WATER RESOURCES RESEARCH, VOL. 35, NO. 11, PAGES 3507-3521, NOVEMBER
Effects of hydraulic roughnesson surface textures
of gravel-bed rivers
John M. Buffington• and David R. Montgomery
1999
This file was created by scanning the printed publication.
Errors identified by the software have been corrected;
however, some errors may remain.
Departmentof GeologicalSciences,Universityof Washington,Seattle
Abstract. Field studiesof forestgravel-bedriversin northwesternWashingtonand
southeastern
Alaska demonstratethat bed-surfacegrain sizeis responsiveto hydraulic
roughnesscausedby bank irregularities,bars, and wood debris.We evaluatetextural
responseby comparingreach-average
mediangrain size(Ds0) to that predictedfrom the
totalbank-fullboundary
shearstress
(r0b), representing
a hypothetical
reference
condition
of lowhydraulic
roughness.
For a givenr%f,channels
withprogressively
greater
hydraulicroughnesshave systematically
finer bed surfaces,presumablydue to reducedbed
shearstress,resultingin lower channelcompetenceand diminishedbed load transport
capacity,both of whichpromotetexturalfining.In channelswith significanthydraulic
roughness,
observedvaluesof Ds0 canbe up to 90% smallerthan thosepredictedfrom
rob.We findthatwooddebrisplaysanimportant
roleat ourstudysites,notonly
providinghydraulicroughness
but alsoinfluencingpool spacing,frequencyof textural
patches,and the amplitudeand wavelengthof bank and bar topographyand their
consequentroughness.Our observations
alsohave biologicalimplications.We find that
texturalfining due to hydraulicroughnesscan createusablesalmonidspawninggravelsin
channels that otherwise would be too coarse.
sorting(o-)andthe sizeof the particleof interest(D•) relative
to its neighbors(Di/D so,whereD sois the medianbed-surface
grain size) [Kirchner et al., 1990; Buffington et al., 1992;
Johnstonet al., 1998].
The broadgrain-sizedistributions
of gravel-bedrivers(typicallysandto cobble)allowfor considerable
selectivetransport
and dynamictexturalresponseto local perturbationsof sediment supplyor transportcapacity[Dietrichet al., 1989]. For
qb= k(r' - rc)n
(1) example,a local transportcapacitygreaterthan supplymay
result in winnowingof fine grains and textural coarsening.
whereqb is the bed load transportcapacity(i.e., the transport Textural coarsening,in turn, createsa rougher surfacewith
rate of a channelunlimitedby sedimentsupply)and k and n greater intergranularfriction angles,increasingcritical shear
are empirical values [du Boys, 1879; O'Brien and Rindlaub,
stresses
for grainsmoving
overthebed(rc•) [Kirchner
et al.,
1934;Meyer-Peterand Mtiller, 1948; Chien, 1956; Wathenet al.,
1990;Buffingtonet al., 1992;Johnstonet al., 1998] thereby
1995].Here r' is that portionof the total boundaryshearstress
retardingbedloadtransportrates(equation(1)). Altered bedthat is appliedto the bed and responsible
for sedimenttranssurfaceroughnessalsoaffectslocalvelocitystructureand bed
port. It is definedas the total boundaryshearstress(to) corshear stress[Naot, 1984; Wibergand Smith, 1991]. Conserected for momentumlossescausedby hydraulicroughness
quently,there is a dynamicfeedbackbetweenbed-surfacetexother than grain skin friction [Einsteinand Banks, 1950;
ture, bed shearstress,and transportcapacity.Given sufficient
Einsteinand Barbarossa,1952;Nelsonand Smith,1989]
time, constancy
of sedimentandwater inputs,and availability
,].t= TO_ T"-- '1'
.......
Tn
(2) of mobile sediment,the above processfeedbackswill ultimately result in equilibrationof the transportrate with the
bed = total - banks - bed forms ....
other
imposedsedimentsupplyrate [Dietrichet al., 1989;Lisleet al.,
1993]. Selectivetransportthat resultsin bed-surfacecoarsenThe critical, grain-mobilizingshear stressfor a grain size of ing and armor developmentalso makesthe bed surfaceless
interest(rc,) is a function
of submerged
grainweight,particle mobile and altersthe timing and total contributionof subsurprotrusion into the flow, and intergranular friction angle
face sedimentto the supplyof bed load material [Milhous,
[Wibergand Smith, 1987;Kirchneret al., 1990;Buffingtonet al.,
1973;ParkerandKlingeman,1982;CarlingandHurley,1987].
1992;Johnstonet al., 1998];the latter two dependon sediment
In thispaperwe examinethe effectsof hydraulicroughness
on bed-surfacegrain size in complexforestchannels.Gravel•Nowat WaterResources
Division,
U.S.Geological
Survey,
Boul- bed rivers in forestedmountain drainagebasinscommonly
der, Colorado.
containnumeroussourcesand scalesof hydraulicroughness
Copyright1999 by the American GeophysicalUnion.
within a singlereach, such as bed-surfaceskin friction; form
drag due to barsand in-channelflow obstructions
(boulders,
Paper number 1999WR900138.
0043-1397/99/1999WR900138509.00
wood debris,and bedrockprojections);skinfriction and form
1.
Introduction
The surfacegrainsizeof gravel-bedriversreflectsthe caliber
andvolumeof sedimentthat is suppliedto the channel,aswell
as the time-integratedfrequencyand magnitudeof discharge
eventsthat are capableof moving sediment.Surfacegrains
movewhenthe bed shearstress(r') exceedsthe criticalstress
for grain motion (rc)
3507
3508
BUFFINGTON
AND
MONTGOMERY:
EFFECTS
drag causedby riparian vegetationlining the banksand protruding into the flow; and momentumlossesdue to downstream changesin channel width and planform curvature.
These hydraulicroughnesselementscan significantlyreduce
the bed shearstress(equation(2)). For example,form drag
due to bed formsin sand-bedand gravel-bedriverscan result
in bed stressesthat are 10-75% lessthan the total boundary
shearstress[Parkerand Peterson,1980;Prestegaard,
1983;Dietrichet al., 1984;Hey, 1988].
We hypothesizethat channelswith greaterhydraulicroughnesswill have smaller proportionsof total boundaryshear
stressavailablefor sedimenttransport(equation(2)) andthus
will havedecreasedcompetenceand finer bed surfaces.In this
paper we developa frameworkfor examininggrain-sizeresponseto hydraulicroughnessand test the abovehypothesis
usingfield data from forestgravel-bedchannels.
2.
Analysis Framework
Our method for examiningtextural responseto hydraulic
roughness
involvescomparingobservedbed-surface
grainsizes
to thosepredictedfor a hypothetical,low-roughness
reference
state [Dietrichet al., 1996].
2.1.
Reference
State
In the absenceof major hydraulicresistancethe bed shear
stressis approximately
equalto the total boundaryshearstress
(equation (2)). This hypothetical,end-memberconditionof
low hydraulicroughness
is visualizedasa wide,straight,planar
channelwith relativelysmall grain sizesthat do not offer significantform drag.The competent,median,bed-surfacegrain
size(D so) for thischannelcanbe predictedfrom the Shields
[1936]equation,whichis a forcebalancebetweenthe driving
fluid stressesand the resistinggrain weight per unit area at
incipientparticle motion
* = 7c5o/[Dso(p•p)#]
T c50
(3)
where p and Psare the fluid and sedimentdensities(set equal
to 1000and2650kg/m3, respectively),
7cSoand7*c5oare the
dimensionaland dimensionless
criticalshearstresses
for incipient motion of D so, respectively,and # is gravitationalaccel-
eration.We set 7'csoequalto 0.030,a conservative
valuefor
OF HYDRAULIC
ROUGHNESS
The referenceDso is the limit of channelcompetence(maximum mobileDso) for a river with vanishinglysmallhydraulic
roughness
otherthangrainskinfriction(i.e., 7' •- 70,see(2)).
It is importantto note, however,that the referenceDso is a
hypotheticalgrainsize,the actualoccurrenceof whichdepends
on the volumeandcaliberof sedimentsuppliedto the channel,
which,in turn, is a basin-specific
functionof geology,geomorphic processes,
and anthropogenicdisturbance.
2.2. Hypothesis
On the basisof the abovetheorywe hypothesizethat channels with greater hydraulicroughnesswill have lower bed
stresses
(equation(2)) and thereforewill have smallerbedsurfacegrainsizesthan that predictedfrom the total boundary
shearstress(i.e., (4) with 7' < 7oversus7' = 70).Consequently,
for a giventotal boundaryshearstresswe expecta systematic
texturalfiningwith increasinghydraulicroughnessand lower
bedstress.The magnitudeof texturalfiningshouldreflectboth
the decreasein competenceand the degreeof fine-sediment
deposition forced by hydraulic roughnessand lower bed
stresses.Hydraulic roughnessthat lowers 7' shoulddecrease
the bed load transportcapacity(equation (1)), resultingin
reduced bed-surfacegrain size due to depositionof finegrainedparticles(thosetypicallyin transportand comprising
the majorityof the bed load). To examinethe effectsof hydraulicroughnesson surfacegrain size,we predictcompetent
mediangrain sizesfor channelswith low hydraulicroughness
((4), with 7' = 70) and comparethese valuesto observed
mediangrain sizesin channelswith systematically
greaterhydraulicroughness.
2.3. Reach-ScaleApproximation
Becauseflow perturbationscausedby roughnesselements
are spatiallycomplexand inherentlynonlinear,we examine
grain-sizeresponseto hydraulicroughness
at reachscales,simplifyingour analysis.Over sufficientlylongreachesin channels
with relativelyslowlyvaryingdischarges,
the reach-averagetotal boundary shear stresscan be approximatedby a depthslope product (70 = p#hS, where h and S are the reachaveragechanneldepth and slope,respectively).
