources Association 38(6):1581-1599.

Chang, H.H., 1988. Fluvial Processes in River Engineering, JohnWiley and Sons, New York and Other Cities, Citing Fortier, S.,and Scobey, F.C. (1926). ‘‘Permissible Canal Velocities,’’ Trans-actions of the ASCE 89:940-984.

Cheng, F. and T. Granata, 2007. Sediment Transport and ChannelAdjustments Associated With Dam Removal: Field Observa-tions. Water Resources Research 43:1-14.

Church, M.A., D.G. McLean, and J.F. Wolcott, 1987. River BedGravels: Sampling and Analysis. In: Sediment Transport inGravel-bed Rivers, C.R. Thorne, J.C. Bathurst, and R.D. Hey(Editors). John Wiley and Sons Ltd., Chichester, UnitedKingdom, New York, pp. 43-79.

Cui, Y., C. Braudrick, W.E. Dietrich, B. Cluer, and G. Parker,2006b. Dam Removal Express Assessment Models (DREAM).Part 2: Sample Runs ⁄ Sensitivity Tests. Journal of HydraulicResearch 44(3):308-323.

Cui, Y., G. Parker, C. Braudrick, W.E. Dietrich, and B. Cluer,2006a. Dam Removal Express Assessment Models (DREAM).Part 1: Model Development and Validation. Journal of Hydrau-lic Research 44(3):291-307.

Downs, P.W., Y. Cui, J.K. Wooster, S.R. Dusterhoff, D.B. Booth, D.E.Dietrich, and L.S. Sklar, 2009. Managing Reservoir SedimentReleases in Dam Removal Projects: An Approach Informed byPhysical and Numerical Modeling of Non-Cohesive Sediment.International Journal of River Basin Management 7(4):433-452.

Doyle, M.W., E.H. Stanley, and J.M. Harbor, 2002. GeomorphicAnalogies for Assessing Probable Response to Dam Removal.Journal of the American Water Resources Association 38(6):1567-1579.

Doyle, M.W., E.H. Stanley, and J.M. Harbor, 2003a. ChannelAdjustments Following Two Dam Removals in Wisconsin. WaterResources Research 39(1):1-15.

Doyle, M.W., E.H. Stanley, J.M. Harbor, and G.S. Grant, 2003b.Dam Removal in the United States: Emerging Needs for Scienceand Policy. Transactions of the American Geophysical Union84(4):32-33.

Ebersole, J.L., W.J. Liss, and C.A. Frissell, 2003. Thermal Hetero-geneity, Stream Channel Morphology, and Salmonid Abundancein Northeastern Oregon Streams. Canadian Journal of Fisheriesand Aquatic Science 60:1266-1280.

Elston, D., 2009. Characterizing Community Impacts of Small DamRemoval: A Case Study of the Brownsville Dam. M.S. Thesis,Oregon State University, Corvallis, Oregon, 116 pp.

Evan, J.F., S.D. Mackey, J.F. Gottgenn, and W.M. Gill, 2000.Lesson From a Dam Failure. Ohio Journal of Science 100(5):121-131.

Faustini, J.M. and P.R. Kaufmann, 2007. Adequacy of VisuallyClassified Particle Count Statistics From Regional StreamHabitat Surveys. Journal of the American Water ResourcesAssociation 43(5):1293-1315.

Ferguson, R.I. and C. Paola, 1997. Bias and Precision of Percentilesof Bulk Grain Size Distributions. Earth Surface Processes andLandforms 22:1061-1077.

Gleick, P.H., H. Cooley, M.J. Cohen, M. Marikawa, J. Morrison,and M. Palanappn, 2009. Dams Removed or Decommissioned inthe United States, 1912 to Present. In: The World’s Water 2008-2009, P.H. Gleick (Editor). Pacific Institute for Studies in Devel-opment, Environment, and Security, Island Press, Washington,D.C., pp. 239-264.

