forest research technical report - british columbia...forest research vancouver forest region 2100...

Technical Report

Research Disciplines: Ecology ~ Geology ~ Geomorphology ~ Hydrology ~ Pedology ~ Silviculture ~ Wildlife

Forest Research

Vancouver Forest Region2100 Labieux Road, Nanaimo, BC, Canada, V9T 6E9, 250-751-7001

TR-009 Hydrology March 2001

Storm-Based Sediment Budgetsin a Partially Harvested Watershed

in Coastal British Columbia

ByRobert Hudson

Research Hydrologist

Robert Hudson, P.Geo., Ph.D.

Research HydrologistVancouver Forest RegionBC Ministry of Forests2100 Labieux RoadNanaimo, British Columbia V9T 6E9250-751-7001

[email protected]

http://www.for.gov.bc.ca/vancouvr/research/research_index.htm

Cover photo : Russell Creek is a sub-basin of the Tsitika Riverwatershed on northeastern Vancouver Island, British Columbia.

Funding for this extension was provided by Forest Renewal BC—apartnership of forest companies, workers, environmental groups, FirstNations, communities, and government. Forest Renewal BC funding—fromstumpage fees and royalties that forest companies pay for the right toharvest timber on Crown lands—is re-invested in the forests, forestworkers, and forest communities. Funding assistance by Forest Renewal BCdoes not imply endorsement of any statements or information contained herein.

Technical Report TR-009 March 2001 Research Section, Vancouver Forest Region, BCMOF


CONTENTS

ABSTRACT ......................................................................................................................................................................................... 2

KEYWORDS ....................................................................................................................................................................................... 2

ACKNOWLEDGMENTS ................................................................................................................................................................... 2

INTRODUCTION ............................................................................................................................................................................... 3

BACKGROUND .................................................................................................................................................................................. 3

Sediment-Budget Research in the Tsitika River Watershed ..................................................................................................... 4

STUDY AREA ..................................................................................................................................................................................... 5

Physical Characteristics ............................................................................................................................................................... 5Automatic Gauging Sites ............................................................................................................................................................. 5Sediment Source Inventory and the Selection of Sampling Sites ............................................................................................. 6

METHODS ........................................................................................................................................................................................... 7

Instrumentation ............................................................................................................................................................................ 7Data Interpretation ...................................................................................................................................................................... 7

RESULTS ............................................................................................................................................................................................. 9

Sediment-Production Characteristics of Selected Sources in Russell Creek ........................................................................ 10Road-Crossing Model ................................................................................................................................................................ 15Gully Model ................................................................................................................................................................................ 19Landslide Model ......................................................................................................................................................................... 21Application of the Sediment-Budget Model ............................................................................................................................. 22

DISCUSSION ..................................................................................................................................................................................... 24

Forest Management Implications ............................................................................................................................................. 28

CONCLUSIONS ................................................................................................................................................................................ 30

REFERENCES .................................................................................................................................................................................. 30

APPENDICES

Appendix A: Sediment Source Inventory ................................................................................................................................. 33Appendix B: Flow Rating and Verification .............................................................................................................................. 39

TABLES

Table 1. Morphological factors of Russell Creek and sub-basins .................................................................................................. 6Table 2a. Attributes of road-crossing sites selected for monitoring ............................................................................................... 8Table 2b. Attributes of gullies selected for monitoring .................................................................................................................. 8Table 3. Characteristics of storms for which sediment budgets are calculated ............................................................................ 10Table 4. Measured storm sediment yields at gauging sites and sediment source sites ................................................................. 11Table 5. Measured and calculated crossing yields ........................................................................................................................ 18Table 6. Gully yield data set for modeling ................................................................................................................................... 20Table 7. Table of measured landslide yields and storm rainfall characteristics ............................................................................ 21Table 8. Mean grain-size distribution of sediment sources, by type............................................................................................. 22Table 9a. Calculated proportion of sand in sediment yield for selected storms at Stephanie and Russell Creeks ....................... 23Table 9b. Calculation of proportion of sediment production transported by gullies across the valley flat .................................. 23Table 10. Summary of sediment-budget results for four storms ..............................................................................................24-26Table 11a. Summary of relative contributions of harvested sources & roads to the overall sediment yield of Russell Creek ..... 27Table 11b. As above, but excluding gully G76 and landslide LS191 ........................................................................................... 27Table 11c. Summary of contributions from Confluence gullies to the sediment yield of Russell Creek at Russell Main......... 27Table 12. Potential effects of several deactivation/water management scenarios ........................................................................ 30