Applyingthis
approximation
andremembering
that7cSo
= 7;œ• 70¾for
our low-roughness
referencechannel,(4) is rewrittenas
p#hS
visuallybasedmethodsof determiningincipientmotion and
Ds0
=
7'c50(Psp)#
(5)
one whichmay minimizeerror causedby neglectof roughness
elements[Buffington
and Montgomery,
1997].Rearrangingthe whereh andS arereach-average
bank-fullvalues.We leave(5)
Shieldsequation allowsdeterminationof the competentme- unsimplifiedbecauseour ultimategoalis to predicta reference
dian grain sizefor a givenbed stress
D so as a function of the bank-full shear stress.Use of the
Ds0= 7'/[7c*s0(p•-p)#]
(4) depth-slopeproductassumessteady,uniform flow.Paola and
Mohtig [1996]arguethat to assumequasi-steady
flow, signifiwhere 7' -- 7cSo. We define 7cSo specificallyas the bank-full cantdischarge
fluctuationsshouldoccuron timescales>>u/#S
bed stress(7;f), whichis approximately
equalto the total (where u is the reach-averagedownstreamvelocity),while
quasi-uniformflow requires study reach lengths >>h/S. In
bank-full
boundary
shear
stress
(7o¾)
forourlow-roughness
reference channel. We choose bank-full flow as our reference
hydraulicallycomplexforestchannelsthe depth-slopeproduct
conditionbecauseit is the practical limit of shear stressin is valid only as a reach-averageapproximationof the total
natural channelswith well-developedfloodplainsand self- boundaryshearstress.Large, frequentlyspacedroughness
elformedbeds(i.e., all sizesmobileat typicalhighflows).Fur- ementscauselocallynonuniformflow,makingthe depth-slope
thermore, the bank-full flow is believedto be a morphologi- productinappropriatefor subreachscales.
cally significantdischargefor gravel-bedrivers[Wolmanand
Miller, 1960;Henderson,1963;Li et al., 1976; Carling,1988],
manyof whichexhibita near-bank-fullthresholdfor significant 3. Study Sites and Methods
To examinetexturalresponseto hydraulicroughness,
a field
sedimenttransport[Leopoldet al., 1964;Parker, 1978, 1979;
Howard, 1980;Anderews,1984].
studyof plane-bedchannels(definitionof Montgomeryand
BUFFINGTON
AND
MONTGOMERY:
EFFECTS
OF HYDRAULIC
ROUGHNESS
3509
Buffington[1997]),wood-poorpool-rifflechannels,andwoodrich pool-riffle channelswas conductedin forest mountain
drainagebasinsof northwesternWashingtonand southeastern
Alaska(Figure1). Thesethreechanneltypesrepresenta general cumulativeadditionof bank,bar, andwood-debris
roughnessanda progressive
decrease
in z' for a givenZo(equation(2)).
3.1.
Olympic Peninsula
We surveyed
fourteenchannelson the OlympicPeninsulaof
northwesternWashington.The OlympicPeninsulais characterized by mountainousterrain and a coastalrain forest of
Sitkaspruce(Piceasitchensis),
westernhemlock(Tsugaheterophylla),redcedar(Thujaplicata),andDouglasfir (Pseudotsuga
taxirolla).Bedrockgeologyof the peninsulais predominantly
composedof Eoceneto Miocenemarinebasaltsand sediments
[Taborand Cady,1978a,b]. The Olympicstudysitesoccupy
watersheds
influencedby Pleistocene
glaciationand are typically incisedinto fluvioglacialdeposits.Channelwidths and
slopesof our studysitesrangedfrom 5 to 13 rn andfrom 0.0040
to 0.0265,respectively
(Table 1).
Loggingactivityon the OlympicPeninsulais generallyintensive,with some forestson their third harvest rotation within
the last 100years.Most studysiteshad riparianbufferscomposedof mixedshruband conifer,althoughsomestudysites
had been dear cut to channelmargins,resultingin riparian
forestsof mixedshrubandred alder(Alnusrubra).Coniferous
buffersweretypicallysecondgrowth,indicatinga longhistory
of loggingdisturbance.Hillslopefailuresare commonin the
OlympicPeninsula,andrelict debris-flowdepositsform channel-marginterracesin three of the studyreaches,indicating
both long run-outpathsand the potentialfor periodiclarge
sedimentinputs.
At eachstudysite,three crosssectionsand a center-linebed
profile were surveyedin reachesthat were 8 to 18 channel
widthslong.Detailed topographic,
textural,and wood maps
alsowere constructedusinga digital theodolite.Bed surfaces
were commonlycomposed
of spatiallydistincttexturalpatches
(i.e., grainsizefacies)of differingparticlesizeand sorting.A
standardprocedurewasdevelopedto classifytexturalpatches
[Buffington
andMontgomery,
1999].Surfacegrainsizesof each
patchtypewere determinedfrom randompebblecounts[Wol-
man,1954]of 100+ grains.Thesesamples
were,in turn,spatiallyaveragedby patchareato determinereach-average
grainsizestatistics.
The lowerlimit of grain-sizemeasurement
was2
mm, with smallersizesgroupedas a singlecategory.We did
not truncatedata collectionat the lower limit of grain-size
measurementas is commonlyrecommended[Kellerhalsand
Bray, 1971;Churchet al., 1987]becauseit can distortthe size
distribution.
Particlesizessuspendable
at bank-fullstagewere
removedfromthe grain-sizedistributions
to separatebedload
from suspendedload. The maximumsuspendablesize was
calculatedfrom Dietrich's[1982] settlingvelocitycurves,assuminga Coreyshapefactorof 0.7, a Power'sroundness
of 3.5,
and a settlingvelocityequal to the bank-fullshearvelocity.
Suspendable
grain sizesrangedfrom 2 to 8 mm.
3.2.
Southeastern
Alaska
We supplementedour Olympicsurveywith data from 27,
southeastAlaskan,coastalchannelsstudiedby Wood-Smith
Figure 1. Photographs
of (a) plane-bed,(b) wood-poorpooland Buffington[1996] (Table 1). The Alaskanstudyareasare
riffle, and (c) wood-richpool-rifflechannelsof northwestern
characterized
by steepglaciatedterrain,maritimeclimate,and Washingtonand southeasternAlaska.
rain forestspredominantly
composed
of Sitkaspruceandwestern hemlock.Hillslopesare commonlygroovedby avalanche
3510
BUFFINGTON
Table la.
AND
MONTGOMERY:
OF HYDRAULIC
ROUGHNESS
Reach-AverageOlympic Channel Characteristics
W,
h,*
L,
Ds0,?
S
m
m
m
mm
Dry 2{}
Alderll
Hoko 2
Hoh 1
Hoh 2
0.0126
0.0265
0.0160
0.0059
0.0114
6.84
8.68
5.12
11.53
10.08
0.59(0.69)
0.57(0.64)
0.34 (0.37)
0.55 (0.62)
0.51 (0.55)
60
154
60
142
95
Skunk2
Hoko 1
Pins 1
0.0126
0.0085
0.0090
6.79
12.81
6.75
0.83(0.89)
0.78 (0.89)
0.86 (0.92)
85
134
64
Pins2
Flu hardy
Mill
Dry 1{}
Skunk1
Cedar
0.0133
0.0105
0.0143
0.0222
0.0040
0.0046
6.90
6.64
8.41
7.96
13.39
10.95
0.61(0.66)
0.53(0.55)
0.83(1.04)
0.60(0.73)
0.78(0.85)
0.72 (0.85)
70
69.5
73
70
140
100
Channel
EFFECTS
SEs0,?
LWD Per
mm
SquareMeter
(a/X)t,
(a/X)w
0.0 (0.0)
5.9 (5.5)
0.0 (0.0)
8.3 (0.8)
7.6 (2.5)
0.0000
0.0015
0.0033
0.0006
0.0052
0.000
0.010
0.008
0.008
0.010
0.018
0.023
0.024
0.008
Wood-PoorPool-Riffle
39.3(41.2)
1.11(0.98)
35.0(35.7)
1.14(1.09)
31.4(34.0)
1.55(1.29)
9.1 (8.4)
6.1 (6.1)
3.3 (3.4)
0.0243
0.0157
0.0278
0.012
0.033
0.012
0.056
0.060
0.042
Wood-RichPool-Riffle
36.0(39.5)
0.96 (0.92)
24.4 (29.1)
0.89 (0.85)
19.4(24.0)
0.98(0.87)
53.4 (55.4)
0.96(0.91)
19.8(23.8)
0.95(0.86)
27.2(29.7)
0.72(0.72)
7.8 (7.2)
5.2 (2.6)
4.8 (4.1)
10.8(10.8)
4.9 (2.2)
5.9 (9.5)
0.0580
0.0455
0.1450
0.0431
0.0352
0.0438
0.026
0.044
0.048
0.029
0.030
0.037
0.112
0.118
0.120
0.079
0.072
0.085
O'g(tb)
t,$
Plane-Bed/Incipient
Pool-Riffle
67.1(69.2)
1.24(1.17)
52.8 (58.0)
1.60(1.28)
54.8 (55.2)
1.34(1.31)
19.3(41.9)
2.17 (1.19)
56.8 (62.1)
1.59(1.48)
S (center-linebed slope),W (bank-fullchannelwidth),h (crosssectionally
averagedbank-fullchanneldepth),Ds0 (medianbed-surface
grain
size),% (graphicstandarddeviation[Folk,1974]),LWD per squaremeter(totalwoodloading,piecesper squaremeter),(a/X)t, (bar
amplitude/wavelength),
and (a/X)w (streamwisebank topography,amplitude/wavelength,
one sideof channel)are reach-average
values.L is
reachlength.SEs0is the standarderror of the reach-average
Ds0 (seetext).
*First valuesare crosssectionallyaveragedoverthe channelbed andbanks,while thosein parentheses
are averagedoverthe channelbed only.
We usethe latter for calculatingthe bank-fulldepth-slopeproductin equation(5).
?Valuesin parentheses
are for grain-sizedistributions
with suspendable
sizesremoved(seetext).
SHere% isthegraphic
standard
deviation,
defined
as((bs4
- qb•6)/2
[Folk,1974],where(b84
andqb16
arethelog2grainsizes[Krumbein,
1936]
for which 16% and 84%, respectively,of the surfacegrain sizesare finer.
{}Evidenceof ancient,catastrophicsedimentinputsfrom debrisflows.
IIEvidenceof recent debris-flowinputs.