Graf, W.L., 1993. Landscapes, Commodities, and Ecosystems: TheRelationship Between Policy and Science for American Rivers.In: Sustaining Our Water Resources, Water Science and Tech-nology Board and National Research Council (Editors). NationalResearch Council, National Academy Press, Washington, D.C.,pp. 11-42.

Grant, G.E., J.D.G. Marr, C. Hill, S. Johnson, K. Campbell, O.Mohseni, J.R. Wallick, S.L. Lewis, J.E. O’Connor, J.J. Major,and B.K. Burkholder, 2008. Experimental and Field Observa-tions of Breach Dynamics Accompanying Erosion of MarmotCofferdam, Sandy River, Oregon. Proceedings of the WorldEnvironmental and Water Resources Congress, Ahupua’aHawaii, R.W. Babcock, and R. Walton (Editors). AmericanSociety of Civil Engineers, Reston, Virginia, pp. 1-10.

Hart, D.D., T.E. Johnson, K.L. Bushaw-Newton, R.J. Horwitz,A.T. Bednarek, D.F. Charles, D.A. Kreeger, and D.J. Velinsky,2002. Dam Removal: Challenges and Opportunities for EcologicalResearch and River Restoration. BioScience 52(8):669-681.

Heinz Center, 2002. Dam Removal: Science and Decision Making.Heinz Center for Science, Economics, and the Environment,Washington, D.C.

Hurlbert, S.H., 1984. Pseudoreplication and the Design of Ecologi-cal Field Experiments. Ecological Monographs 54(2):187-211.

Kanehl, P.D., J. Lyons, and J.E. Nelson, 1997. Change in theHabitat and Fish Community of the Milwaukee River, Wiconsin,Following Removal of the Woolen Mills Dam. North AmericanJournal of Fisheries Management 17:387-400.

Kibler, K.M., D.D. Tullos, and G.M. Kondolf, 2010. Learning FromDam Removal Monitoring: Challenges to Selecting Experimen-tal Design and Establishing Significance of Outcomes. RiverResearch and Applications, doi: 10.1002/rra.1415.

Kondolf, G.M., T.E. Lisle, and G.M. Wolman, 2003. Bed SedimentMeasurement. In: Tools in Fluvial Geomorphology, G.M.Kondolf and H. Piegay (Editors). John Wiley & Sons, New York,pp. 347-395.

Kondolf, G.M. and M.G. Wolman, 1993. The Sizes of SalmonidSpawning Gravels. Water Resources Research 29(7):2275-2285.



Lisle, T.E., Y. Cui, G. Parker, J.E. Pizzuto, and A.M. Dodd, 2001.The Dominance of Dispersion in the Evolution of Bed-MaterialWaves in Gravel-bed Rivers. Earth Surface Processes and Land-forms 26:1409-1420.

Lorang, M.S. and F.R. Hauer, 2003. Flow Competence and Stream-bed Stability: An Evaluation of Technique and Application.Journal of the North American Benthological Society 22(4):475-491.

Madej, M.A., 2001. Development of Channel Organization andRoughness Following Sediment Pulses in Single-Thread,Gravel-Bed Rivers. Water Resources Research 37(8):2259-2272.

McHugh, P. and P. Budy, 2005. A Comparison of Visual andMeasurement-Based Techniques for Quantifying CobbleEmbeddedness and Fine-Sediment Levels in Salmon-BearingStreams. North American Journal of Fisheries Management25:1208-1214.

Nielson, L.A. and D.L. Johnson, 1983. Fisheries Techniques. Ameri-can Fisheries Society, Bethesda, Maryland.

Norris, R.H. and C.P. Hawkins, 2000. Monitoring River Health.Hydrobiologia 435(3):5-17.

Northwest Geophysical Associates, Inc., 2006. Seismic RefractionStudy Brownsville Dam. Northwest Geophysical Associates,Inc., Corvallis, Oregon.

Pejchar, L. and K. Warner, 2001. A River Might Run Through ItAgain: Criteria for Consideration of Dam Removal and InterimLessons From California. Environmental Management 28(5):561-575.