1



FIGURES

Figure 1. Map of Russell Creek ...................................................................................................................................................... 5Figure 2. Russell Main channel ...................................................................................................................................................... 6Figure 3. Stephanie Creek .............................................................................................................................................................. 6Figure 4. The Confluence ............................................................................................................................................................... 7Figure 5. Comparison of normal regime SSC vs turbidity relationships at Russell Main, Stephanie, and the Confluence............9Figure 6. Streamflow and SSC, Russell Creek, 12 November 1998 ............................................................................................ 12Figure 7a. Streamflow and SSC at Site 2 above and below the crossing ..................................................................................... 12Figure 7b. Gully G61 .................................................................................................................................................................... 12Figure 8a. Streamflow and SSC at Site 12 above and below the crossing ................................................................................... 13Figure 8b. Gully G65 looking upstream from Site 12 .................................................................................................................. 13Figure 9a. Russell Confluence, north slope, as of summer 1994, prior to debris flow in gully G74............................................ 13Figure 9b. Russell Confluence, north slope, as of spring 2000 .................................................................................................... 14Figure 10. Streamflow and SSC at Site 27 above and below the crossing ................................................................................... 14Figure 11. Streamflow and SSC at Site 34 above and below the crossing ................................................................................... 14Figure 12a. Streamflow and SSC in gullies G75 and G76 ........................................................................................................... 15Figure 12b. Gully G76 looking upstream from Site 46 ................................................................................................................ 15Figure 13a. Site 206, the landslide at upper Stephanie monitored for sediment yield ................................................................. 16Figure 13b. Streamflow and SSC at upper Stephanie above and below Site 206, 12 November ................................................ 16Figure 14a. Ditch above Site 45 after the washout of November 1995 and prior to road deactivation ....................................... 16Figure 14b. Streamflow in gully G75, and SSC above and below Site 45 in its post-washout state, April 1996 storm .............. 16Figure 15. Sediment yield at road crossings vs area of cut-and-fill slopes for selected storms ................................................... 17Figure 16. Measured vs calculated sediment yield from crossings .............................................................................................. 19Figure 17. A model of sediment yield under “normal” conditions from ditches & from running surfaces of mainline roads ..... 19Figure 18. Gully yield as a function of storm rainfall ................................................................................................................... 20Figure 19. The gully yield model consists of a method of predicting the equation coefficients of the curves in Figure 18 ........ 20Figure 20. Landslide yield per unit area, as a function of the maximum 24-h storm rainfall intensity ........................................ 21Figure 21. Relationships between storm yields at Russell Main, the Confluence, and upper Russell .......................................... 22Figure 22. Determination of Equation 11 ..................................................................................................................................... 24Figure 23. Landslide LS 199, Stephanie Creek, produces more sediment than any other source in Russell Creek ..................... 26Figure 24. Contribution of road-related sources to the sediment budget of Russell Creek............................................................ 28Figure 25. Evidence of rill erosion on the fill slope at Site 2 ....................................................................................................... 29Figure 26. Site 45 after road deactivation .................................................................................................................................... 29

ABSTRACTA program of sediment-budget research commenced in 1994 atRussell Creek, a sub-basin of the Tsitika River watershed on North-eastern Vancouver Island, British Columbia. Part of the programaimed to determine the relative contribution of different types ofsediment sources to the sediment load of mainstem channel sitesin Russell Creek and its main tributary, Stephanie Creek, which aretwo typical Coastal British Columbia streams.