ôBed load and suspendedload material could not be differentiatedas the maximumsuspendablegrain size was <2 ram, the minimum
resolutionusedfor our surfacepebblecounts.
chutesand slopefailures.The geologyof southeastern
Alaska
is characterizedby all major rock typesof agesrangingfrom
Proterozoic(?)to Quaternary[GehrelsandBerg,1992],largely
accretedduring the Cretaceousto Eocene [Goldfarbet al.,
1988;Gehrelset al., 1990].Southeastern
Alaskawasprofoundly
influencedby Pleistoceneglaciation[Reed,1958], and many
channelsdrain glaciallycarvedvalleysand cirques.The Alaskan studysitesincludeboth pristineold growthenvironments
and areas heavily disturbed by timber harvesting.Pristine
channelswere characterizedby high wood loading,while disturbed channelswere generallyclear-cutto the streambanks
and had mostor all of their in-channelwoodremoved[WoodSmithand Buffington,1996].Channelwidthsand slopesof our
studysitesranged from 5 to 29 m and from 0.0017 to 0.0267,
respectively
(Table 1).
At eachAlaskan studysite,five crosssectionsand a centerline bed profile were surveyedwith an engineer'slevel over
reachesthat were 8 to 23 channelwidthslong.A bank-to-bank
randompebblecount[Wolman,1954]of 100+ grainswasconducted at each crosssection,with grainssmaller than 2 mm
groupedas a singlecategory.Reach-averagegrain-sizestatisticswere determinedfrom simple averagesof thesesamples,
with the suspendable
load removedasdescribedin section3.1.
Suspendablegrain sizesranged from 1 to 9 mm.
3.3.
Wood Loading
The studysitesexhibit a continuumof wood loading that
stronglyinfluencespool spacing(Figure2). We usedthe me-
dianvalueof thiscontinuum
(0.03pieces/m
2) to dividethe
channelsinto wood-poorand wood-richcategories.
4.
Results
4.1.
Reach-Scale Response
Our field data demonstratethat for a given reach-average,
totalbank-full
shear
stress
(roy),channels
withgreater
roughness(andthereforelowerr;f (equation(2)) havesystematicallysmallerreach-average
surfacegrainsizes(Figure 3). The
progressiveincreasein bank, bar, and wood roughnesscauses
a correspondingreductionin reach-averageD so at our study
sites. This result is consistent with several other studies that
demonstrate
that wood
removal
from
forest
channels
causes
bed-surfacecoarsening,presumablybecauseof increasedbed
shearstressresultingfrom lossof woodroughness
(seereview
byLisle [1995]).The solidline in Figure3 is the low-roughness
prediction
of competence
(see(5), r*c5o= 0.03). Asexpected,
the data showthat surfacegrain sizesapproachthe theoretical
competencecurveat low hydraulicroughness
(plane-bedmorphology),as r;f approachesro ß
There is considerable
variabihtyin the magnitudeof textural
fining causedby bank, bar, and wood roughness(Figure 4).
Although the distributionsof textural responseoverlap, the
centraltendencies
of texturalfining(25th-75thpercentiles)are
distinctfor each channeltype (Figure 4). The ratio of observed-to-predicted
Dso has a medianvalue of 0.53 for planebed channels,0.30for wood-poorpool-rifflechannels,and 0.18
for wood-richpool-rifflechannels;this representsa reduction
of roughly40% from one channeltypeto the next.In channels
with significanthydraulic roughness(wood-rich pool-riffle
channels),observedD so can be up to 90% lessthan the predicted competentvalue for the bank-full stage.
BUFFINGTON
Table lb.
AND
MONTGOMERY:
EFFECTS
ROUGHNESS
3511
Reach-AverageAlaskan Channel Characteristics
W,
h,*
L,
Ds0,?
S
rn
rn
rn
mm
Maybeso3
Maybeso4
Indian
Weasel[I
0.0024
0.0036
0.0122
0.0025
27.07
24.48
24.60
15.10
1.03(1.11)
1.06(1.12)
1.17(1.32)
0.92(1.01)
324
436
480
187
36.4(39.6)
46.4(47.7)
79.4 (85.1)
25.6(48.2)
12 Mile 1
12 Mile 2
Maybeso1
Maybeso2
Cable
FUBAR 1
FUBAR 2
Muff
Bambi
0.0021
0.0028
0.0095
0.0065
0.0017
0.0106
0.0127
0.0150
0.0102
23.34
22.47
22.31
29.12
16.89
17.84
16.32
14.29
4.6
0.95(1.05)
1.03(1.10)
1.13(1.24)
0.97(1.10)
0.88(0.96)
0.62(0.66)
0.79(0.85)
0.59(0.64)
0.32
360
170
400
500
300
360
300
300
80
24.3(26.6)
21.9(24.9)
49.6(53.2)
36.2(38.6)
20.6ô
42.9(47.5)
57.6(60.8)
44.3(53.5)
16.3(17.7)
Hook{}
Trap 1
Trap 2
Trap 3
Trap 4
Trap 5
Trap 6
E Fk Trap 1
E Fk Trap 2
Fowler1
Fowler2
Fish 1
Fish2
Greensl[
0.0110
0.0055
0.0074
0.0072
0.0071
0.0102
0.0120
0.0133
0.0127
0.0063
0.0054
0.0267
0.0224
0.0220
21.37
12.92
15.59
11.84
9.67
14.11
15.76
15.72
10.65
18.03
11.46
19.18
12.88
12.90
0.82(0.88)
0.84(0.95)
0.66(0.77)
0.60(0.66)
0.68(0.76)
0.68(0.75)
0.71(0.78)
0.78(0.87)
0.58(0.66)
0.65(0.74)
0.68(0.83)
1.12(1.33)
0.56(0.60)
0.66(0.85)
250
165
220
220
220
175
200
200
172
225
210
167
280
260
27.4(31.9)
17.0(17.8)
15.4(16.0)
13.9(16.5)
13.5(16.2)
17.6(21.2)
13.4(18.7)
19.9(23.5)
24.5(31.0)
14.1(18.5)
19.0(24.5)
30.6(40.9)
45.8(48.3)
34.9(47.0)
Channel
OF HYDRAULIC
o¾((b)
?,$
SE50,?
LWD Per
mm
SquareMeter
(a/X)•,
(a/X)w
2.1 (2.8)
4.1 (3.9)
10.0(9.3)
8.8 (6.0)
0.0132
0.0058
0.0025
0.0230
0.007
0.004
0.014
0.014
0.021
0.020
0.054
2.6 (2.1)
3.2 (3.0)
5.1 (5.6)
4.3 (4.2)
6.4ô
9.2 (7.7)
4.5 (4.4)
5.2 (4.3)
2.0 (1.7)
0.0106
0.0108
0.0000
0.0112
0.0234
0.0290
0.0070
0.0123
0.0136
0.010
0.016
0.015
0.013
0.023
0.012
0.014
0.015
0.025
0.045
0.040
2.1 (1.6)
3.0 (2.8)
1.7(1.4)
1.7(0.7)
1.1(0.8)
1.4(1.1)
1.5(2.3)
1.4(1.3)
3.9 (1.9)
3.6 (3.6)
7.6 (9.5)
4.0 (3.3)
2.5 (2.4)
6.8 (5.7)
0.0405
0.0644
0.0635
0.0639
0.0698
0.0591
0.0566
0.040
0.040
0.037
0.045
0.032
0.048
0.038
0.038
0.051
0.023
0.034
0.052
0.034
0.039
Plane-Bed/Incipient
Pool-Riffle
>1.50 (1.17)
0.88(0.82)
1.67(1.54)
>2.34 (1.45)
Wood-PoorPool-Riffle
1.44(0.89)
1.90(0.90)
1.76(1.51)
1.30(1.16)
> 1.97ô
> 1.50(1.15)
1.58(1.41)
>2.13 (1.29)
>1.08 (0.85)
0.083
0.071
0.024
0.060
0.075
Wood-RichPool-Riffle
>1.78 (1.07)
1.04(0.97)
1.30(1.11)
> 1.72(1.21)
>1.83 (1.34)
>1.54 (0.98)
>1.88 (1.10)
> 1.84(1.00)
1.79(1.20)
>2.03 (1.55)
>2.06 (1.57)
1.32(0.90)
1.33(1.23)
>2.16 (1.58)
0.0531
0.0313
0.0358
0.0601
0.0311
0.0303
0.100
0.058
0.048
0.126
0.077
0.055
See Table la for explanatoryfootnotes.
4.2. RoughnessConfiguration and Magnitude
marilythroughform drag,the magnitudeof whichdependson
the frequency,size, orientation,and height abovethe bed of
Thescatter
ofD5ovalues
in Figure3 fora givenroyand
channeltypereflectssite-specific
differencesin roughnesscon- the in-channellogs and rootwads.Similarly,the form drag
figuration.For example,woodcreateshydraulicroughness
pri- causedby bars dependson their amplitudeand wavelength
[Nelsonand Smith, 1989]. Severalsourcescontributeto what
we collectivelycall bank roughness:
(1) proximityof channel
banks (so-calledwidth-to-deptheffects) and associatedmo20
mentumdiffusion[Leighly,1932;Parker,1978];(2) roughness
Olympic Alaskan
o
ß
plane-bed/
length scale(i.e., skin friction) of the material forming the
incipientpool-riffle
banks
[Einstein,1934, 1942;Houjou et al., 1990]; (3) down[]
ß
pool-riffle
streamvariationsin channelwidth that effectivelyforce lateral
form draganalogousto bed form drag;and (4) riparianvegetation protrudingfrom the banksand causingadditionalform
drag.
The physicalchannel characteristicsthat causebank, bar,
wood-poor • wood-rich
and woodroughnessshowa statisticallysignificantincreasein
o
magnitudeacrossthe three channeltypesstudied(Figure5).
For example,while there is no significantdifferencein widthto-depthratios (W/h) amongstthe channels(Figure 54 and
[]
Table 2), there is a significantincreasein streamwisebank
0
!
!
0.00
0.05
0.10
0.15
topographyand consequentform drag (Figure 5b and Table
2). Streamwisevariationof channelwidth at the studysites
Woodloading(pieces/m
2)
resultsin undulatingbanksand lateral form drag,the magniFigure 2. Pool spacingas a functionof wood loading.Pool tude of which dependson the amplitude-wavelength
ratio of
spacingis expressedin channelwidths per pool, defined as
the bank undulations((a/X)w, a reach-average
valuefor each
(L/W)/number of pools,where L is reach length and W is
bank-fullwidth.Wood loadingis definedasnumberof pieces/ side of the channel).This ratio increasessignificantlyacross
(WL). Although the ordinate and abcissacontain common the three channeltypes(Figure5b and Table 2), indicatinga
increasein hydraulicroughness
becauseof greater
factors,the observedrelationshipis not spurious[Buffington, progressive
1998].
bank form drag.