Pitlick, J., 1993. Response and Recovery of a Sub-Alpine StreamFollowing a Catastrophic Flood. Geological Society of AmericaBulletin 105:657-670.

Pizzuto, J., 2002. Effects of Dam Removal on River Form andProcess. BioScience 52(8):683-691.

Poff, N.L. and D.D. Hart, 2002. How Dams Vary and Why ItMatters for the Emerging Science of Dam Removal. BioScience52(8):659-668.

Poole, G.C., C.A. Frissell, and S.C. Ralph, 1997. In-Stream HabitatUnit Classification: Inadequacies for Monitoring and Some Con-sequences for Management. Journal of the American WaterResources Association 33(4):879-896.

Randle, T.J. and J.A. Daraio, 2003. Sediment and GeomorphicAssessment for the Potential Removal of the Chiloquin Dam,Sprague River, Oregon. U.S. Bureau of Reclamation, Denver,Colorado.

Runyon, J., C. Andrus, and R. Schwindt, 2004. Calapooia RiverWatershed Assessment. Calapooia Watershed Council, Browns-ville, Oregon.

Shirazi, M.A., J.M. Faustini, and P.R. Kaufmann, 2009. Stream-bed Gravel Sampling and Frequency Base Conversion: A Solu-tion to Dataset Sharing. Water Resources Research 45(1):W01414.

Shuman, J.R., 1995. Environmental Considerations for AssessingDam Removal Alternatives for River Restoration. RegulatedRivers Research and Management 11:249-261.

Stanley, E.H. and M.W. Doyle, 2003. Trading Off: The EcologicalEffects of Dam Removal. Frontiers in Ecology and the Environ-ment 1(1):15-22.

Stanley, E.H., M.A. Leubke, M.W. Doyle, and D.W. Marshall,2002. Short-Term Changes in Channel Form and Macroin-vertebrate Communities Following Low-Head Dam Removal.Journal of the North American Benthological Society 21(1):172-187.

Stewart, G., 2005. Patterns and Processes of Sediment TransportFollowing Sediment-Filled Dam Removal in Gravel Bed Rivers.Ph.D. Thesis, Oregon State University, Corvallis, Oregon, 100pp.

Stewart, G. and G.E. Grant, 2005. Potential Geomorphic andEcological Impacts of Marmot Dam Removal, Sandy River,

OR. Report prepared for Portland General Electric, Corvallis,Oregon.

Walter, C. and D.D. Tullos, 2009. Downstream Channel ChangesAfter a Small Dam Removal: Using Aerial Photos and Measure-ment Error for Context; Calapooia River, Oregon. RiverResearch and Applications 26(10):1220-1245.

Williams, G.P. and M.G. Wolman, 1984. Downstream Effects ofDams on Alluvial Rivers. U.S. Geological Survey ProfessionalPaper 1286, Reston, Virginia.

Wohl, E.E. and D.A. Cenderelli, 2000. Sediment Deposition andTransport Patterns Following a Reservoir Sediment Release.Water Resources Research 36(1):319-333.

Wolman, M.G., 1954. A Method of Sampling Coarse River BedMaterial. Transactions of the American Geophysical Union35(6):951-956.

Wolman, M.G. and A.P. Schick, 1967. Effects of Construction onFluvial Sediment, Urban and Suburban Areas of Maryland.Water Resources Research 3(2):451-464.

Wong, M., A. Cantelli, C. Paola, and G. Parker, 2004. ErosionalNarrowing After Dam Removal: Theory and Numerical Model.Proceedings of the ASCE World Water and EnvironmentalResources 2004 Congress, Salt Lake City, June 27 to July 1, 10pp.

Yarnell, S.M., J.F. Mount, and E.W. Larsen, 2006. The Influence ofRelative Sediment Supply on Riverine Habitat Heterogeneity.Geomorphology 80:310-324.



evolving expectations of dam removal outcomes: …

Documents