I
I
I
I
3512
BUFFINGTON
AND
MONTGOMERY:
EFFECTS
OF HYDRAULIC
ROUGHNESS
lOOO
Species
1.
Brook
•out
(Salvelinusfontinalis)
2. hnk salmon(Oncorhynchus
gorbuscha)
3. Coho sa•on (O. kisutch)
5. R•nbow
•out
(O.
mykiss)
4.
Steelhead
•out
(O.
mykiss)
I 6. Sockeye
salmon
(O.nerka)
7.Brown•out(Salmotru•a)
[8.
Chum
salmon
(O.
keta)
9. CMnook salmon(O. tsha•tscha)
Roughness
sources
Olympic Alaskan
plane-bed/
incipientpool-riffle
10
ß
[]
wood-poorpool-riffle ¸
[]
wood-richpool-riffle O
[]
banks(seetext),
incipient
bars
I
+
Increasing
100
bars
cumulative
+
roughness
wood
•
1000
Reach-average
totalbankfullshearstress
('170bf)
(Pa)
Figure 3. Median bed-surfacegrain sizeversustotal bank-fullboundaryshearstressstratifiedby channel
ty,
pe.=The
solid
lineisthe
low-roughness
prediction
ofcompetent
Dso
asa function
offrom
,%f (equation
(5),
a debris' flow.
* cSo
0.03).Circled
point
isaplane-bed
channel
recently
impacted
bysediment
input
Error barsrepresentthe standarderror;wherenot shown,the erroris smallerthanthe symbolsize(Table 1).
Also shownare rangesof D so valuespreferredby spawningsalmonids[Kondolfand Wolman,1993], the
significance
of whichis discussed
later in the paper.The reportedrangesof spawninggravelslikely contain
considerable
error andmaybe biasedtowardsmallsizesbecause(1) streambedstypicallyweresampledusing
the approachof McNeil andAhnell [1960]whichcombinessurfacean subsurface
materialand (2) in some
casesthe coarsetails of the sizedistributionwere arbitrarilytruncated[Kondolfand Wolman,1993].
Differencesin bank skinfriction and protrusionof riparian againstbanks,causingscourand the developmentof locally
vegetationinto the channelwere not quantified.However,the wide sectionsof channel,or they can armor banksand maingreatestamountof vegetativeprotrusioninto the channeltyp- tain locally narrow channelwidths; both effects commonly
icallyoccurredat thewood-poorpool-rifflesiteswheregrowth occurwithin a singlereach.While woodloadingalsoinfluences
of riparian shrubsand deciduoustreeswas stimulatedby recent clear-cutting.
Like streamwisebank topography,the amplitude-wavelengthratio of bars ((a/X)t,) showsa statisticallysignificant
increaseacrossthe three channeltypes(Figure 5c and Table
2), indicatinga progressive
increasein roughness
due to bed
form drag.Two scalesof bars are includedin Figure5c: macroscalebars (at, > 0.5 h) and mesoscale
bars (0.25 h <
at, < 0.5 h). Bar formswere identifiedvisuallyfrom detrendedbed profiles.
There is also a significantincreasein wood loading across
the threechanneltypes(Figure5d andTable2), resultingin a
n=9
n=12
n=ø20
progressive
increasein hydraulicroughness
dueto woodform
plane-bed/
wood-poor
wood-rich
incipientpool-riffle
pool-riffle
pool-riffle
drag.Wood loadingplaysan importantmorphologicand hydrologicrole at the studysites,not onlyinfluencingpool spac- Figure 4.
Box plotsof texturalfiningdefinedasthe ratio of
ing (Figure2) but alsocontrollingthe amplitude-wavelengthobserved-to-predicted
D so. Predictedvalues are calculated
ratio of both bank and bed topography(Figure 6). Forest fromequation(5). The linewithineachboxisthe medianvalue
channelscommonlyexhibitconsiderable
widthvariationwithin of the distribution,box endsare the inner and outer quartiles,
a singlereach[Trimble,1997]becauseof morphologicforcing whiskersare the inner and outer tenths, and circlesare the exper distribution.
causedby in-channelwood.Wood obstructions
canforceflow trema.Variablen is the numberof observations
,
,
BUFFINGTON
35
i
AND
MONTGOMERY:
EFFECTS
OF HYDRAULIC
ROUGHNESS
3513
I
o-•
b)
o
25,
•
o
20
15,
o
10
o
n=9
5
n=12
•
n=7
.06
i
i
n=10
n=12
plan•-be• woo•-poor wooci-rich
plan•-bed/ woofS-poor wooti-rich
incipientpool-riffle pool-riffle
•
n=20
pool-riffle
pool-riffle
incipientpool-riffle p•l-fiffie
i
.16
• .05
.14,
12,
.10•
•
.03
.08,
.06,
• .02
•c•.01
•
o
n=9
.04,
n=9
plane-bed/
n=12
wood-poor
incipient
pool-riffle pool-riffle
n=12
.02,
n=20
n=19
0
wood-rich
plane-bed/
pool-riffle
wood-poor
incipient
pool-riffle pool-riffle
wood-rich
pool-riffle
Figure 5. Boxplotsof reach-average
(a) width-to-depthratio, (b) amplitude-wavelength
ratio of streamwise
banktopography,
(c) amplitude-wavelength
ratio of bar topography,
and (d) woodloading.See Figure4
captionfor box plot definition.Valuesusedfor theseplots are listedin Table 1.
bed topography(Figure 6b), we find that wood affectsthe
spacingof barsmore so than their amplitude(Figure 7). Our
findingsindicatethat beyondcreatingits own form drag and
hydraulicroughness,
woodforcesgreaterbankandbed topographyand consequently
greaterform drag at our studysites.
4.3.
Subreach-ScaleResponse
Detailed field measurements
at the Olympicsitesallow examination of subreach-scale
textural responseto hydraulic
roughness.
Representative
mapsof eachchanneltypeillustrate
characteristic
variationsin topography,
woodloading,and surfacetexture(Figure8). The numberof texturaltypesobserved
in a reachvariedfrom one to seven,with texturescomposedof
grainsizesrangingfrom silt to smallboulders(Table 3). Twosamplemediantests(a nonparametrickind of t test [Conover,
1971])indicatethat almostall texturaltypeswithin a reachare
significantly
differentfrom one another(P -< 0.05), while
most patchesof the same textural type within a reach are
statisticallysimilar (P > 0.05).
Faciesmappingdemonstrates
that channeltype androughnessconfiguration
stronglyinfluencethe number,frequency,
and spatial arrangementof surfacetextures.Plane-bedchan-
nels exhibitone to four grain-sizefaciesbut are frequently
monotextural(Table 3). Each of the wood-poorpool-riffle
channels
arecomposed
of fourtexturaltypes,whilewood-rich
pool-riffle channelsexhibit three to sevenfacies types per
reach(Table 3). Similarly,the total numberof texturalpatches
within a reachrangesfrom 1 to 8 in plane-bedchannelsbut
increases
to 13-24 in wood-poorpool-rifflechannelsand 17-55
in wood-richpool-rifflestreams(Table 3). Furthermore,the
spatialarrangementof texturesis progressively
more complicatedin the three channeltypesstudied(Figure8).
Textural patcheslikely representspatialdivergenceof sediment supply and transport capacity, causing local, sizeselectivevariationsin sedimentflux that lead to patch development. Our field observations suggest that increased
frequencyand magnitudeof flow obstructions(i.e., bars and
wood) enhancethe spatial divergenceof sedimentflux and
patch development.For example, the number of textural
patchesin a reachis relatedto the frequencyof woodobstructions (Figure 9). Although the stochasticnature of wood re-
Table 2. Comparisonof Roughness
Characteristics
BetweenChannelTypes
P Values
Wood
Plane-bedversuswood-poorpool-riffle
Wood-poorpool-riffleversuswood-richpool-riffle
Plane-bedversuswood-richpool-riffle
Pieces Per
W/h
(a/X)w
(a/X)t,
SquareMeter
0.801
0.465
0.599
0.024
0.029
0.002
0.044
<0.001
0.001
0.044
<0.001
<0.001
ReportedP valuesare for two-samplemediantests(a nonparametric
sort of t test [Conover,1971])
evaluated
witha one-tailed
X2statistic.
Differences
between
distribution
means
areconsidered
statistically
significantwhen P -< 0.05.
3514
BUFFINGTON
0.14
•
AND
' •A
MONTGOMERY:
I
EFFECTS
'
0.12
0.0
0.08
0.061
[
•• 0.04'
[•
•O
0.02
wood-poorpool-riffle
/ [
•Awood-rich
pool-riffle
/[
•
0.00
.
.
•
Y=&23•
1•
0.05
.
0.10
0.15
Reach-average
woodloading
(pieces/m
2)
b)
I
0.06
,
I
0.05
0.04
ß
AA
.
0.03
0.02
I•
0.01
ROUGHNESS
Discussion
Our resultsdemonstratethat reach-averageD so is system-
aticallyfiner in channelswith greater hydraulicroughness.
However, the overall trend of the data in Figure 4 has a
somewhatlower slope than the reference D so prediction
(equation(5)). This apparentdisagreementbetweentheory
and observationmay be due to the small size of our data set.
For example,data from gravel-bedrivers in Colorado JAndrews,1984] supportthe grain-sizepredictionquite well (Figure 11a); the data trend with the prediction,and the limit of
competencein thesechannelsis well-describedby the prediction. Thesedata alsosupportour hypothesis
regardingtextural
responseto hydraulicroughness,
demonstrating
that channels
with thickerriparianvegetation(and thereforegreaterhydraulic resistanceoffered by the banks) have relatively smaller
bed-surfacegrain sizes.Thicker riparian vegetation also increasesbank strength,promoting smallerwidth-depth ratios
that may reduceboth bed shearstressand surfacegrain size.
Data from gravel-bedriversin the United Kingdom[Heyand
Thorne,1986] alsotrend with the grain-sizepredictionwhich,
again,forms a good upper envelopeof channelcompetence
(Figure lib).
It is importantto note, however,that only gravel-bedchannelswith plane-bedmorphologies(definitionof Montgomery
and Buffington[1997]) have the potentialto realize the pre-
•ane-•incipient
pool-fiffl
•
0.00
5.
OF HYDRAULIC
_
O
O
•1O •1
plane-bed/incipient
pool-riffle
dicted referencevaluesof Dso. Both lower-gradientsand-bed
riversand steeper-gradient
boulder-bedrivershavecharacteristic channelmorphologiesand accompanying
roughnesselements that cause r' << to, and therefore observedvalues of
.
'A •vood-poor
pool-riffle
wood-rich
pool-riffle
y = 0.009 + 0.55x, R^2 = 0.70
•
0.00
0.00
0.05
0.10
6
0.15
Reach-average
woodloading
(pieces/m
2)
Figure 6. Reach-averageamplitude-wavelength
ratio of (a)
banktopographyand(b) bar topographyasa functionof wood
loading.Mill Creek (far right triangle) is excludedfrom the
curve fits.
n=8
cruitmentcreatescomplex,irregulartexturalpatternswithin
our studysites,more predictablepatternsof texturalpatches
have been documentedin channelswith lesschaoticarrangementsof flow obstructions.
For example,regular patternsof
texturalpatchesare foundin self-formedmeanderingchannels
becauseof systematicdownstreamand cross-channel
variations in shearstressand sedimentflux causedby channelcurvature, topographicallyinducedconvectiveaccelerations,
and
lateralbed slope[Dietrichet al., 1979;Dietrichand Smith,1984;
Parker and Andrews,1985].
Paired surfaceand subsurfacesamplingof textural patches
demonstrates
a strongcorrelationbetweensurfaceand subsur-
pland-bed/
incipientpool-riffle
n=12
n=20
wooci-poor
woo•l-rich
pool-riffle
pool-riffle
I
-½
face mediangrain sizes(Figure 10). Coarsertexturalpatches
havecorrespondingly
coarsersubsurface
sizes[seealsoBuffingtonand Montgomery,1999].Moreover,we observelittle to
n=9
n=12
n=20
no armoring at our sites.We find that ratios of surface-towoo•-rich
plane-bed/
wood-poor
subsurface
D5o are within the typicallyobservedrange of 1-3
pool-riffle
incipientpool-riffle
pool-riffle
[Milhous,1973;Bathurst,1987;Kinerson,1990;Pitlickand Van
Streeter,1998] but tend to clusternear 1 for our studysites, Figure 7. Box plots of reach-average,dimensionlessbar
indicatingpoorly armoredsurfaces.Hydraulicroughnessand wavelengthand amplitude.See Figure 4 captionfor box plot
reducedr' likely inhibit armor developmentin thesechannels. definition.
O
BUFFINGTON
AND
MONTGOMERY:
EFFECTS
OF HYDRAULIC
ROUGHNESS
3515
a)
I
Flow
i i• i •11•
i i• i •ll•J
ß.•,•.!.•'•
-,..-,,,,.,...
.....•.•
cG,D50=54.8
mm,
tSg=l.34
4)
Contour interval=0.1
0
On-bank
treeprojecting
intochannel
5
m
10
I
meters
F/ow
b)
Pool
•
Pool
logs,
rootwadsContour
interval=0.1
ml
•cGcvc,D50=47.4
mm,
(•g=0.92
(D • Discrete
Debris
pile
(branches,
wood
chips)
0 5 1•0
.'"'i•cGmcvc,
D50=38.4
mm,
•g=1.24
•
'$4•G,D5O=13.1
mm,
C•g=0.98
(D
sG,
D5o--6.0
mm
• On-bank
tree
projecting
into
channel
meters
Figure 8. Morphologicand textural plan maps of typical (a) plane-bed(Hoko River 2), (b) wood-poor
pool-riffle(Hoko River 1), and (c) wood-richpool-riffle(Mill Creek) channelsstudiedon the Olympic
Peninsula[from Buffingtonand Montgomery,1999]. See Table 3 for texture definitions.Boundarybetween
channelbed and walls definesthe lateral margin of the maps.
D so much less than those predictedfrom (5) (Figure 12).
Sand-bedriverstypicallyhavea dune-ripplemorphologycharacterizedby multiple scalesof bed forms(ripples,dunes,and
bars)that providesignificantform drag,while steeper-gradient
boulder-bed channelstypically have step-pool and cascade
morphologiescharacterizedby tumbling flow, low width-todepth ratios,and boulder form drag, all of which create considerable channel roughness[Montgomeryand Buffington,
1997].Dune-ripplechannelsalsoexhibitsedimenttransportat
stagessignificantlylessthan bank-full, indicatingthat the observedDso shouldbe muchlessthan a theoreticalcompetent
Dso predictedfrom bank-full shearstress.
Even when limited to gravel-bedmorphologies(i.e., planebed and pool-rifflechannels),(5) overpredictsthe competent
Dso in channelswith steepslopesbecauseof unaccounted-for
effectsof particle form drag. As grain size becomesa significant proportion of the flow depth, it createsform drag that
diminishesz' and the competentD5o, an effect that is not
3516
BUFFINGTON
C) Pqol
AND
x••
MONTGOMERY:
EFFECTS
OF HYDRAULIC
ROUGHNESS
Flow
Pool
pile
(branches,
wood
chips)
'.'".•(i•Gmvcc,
D50=23.7
mm,
C•g=1.04
0 •[l• Debris
•Gcfm_vffm,
D50=8.4
mm,
C•g=0.93
0 • On-bank
tree
projecting
into
channel
iS, D50--2.0
mm
Contour interval=0.1 m
0
5
10
,•Z, D50•0.065
mm
•' Discrete
logs,
rootwads
i
meters
Figure 8c. (continued)
accountedfor in (5). Particleform dragis representedby the
relative roughnessratio (D so/h). For givenvaluesof p, Ps,
and z *cso,(5) specifies
a uniquevalueof D sos/hfor eachvalue
of S
3). For the most part the standarderrors are smallbecause
evenin very patchyreachesthere is typicallya dominanttextural type that occupies->50% of the bed-surfacearea (Table
3), whichskewsthe distributionof texturaltypesandresultant
reach-averageD so toward the grain size characteristics
of the
dominanttexturaltype.
Becausewe did not quantify either the rate or caliber of
sedimentsupplyat the field sites,one may wonder if the textural fining observedin Figures3 and 4 is due to systematic
changesin either of these factors, rather than a systematic
increasein hydraulicroughnessand lowered z'. In particular,
laboratorystudiesdemonstratethat bed-surfacegrainsizevaries inverselywith sedimentsupplyrate [Buffingtonand Montgomery,this issue].However, it is unlikely that the textural
fining observedhere representseither an underlyingincrease
in sedimentsupplyrate or a decreasein supplycaliber for
severalreasons.(1) The highestsedimentloadingis expected
for sites in logged basins,where high sedimentproduction
D5o/h=S/Iz*c5o(•-l)]
(6)
Consequently,large particle form drag is predictedfor channels with steep slopes.Channel slope ranged from 0.0017 to
0.027 at our studysites,correspondingwith values of D so/h
equal to 0.03-0.5 (equation(6)) and a 2-32% overprediction
of competentDso becauseof unaccounted-foreffectsof grain
form drag [Buffington,1998].However,more than 80% of our
studysiteshad slopes<0.015, indicatinglessthan 17% overpredictionof competencefor the majorityof our data.
Becauseour reportedvaluesof Dso are reach averages,it is
important to examine whether the standard error of these
averagevalues exceedsthe inferred textural responseto hydraulic roughnessthus influencingour interpretationof the rates occur because of two factors: fluvial erosion and mass
data. The standarderror of the reach-average
Dso is definedas wastingcausedby poor road design[Montgomery,
1994;Bestet
al., 1995]andhillslopefailurescausedby reducedroot strength
SE0=
(7) following timber harvest [Sidle et al., 1985]. However, the
wood-poorpool-rifflesites)do not
For the Washingtonchannels,Ssois the standarddeviationof loggedsites(predominantly
show
the
highest
degree
of
textural fining (Figure 3). The
median grain sizesof texturesweightedby their area, and n is
with the greatestamountof texturalfining(wood-rich
the numberof texturaltypesper reach.For the Alaskanchan- channels
nels,Ssois the standarddeviationof mediangrain sizesof the pool-riffle channels)are predominantlyroadlessold growth
channel-spanning
pebblecounts,and n is the numberof peb- siteswith a lowerfrequencyof masswastingevents.Therefore
ble countsper reach. We find that the standarderrors are the observedtextural fining is not due to an increasingsedisubstantialat somesites,but the pattern of decreasinggrain ment supplyrate. (2) Lithology (a strongcontrol on grain
sizewith increasingroughnessdue to banks,bars,and wood is strengthand caliber)is highlyvariableacrossthe sitesand is
maintainedand is not obscuredby the standarderrors(Figure uncorrelated
with the threechanneltypes.(3) Drainagearea(a
BUFFINGTON
AND
MONTGOMERY:
EFFECTS
OF HYDRAULIC
ROUGHNESS
3517
Table 3. SurfaceTexture Compositionof the OlympicChannels
Percentage
Channel
Frequency
Texture*
of Bed
Dry 2
gC
Plane-Bed/Incipient
Pool-Riffle
100
1
Alder
G
Hoko 2
Hoh 1
Hoh 2
scG
cG
cG
S
cgS
cG
S
cgS
cG
gC
Skunk2
S
G
csG
cG
Hoko
sG
1
Pins1
Flu Hardy
Dry 1
Skunk1
Cedar
4
1
•
2
3
3
2
1
1
1
Wood-PoorPool-Riffle
9.33
9
11.19
9
8.22
4
71.24
2
67.1(69.2)
ira(
1.24(1.17)
-11
21.8(29.7)
56.1 (61.0)
54.8(55.2)
-2.0 (NA)
9.0 (42.7)
38.9 (40.3)
-2.0 (NA)
9.0 (42.7)
38.9 (40.3)
61.4(63.1)
1.76(1.39)
1.58(1.27)
1.34(1.31)
2.76 (1.26)
1.10(1.05)
1.54(1.49)
<2.0 (8.5)
9.9 (12.8)
25.0 (27.8)
50.4 (51.5)
>0.65 (0.40)
1.41(1.05)
1.37(0.98)
1.10(1.05)
2.76 (1.26)
1.10(1.05)
5.10
7
-6.0
9
4
3
3
13.1(13.8)
38.4(39.2)
47.4(47.7)
-2.0 (NA)
0.98 (0.87)
1.24(1.20)
0.92 (0.92)
4.15
2.58
6
3
9.2(11.0)
12.7(14.1)
1.32(1.02)
1.15(1.04)
cG
92.11
1
33.3(35.6)
1.57(1.31)
S
sG
cGm•cc
cGmc•c
Z
S
Wood-RichPool-Riffle
7.71
4
3.22
3
48.47
3
40.61
7
3.98
2
11.79
9
Gcfm
Mill
9.0
90.6
100
2
63
35
4
4
0.5
91.5
1
mm
13.87
61.58
19.45
1.16
G
cGmcvc
cGcvc
S
Gcfm
Gfcm_fm
c
Pins2
0.4
D so,
(Number/Reach)
-2.0 (NA)
7.1 (8.7)
29.6 (30.4)
52.4 (52.8)
-0.06 (NA)
-2.0 (NA)
0.92 (0.47)
0.93 (0.88)
1.01(1.00)
3.70
6
11.7(12.0)
0.92(0.90)
80.53
1.63
15.89
i
2
21
29.5 (29.9)
•0.06 (NA)
-2.0 (NA)
0.89 (0.85)
Gcfm_vffm
11.77
25
8.4(10.4)
0.93(0.67)
Gm•cc
Gcvc
S
scG
G
cG
gC
S
G ....
Gmvcc
Z
S
sG
G ....
Gmvcc
cG
gC
62.57
8.15
0.94
4.79
7.52
39.28
47.48
16.15
62.31
21.54
5.61
2.64
1.08
12.80
64.19
6.15
7.53
2
5
3
2
5
3
8
32
6
4
2
3
6
11
1
1
3
23.7 (24.5)
39.6 (39.6)
-2.0 (NA)
9.3 (33.9)
13.8(15.7)
39.1 (39.6)
77.0 (77.0)
-2.0 (NA)
21.2 (22.0)
29.2(29.2)
-0.06 (NA)
-2.0 (NA)
6.6 (7.2)
17.6(17.6)
26.6 (26.8)
35.3(35.3)
74.4(74.4)
1.04(0.94)
0.58 (0.58)
Gmc•c
Z
S
2.34(1.90)
1.09(0.87)
1.12(1.09)
0.66 (0.66)
0.96 (0.84)
0.91(0.91)
0.98(0.84)
0.65 (0.65)
0.70 (0.70)
0.82 (0.82)
0.89 (0.89)
*Texturesare namedusingthe Buffingtonand Montgomery[1999] classification
scheme.Capital letters
representthe dominantgrainsize(Z, silt;S, sand;G, gravel;and C, cobble),precedinglowercaseletters
representlessabundantgrainsizes,read asadjectives
modifyingthe uppercasenoun(s, sandy;g, gravelly;
and c, cobbley),and succeeding
lowercasesubscripts
further describethe grainsizecomposition
of the
dominantsizeclass(vf, veryfine;f, fine;m, medium;c, coarse;and vc, verycoarse).Order of lowercase
lettersindicates
relativeabundance
(leastto greatest).
Forexample,
sGfmissandy,
fineto mediumgravel.
Lowercasesubscripts
are usedto distinguish
otherwiseidenticaltexturalnames(e.g.,distinguishing
coarse
versusfine graveltextures).Sedimenttermscorrespondwith standardgrain sizeclasses[Buffington
and
Montgomery,1999, Table 1]. NA indicatesthe entire suspensionof a patch at bank-full flow.
?Valuesin parentheses
are are for grain-sizedistributions
with suspendable
sizesremoved(seetext).
$Heretraisthegraphic
standard
deviation,
defined
as((b84
- (•16)/2
[Folk,1974],where4)84
and4)16
are
the log2grain sizes[Krumbein,1936]for which 16% and 84%, respectively,of the surfacegrain sizesare
finer.
3518
BUFFINGTON
AND
MONTGOMERY:
EFFECTS
OF HYDRAULIC
ROUGHNESS
and resultant debris flows are characteristicof the steep,
mountainousterrain of northwesternWashingtonand southeasternAlaska.Evidencefor suchimpactsobservedat our field
sitesinclude(1) landslidetracksenteringa channel,in some
casesaccompanied
by a debrisjam, and (2) freshdebrisflow
levees,with inundated and scouredriparian vegetation.Although six of our studysitesshowedevidenceof debris-flow
input (Table 1), only one of the mostrecentlyaffectedchannelsshoweda strongtexturalresponseto debris-flowloading.
A more detailed studyof the influenceof hydraulicroughnesson reach-averagebed-surfacegrain sizewasconductedat
one of our wood-poor pool-riffle channelsand one of our
wood-richpool-riffle channels[Buffington,1998]. Bed shear
stresses for
0
20
40
60
80
100
Woodpiecesperreach
Figure 9. Textural patch frequency as a function of wood
frequencyat the Olympicstudysites.The fitted curveis forced
to one patch in plane-bedreacheswith zero piecesof wood.
these
sites were
calculated
from
a theoretical
stress-partitioning
model that wasverified througha courseof
field study.Resultsshowthat observedreach-averagevaluesof
D so are within 1-10% of thosepredictedfrom the bank-full
bed stress.Thesefindingsindicatethat bed-surfacegrain sizes
at thosesitesare in quasi-equilibriumwith bank-full channel
hydraulicsand suggestthat hydraulicroughness,rather than
sedimentsupply,is the dominantcontrol on grain size.
crudesurrogatefor sedimentsupplyrate andcaliber)is uncorrelated with channeltype.
The stronggrain sizeresponseto hydraulicroughness(Figure 3), and the lack of evidencefor an underlyingcovariance
with sedimentsupplythat would explainthe observedtextural
fining,suggests
that site-specific
differencesin sedimentsupply
rate and caliber are overwhelmedby bank, bar, and wood
roughnessat our studysites,exceptwhere recentcatastrophic
sedimentinputshaveoccurred.For example,one of the planebed studysitesrecentlyimpactedby a debrisflow has a reachaveragemedian grain size considerablyfiner than the other
plane-bedchannels(circledpoint in Figure 3), presumablyin
responseto catastrophicsedimentloading of the debrisflow.
Infrequent, catastrophicsedimentinputsfrom hollow failures
Andrews [1984]
thin bankvegetation
thickbankvegetation
10
100
Reach-average
bank-full
shear
stress
(Z0bf)
(Pa)
80
i ' i i i i i !i , i . i. ,l:f'"
i.
70
so
u
]
•/ I
,,'""
..--"'" ...
1
i
0
ß
i
10
ß
i
20
ß
i
30
-
i
40
-
i
50
ß
i
60
ß
•
,
, 11,
ß
100-
i
70
i
b)
2:1'
20
] ß,,4,..,&'"
......
'*"ß
0
I
1000
80
surfaceDs0 (mm)
Figure 10. Surfacemediangrainsizeversussubsurface
value
for textural patchessampledat the Olympicstudysites.Particle sizesthat are suspendable
at bank-fullstagewere removed
from both grain-sizedistributions(see section3). Subsurface
grain-sizedistributionswere determinedfrom sievedbulk samples,followingthe Churchet al. [1987]samplingcriterion(i.e.,
the largestgrain is -<1% of the total sampleweight).
:,d.½;:
"'
ß
10
[HeyandThorne,
1986]
100
1000
Reach-average
bank-fullshearstress
(Z0bf)
(Pa)
Figure 11. Median bed-surfacegrain sizeversustotal bankfull shearstressfor gravel-bedchannelsin (a) Colorado[Andrews,1984] and (b) the United Kingdom[Hey and Thorne,
1986]. Solid line is that of Figure 3.
BUFFINGTON
AND
MONTGOMERY:
EFFECTS
ROUGHNESS
3519
averageof 3) and does not satisfythe criterion for quasiuniformflow.Thereforea depth-slope
productmay onlypartially approximatethe actualreach-average
shearstressat each
10000-
ß dune-tipple
o pool-riffle
ß plane-bed
ß
step-pool
studysite.
•_
cascad•e
1
OF HYDRAULIC
6.
Conclusions
We find that surfacegrain sizesof gravel-bedrivers are
responsive
to hydraulicroughness
causedby bank irregularities,bars,andwood debris.Progressive
increasesin hydraulic
roughness
causesystematic
texturalfining,presumablybecause
of loweredbed stresses
which,in turn, reducechannelcompetenceand diminishbed load transportcapacity,both of which
promotetexturalfining.In hydraulicallyroughforestchannels
the observedreach-average
Dso canbe aslow asone tenth the
competentvalue predictedfrom the total bank-fullboundary
shearstress.This suggests
a corresponding
differencebetween
the bed shearstressandthe total boundaryshearstressfor bed
surfacesthat are in equilibriumwith channelhydraulicsand
highlightsthe importanceof accountingfor hydraulicroughness in complexalluvial channels.Becausemany bed load
Reach-averagebank-full
transport equationsare power functionsof the differencebetweenthe appliedand criticalshearstresses,
smallerrorsin the
bed stresscan causelarge errorsin calculatedbed load trans-
shearstress(%bf)(Pa)
port rates.
10
100
1000
Figure 12. Median bed-surfacegrain sizeversustotal bankfull shearstress,stratifiedby reachmorphology(definitionsof
Montgomeryand Buffington[1997]). Data sourcesare as follows:dune-ripple[Simonsand Albertson,1963; Chitale, 1970;
Higginson
andJohnston;
1988];plane-bedandpool-riffle[Lisle,
1989; Lisle and Madej, 1992; this study];and step-pooland
cascade[Montgomery
andBuffington,1997].Solidline is that of
Figure 3.
Textural fining causedby hydraulicroughnessalso has important implicationsfor the availabilityof salmonidspawning
habitat.Salmonidsselectspecificgrainsizesin whichto spawn
[Kondolfand Wolman,1993].Comparisonof preferredspawning-gravelsizesand our field data suggestthat texturalfining
causedby banks,bars,and wood can createusablespawning
gravelsin channelsotherwisetoo coarseto be hospitablefor
spawning(Figure 3). Furthermore,bar and wood roughness
createa greatervarietyof texturalpatches(seeFigures8a-8c),
offeringa rangeof aquatichabitatsthat maypromotebiologic
diversityor be of useto specificanimalsat differentlife stages.
Macroinvertebratesalso exhibit grain-sizepreferenceswhen
selectingaquatic habitat [Cumminsand Lauff, 1969; Reice,
19801.
Our approachfor predictinga low-roughness
referenceDso
usesa reach-averagedepth-slopeproductto approximatethe
total boundary shear stress.This approximation assumes
steady, uniform flow at reach scales.These conditions are
satisfiedwhen dischargefluctuations occur on timescales
>>u/#S and when studyreach lengthsare >>h/S [Paola and
Mohrig, 1996].Although our studysitesare characterizedby
high velocities(---1 m/s at bank-full stage)and flashyhydro-
graphs,the steepchannelslopes(---10-3-10
-2) makeu/#S
We alsofind that at subreachscalesour studychannelsare
composedof discretetexturalpatchesthat vary with channel
morphologyand roughnessconfiguration.Previous studies
demonstratethat texturalpatchesalso developand evolvein
responseto alteredsedimentload in plane-bedandpool-riffle
channels[Dietrichet al., 1989;Lisle et al., 1993]. Despitethe
commonoccurrenceof textural patchesin both natural [Dietrichand Smith, 1984; Fergusonet al., 1989; Kinerson,1990;
Wolcottand Church, 1990;Lisle and Madej, 1992;Paola and
Seal,1995;PowellandAshworth,1995]andlaboratorychannels
[Iseyaand Ikeda, 1987;Dietrichet al., 1989;Lisleet al., 1993],
little is known about the processesand mechanicsof patch
development,patch interactions,and their role in bed load
transportand channelstability.
The analysisframeworkpresentedhere providesa theoreticalreferencepointfor examiningtexturalresponseto hydraulic roughness.
However,surfacegrainsizealsois responsive
to
bed load sedimentsupply[Buffingtonand Montgomery,this
issue]. Consequently,it may be difficult to assessrelative
causesfor texturalfiningwhen channelshaveboth high sediment loading and significanthydraulicroughness.To isolate
the effectsof sedimentsupply,a methodfor partitioningchannel shearstressis required,suchthat the competentmedian
grain size can be calculatedfrom the bed shearstress(z')
ratherthanthe total boundaryshearstress(zo) [Buffington
and
Montgomery,
thisissue].Moreover,useof surfacegrainsizeto
assessmagnitudesof hydraulicroughnessand bed load sediment supplyrequire channelsto be in quasi-equilibrium.
Surface texturesthat have not had sufficienttime to equilibrate
with channelhydraulicsand imposedsedimentloadsmay not
be goodindicatorsof thesequantities.Nevertheless,
our approachprovidesa physicallybasedframeworkfor examining
controlson bed-surfacegrain size.With the abovecaveatsin
mind our frameworkcanbe usedas a startingpoint for interpreting physicalprocessesand assessing
channel condition
basedon inspectionof bed-surfacegrain size.
quite small(<2 min) and considerably
lessthan the timescale
for significantdischargevariationsduringflood events(rising
limb of hydrographis typically->5 hours [Estepand Beschta,
1985; Smithet al., 1993]). Consequently,a quasi-steady
flow
approximationis valid for our studysites.However,the ratio of
Acknowledgments.Financialsupportwas providedby the Washstudy reach length to h/S rangesfrom 0.4 to 10 (with an ingtonStateTimber,FishandWildlife agreement(TFW-SH10-FY93-
3520
BUFFINGTON
AND
MONTGOMERY:
EFFECTS
004 and FY95-156) and the PacificNorthwestResearchStationof the
USDA Forest Service(cooperativeagreementPNW 94-0617).ITT
RayonierCorporationgraciouslyprovidedaccessto their land on the
Olympic Peninsula.We thank Mike Church, Rob Ferguson,John
Pitlick, Jim Pizzuto, Peter Wilcock, and an anonymousreviewer for
insightfulcriticismsof earlier draftsof thiswork.
References
Andrews,E. D., Bed-material entrainment and hydraulicgeometryof
gravel-bedriversin Colorado,Geol. Soc.Am. Bull., 95, 371-378,
1984.
Bathurst,J. C., Measuringand modelingbedloadtransportin channels
with coarsebed materials, in River Channels:Environmentand Process,edited by K. S. Richards,pp. 272-294, Blackwell,Cambridge,
Mass., 1987.
Best,D. W., H. M. Kelsey,D. K. Hagans,andM. Alpert, Role of fluvial
hillslopeerosionand road constructionin the sedimentbudgetof
Garrett Creek, HumboldtCounty,California,U.S. Geol.Surv.Prof.
Pap., 1454-M, 9 pp., 1995.
Buffington,J. M., The use of streambedtextureto interpret physical
and biologicalconditionsat watershed,reach,and subreachscales,
Ph.D. dissertation,147 pp., Univ. of Wash., Seattle, 1998.
Buffington,J. M., and D. R. Montgomery,A systematicanalysisof
eight decadesof incipientmotion studies,with specialreferenceto
gravel-beddedrivers,WaterResour.Res.,33, 1993-2029,1997.
Buffington,J. M., and D. R. Montgomery,A procedurefor classifying
textural facies in gravel-bedrivers, WaterResour.Res., 35, 19031914, 1999.
Buffington,J. M., and D. R. Montgomery,Effectsof sedimentsupply
on surface textures of gravel-bed rivers, Water Resour.Res., this
issue.
Buffington,J. M., W. E. Dietrich, and J. W. Kirchner,Friction angle
measurementson a naturallyformedgravelstreambed:Implications
for criticalboundaryshearstress,WaterResour.Res.,28, 411-425,
1992.
Carling,P., The conceptof dominantdischargeappliedto two gravelbed streamsin relation to channelstabilitythresholds,Earth Surf.
Processes
Landforms,13, 355-367, 1988.
Carling, P. A., and M. A. Hurley, A time-varyingstochasticmodel of
the frequencyand magnitudeof bed load transporteventsin two
small trout streams,in SedimentTransportin Gravel-BedRivers,
editedby C. R. Thorne,J. C. Bathurst,and R. D. Hey, pp. 897-920,
John Wiley, New York, 1987.
Chien, N., The presentstatusof researchon sedimenttransport,Trans.
Am. Soc. Civ. Eng., 121,833-884, 1956.
Chitale, S. V., River channelpatterns,J. Hydraul.Div. Am. Soc. Civ.
Eng., 96, 201-221, 1970.
Church, M. A., D. G. McLean, and J. F. Wolcott, River bed gravels:
Samplingand analysis,in SedimentTransportin Gravel-BedRivers,
edited by C. R. Thorne, J. C. Bathurst,and R. D. Hey, pp. 43-88,
John Wiley, New York, 1987.
Conover,W. J., PracticalNonparametricStatistics,
462 pp., JohnWiley,
OF HYDRAULIC
ROUGHNESS
Einstein,A., Der hydraulischeoder proill-radius,Schweiz.Bauztg.,103,
89-91, 1934.
Einstein,H. A., Formulasfor the transportationof bed load, Trans.
Am. Soc. Civ. Eng., 107, 561-597, 1942.
Einstein, H. A., and R. B. Banks,Fluid resistanceof compositeroughness,Eos Trans.AGU, 31,603-610, 1950.
Einstein,H. A., and N. L. Barbarossa,River channelroughness,Trans.
Am. Soc. Civ. Eng., 117, 1121-1146, 1952.
Estep,M. A., and R. L. Beschta,Transportof bedloadsedimentand
channelmorphologyof a southeastAlaska stream,For. Serv.Res.
Note PNW-430, 15 pp., U.S. Dep. of Agric., Portland,Oreg., 1985.
Ferguson,R. I., K. L. Prestegaard,and P. J. Ashworth,Influenceof
sandon hydraulicsand graveltransportin a braidedgravelbed river,
Water Resour. Res., 25, 635-643, 1989.
Folk, R. L., Petrologyof Sedimentary
Rocks,182 pp., Hemphill, Austin,
Tex., 1974.
Gehrels,G. E., and H. C. Berg, Geologicmap of southeastern
Alaska,
U.S. Geol. Surv. Misc. Invest. Ser., 1-1867, 1992.
Gehrels, G. E., W. C. McClelland, S. D. Samson, P. J. Patchett, and
J. L. Jackson,Ancient continentalmargin assemblagein the northern Coast Mountains, southeastAlaska and northwest Canada, Geology,18, 208-211, 1990.
Goldfarb, R. J., D. L. Leach, W. J. Pickthorn, and C. J. Paterson,
Origin of lode-golddepositsof the Juneaugold belt, southeastern
Alaska, Geology,16, 440-443, 1988.
Henderson,F. M., Stabilityof alluvial channels,Trans.Am. Soc. Civ.
Eng., 128, 657-720, 1963.
Hey, R. D., Bar form resistancein gravel-bedrivers,J. Hydrol.Eng.,
114, 1498-1508, 1988.
Hey, R. D., andC. R. Thorne,Stablechannelswith mobilegravelbeds,
J. Hydraul.Eng., 112, 671-689, 1986.
Higginson,N. N.J., and H. T. Johnston,Estimationof friction factor
in natural streams,in International Conferenceon River Regime,
editedby W. R. White, pp. 251-266, JohnWiley, New York, 1988.
Houjou, K., Y. Shimizu,and C. Ishii, Calculationof boundaryshear
stressin open channelflow, J. Hydrosci.Hydraul.Eng., 8, 21-37,
1990.
Howard, A.D., Thresholdsin river regimes,in Thresholds
in Geomorphology,editedby D. R. Coatesand J. D. Vitek, pp. 227-258, Allen
and Unwin, Winchester, Mass., 1980.
Iseya,F., and H. Ikeda, Pulsationsin bedloadtransportrates induced
by a longitudinalsedimentsorting:A flume studyusingsand and
gravelmixtures,Geogr.Ann., 69A, 15-27, 1987.
Johnston,C. E., E. D. Andrews, and J. Pitlick, In situ determination of
particle friction angles of fluvial gravels,Water Resour.Res., 34,
2017-2030, 1998.
Kellerhals,R., and D. I. Bray, Samplingproceduresfor coarsefluvial
sediments,J. Hydraul.Div. Am. Soc.Civ.Eng., 97, 1165-1180, 1971.
Kinerson,D., Bed surfaceresponseto sedimentsupply,M.S. thesis,
420 pp., Univ. of Calif., Berkeley, 1990.
Kirchner,J. W., W. E. Dietrich, F. Iseya,and H. Ikeda, The variability
of critical shearstress,friction angle,and grain protrusionin water
worked sediments,Sedimentology,
37, 647-672, 1990.
Kondolf, G. M., and M. G. Wolman, The sizesof salmonidspawning
New York, 1971.
gravels,WaterResour.Res.,29, 2275-2285, 1993.
Cummins,K. W., and G. H. Lauff, The influenceof substrateparticle
size on the microdistributionof streammacrobenthos,
Hydrobiolo- Krumbein, W. C., Application of logarithmic moments to size frequencydistributionsof sediments,J. Sediment.Petrol.,6, 35-47,
gia, 34, 145-181, 1969.
1936.
Dietrich, W. E., Settling velocity of natural particles, Water Resour.
Leighly, J. B., Toward a theory of the morphologicsignificanceof
Res., 18, 1615-1626, 1982.
turbulencein the flow of water in streams,Univ. Calif.Publ. Geogr.,
Dietrich, W. E., and J. D. Smith, Bed load transportin a river meander, Water Resour.Res., 20, 1355-1380, 1984.
Dietrich, W. E., J. D. Smith, and T. Dunne, Flow and sediment trans-
6, 1-22, 1932.
Leopold,L. B., M. G. Wolman, and J.P. Miller, FluvialProcesses
in
Geomorphology,
522 pp., W. H. Freeman,New York, 1964.
port in a sandbeddedmeander,J. Geol.,87, 305-315, 1979.
Dietrich, W. E., J. D. Smith, and T. Dunne, Boundary shear stress, Li, R., D. B. Simons,and M. A. Stevens,Morphologyof cobblestreams
in smallwatersheds,J. Hydraul.Div. Am. Soc. Civ. Eng., 102, 1101sediment transport and bed morphologyin a sand-beddedriver
meanderduringhighand low flow,in RiverMeandering,
Proceedings 1117, 1976.
ofthe Conference
Rivers'83,editedby C. M. Elliot, pp. 632-639, Am. Lisle, T. E., Sedimenttransportand resultingdepositionin spawning
gravels,north coastalCalifornia, WaterResour.Res.,25, 1303-1319,
Soc. of Civ. Eng., New York, 1984.
1989.
Dietrich, W. E., J. W. Kirchner, H. Ikeda, and F. Iseya, Sediment
supplyand the developmentof the coarsesurfacelayer in gravel- Lisle, T. E., Effects of coarsewoody debris and its removal on a
channelaffectedby the 1980 eruptionof Mount St. Helens,Washbedded rivers, Nature, 340, 215-217, 1989.
ington, WaterResour.Res.,31, 1797-1808, 1995.
Dietrich, W. E., M. E. Power, K. O. Sullivan,and D. R. Montgomery,
The useof an analyticalreferencestatein watershedanalysis,Geol. Lisle, T. E., and M. A. Madej, Spatial variation in armouring in a
channelwith high sedimentsupply,in Dynamicsof Gravel-BedRivSoc.Am. Abstr. Programs,28, 62, 1996.
ers, edited by P. Billi et al., pp. 277-293, John Wiley, New York,
du Boys, P., Le Rh6ne et les rivi6rs a lit affouillable,Ann. Ponts
Chauss•es,Mem. Doc., 5, 18, 141-195, 1879.
1992.
BUFFINGTON
AND
MONTGOMERY:
EFFECTS
OF HYDRAULIC
ROUGHNESS
3521
Lisle, T. E., F. Iseya, and H. Ikeda, Responseof a channelwith Reice, S. R., The role of substratum in benthic macroinvertebrate
alternatebars to a decreasein supplyof mixed-sizebed load: A
microdistribution
and litter decomposition
in a woodlandstream,
flume experiment,WaterResour.Res.,29, 3623-3629, 1993.
Ecology,61, 580-590, 1980.
McNeil,W. J., andW. H. Ahnell,Measurement
of gravelcomposition Shields,A., Anwendungder Aehnlichkeitsmechanik
und der Turbuof salmonstreambeds,Circ. 120, 2 pp., Fish. Res. Inst., Coll. of
lenzforschungauf die Geschiebebewegung,
Mitt. Preuss.VersuchFish., Univ. of Wash., Seattle, 1960.
sanst.Wasserbau
Schiffbau,26, 26 pp., 1936.
Meyer-Peter,E., and R. Miiller, Formulasfor bed-loadtransport,in Sidle,R. C., A. J. Pearce,and C. L. O'Loughlin,HillslopeStabilityand
Proceedings
of the 2nd Meetingof the InternationalAssociation
for
Land Use,WaterResour.Monogr.,vol. 11, 140 pp., AGU, Washington, D.C., 1985.
HydraulicStructures
Research,pp. 39-64, Int. Assoc.for Hydraul.
Res., Delft, Netherlands, 1948.
Simons,D. B., and M. L. Albertson,Uniform water conveyance
chanMilhous,R. T., Sedimenttransportin a gravel-bottomed
stream,Ph.D.
nels in alluvialmaterial, Trans.Am. Soc. Civ. Eng., 128, 65-167,
1963.
dissertation,
232 pp., Ore. StateUniv., Corvallis,1973.
Montgomery,D. R., Road surfacedrainage,channel initiation, and Smith, R. D., R. C. Sidle, and P. E. Porter, Effects on bedload transslopeinstability,WaterResour.Res.,30, 1925-1932,1994.
port of experimentalremovalof woodydebrisfrom a forestgravelMontgomery,D. R., andJ. M. Buffington,Channel-reach
morphology
bed stream,Earth Surfi.Processes
Landforms,18, 455-468, 1993.
in mountain drainage basins,Geol. Soc.Am. Bull., 109, 596-611,
Tabor,R. W., andW. M. Cady,Geologicmapof the OlympicPenin1997.
sula,Washington,U.S. Geol. Surv.Misc. Invest.Ser.,1-994, 1978a.
Naot, D., Responseof channelflow to roughnessheterogeneity,J. Tabor,R. W., andW. M. Cady,The structureof the OlympicMounHydraul.Eng., 110, 1568-1587, 1984.
tains, Washington:Analysisof a subductionzone, U.S. Geol. Surv.
Nelson, J. M., and J. D. Smith, Flow in meanderingchannelswith
Prof. Pap., 1033, 38 pp., 1978b.
naturaltopography,
in RiverMeandering,
WaterResour.Monogr.,vol. Trimble, S. W., Streamchannelerosionand changeresultingfrom
12, editedby S. Ikeda and G. Parker,pp. 69-102, AGU, Washingriparian forests,Geology,25, 467-469, 1997.
ton, D.C., 1989.
Wathen,S. J., R. I. Ferguson,T. B. Hoey, andA. Werritty,Unequal
O'Brien,M.P., andB. D. Rindlaub,The transportation
of bed-loadby
mobilityof gravelandsandin weaklybimodalriver sediments,Water
streams,Eos Trans.AGU, 15, 593-603, 1934.
Paola,C., andD. Mohrig,Paleohydraulics
revisited:Palaeoslope
estimation in coarse-grainedbraided rivers, Basin Res., 8, 243-254,
1996.
Resour.Res., 31, 2087-2096, 1995.
Wiberg, P. L., and J. D. Smith, Calculationsof the critical shearstress
for motion of uniform and heterogeneoussediments,WaterResour.
Res., 23, 1471-1480, 1987.
Paola, C., and R. Seal, Grain size patchinessas a causeof selective Wiberg,P. L., andJ. D. Smith,Velocitydistributionandbedroughness
depositionand downstreamfining, Water Resour.Res., 31, 1395in high-gradientstreams,WaterResour.Res.,27, 825-838, 1991.
1407, 1995.
Wolcott,J., and M. Church,Strategiesfor samplingspatiallyheteroParker, G., Self-formedstraightrivers with equilibriumbanks and
geneousphenomena:The example of river gravels,J. Sediment.
Petrol., 61, 534-543, 1990.
mobile bed, 2, The gravelriver, J. Fluid Mech., 89, 127-146, 1978.
Parker,G., Hydraulicgeometryof activegravelrivers,J. Hydraul.Div. Wolman,M. G., A methodof samplingcoarsebedmaterial,Eos Trans.
Am. Soc. Civ. Eng., 105, 1185-1201, 1979.
AGU, 35, 951-956, 1954.
Parker,G., andE. D. Andrews,Sortingof bed load sedimentby flow Wolman,M. G., and J.P. Miller, Magnitudeand frequencyof forces
in meander bends, Water Resour.Res., 21, 1361-1373, 1985.
in geomorphicprocesses,
J. Geol., 68, 54-74, 1960.
Parker, G., and P. C. Klingeman, On why gravel bed streamsare Wood-Smith,R. D., and J. M. Buffington,Multivariategeomorphic
paved, WaterResour.Res.,18, 1409-1423, 1982.
analysisof foreststreams:Implicationsfor assessment
of land use
Parker, G., and A. W. Peterson,Bar resistanceof gravel-bedstreams,
impacton channelcondition,Earth Surfi.Processes
Landforms,21,
377-393, 1996.
J. Hydraul.Div. Am. Soc.Civ. Eng., 106, 1559-1575, 1980.
Pitlick,J., and M. M. Van Streeter,Geomorphology
and endangered
fish habitatsof the upper ColoradoRiver, 1, Historicchangesin
J. M. Buffington,
Water Resources
Division,U.S. Geological
Survey,
streamflow,sedimentload, and channelmorphology,WaterResour. 3215 Marine Street,BuildingRL6, Boulder,CO 80303. (jbuffing@
Res., 34, 287-302, 1998.
usgs.gov)
Powell,D. M., andP. J. Ashworth,Spatialpatternof flowcompetence D. R. Montgomery,Departmentof GeologicalSciences,University
and bed load transportin a dividedgravelbed river, WaterResour. of Washington, Seattle, WA 98195. (dave@bigdirt.geology.
Res., 31, 741-752, 1995.
washington.edu)
Prestegaard,K. L., Bar resistancein gravel bed steamsat bankfull
stage,WaterResour.Res., 19, 473-476, 1983.
Reed, J. C., Southeastern
Alaska,in Landscapes
of Alaska, editedby (ReceivedJuly27, 1998;revisedApril 21, 1999;
H. Williams,pp. 9-18, Univ. of Calif. Press,Berkeley,1958.
acceptedApril 23, 1999.)