The report describes a sediment-budget model that predicts thesediment yield from gullies, forestry-road-related sources, and land-slides based on attributes of the sediment sources and storm rain-fall characteristics. The sediment-budget model will help forestmanagers assess the effects of forestry activities on sediment pro-duction in streams, and the related implications for fish habitat.The model will also be useful to forestry-road-deactivation plan-ners in predicting the effectiveness of road-deactivation sce-narios.

KEYWORDSforestry, forest management, hydrology, streams, sediment, sedimentbudget, sediment production, sediment yield, fish habitat, forestry roads,road deactivation, Vancouver Forest Region, British Columbia

ACKNOWLEDGEMENTSThis research was funded by Forest Renewal BC, and through basefunding from the Operational Division of the BCMOF. RyanHanson and John Fraser assisted with the data collection and sam-pling. Karen Paulig did the sediment source survey work and helpedwith instrumentation. John Tyler assisted with instrumentation anddata collection. The Tsitika River Sediment Monitoring Programwas originally implemented by Dan Hogan and Bruno Tassone. EdMayert, David Paul, and Scott Ferguson (Water Survey of Canada)contributed to the hydrometric work. Thanks are extended toRowland Atkins, Tom Millard, Rob Millar, and Warren Cooper forreview comments. Editing and layout was done by Kathi Hagan.

2



INTRODUCTION

Sediment production in creeks and rivers in Coastal British Co-lumbia is a natural process, and the subject of ongoing researchin the Vancouver Forest Region and elsewhere in the province.One of the most important issues related to this research is theeffects of forest harvesting and related activities on the sedi-ment budgets of creeks.

A program of sediment-budget research commenced in 1994at Russell Creek, a sub-basin of the Tsitika River watershed onNortheastern Vancouver Island, British Columbia. Part of theprogram aimed to determine the relative contribution of dif-ferent types of sediment sources to the sediment load ofmainstem channel sites in Russell Creek and its main tributary,Stephanie Creek, which are typical Coastal British Columbiastreams. This report describes a sediment-budget model thatpredicts the sediment yield from gullies, forestry-road-relatedsources, and landslides based on attributes of the sedimentsources and storm rainfall characteristics. The sediment-bud-get model will help forest managers assess the effects of for-estry activities on sediment production in streams, and the re-lated implications for fish habitat. The model will also be use-ful to forestry-road-deactivation planners in predicting the ef-fectiveness of road-deactivation scenarios.

BACKGROUND

Interest in sediment-production research began in the 1960sbecause of concern for the potential effects of increased sedi-mentation in forest streams on fish habitat and health.

Research on the effects of road construction and harvestingand silvicultural practices on suspended sediment loads instreams was undertaken in watersheds with unstable soils inthe Oregon Cascades, and on more stable granitic soils in Idaho.Harvesting without roads or without slash burning had mini-mal effects on suspended sediment concentration (SSC). How-ever, slash burning relatively immediately after harvesting re-sulted in large increases in SSC in the short-term; these increasesdiminished over time and as the harvested sites became re-veg-etated (Frederiksen 1970; Brown and Krygier 1971). Similarresults were obtained in the Idaho study, except that SSC levelsbefore harvesting were less than in the Oregon studies and theeffects of the burning on SSC were less severe (Megahan et al.1995). Similar large increases in SSC were noted immediatelyafter road construction; peak SSC increased up to 250 timesimmediately after harvesting in the Oregon study, but in theIdaho study, watershed sediment yield increased 5 times duringthe road-construction phase (Megahan et al. 1986). While theeffects of road construction on sediment yield are fairly imme-diate and rather severe, they subside after construction is com-plete; nevertheless, SSC levels in roaded watersheds remain atelevated levels due to a combination of road-surface erosionand road-related landslides (Brown and Krygier 1971; Reid andDunne 1984; Beschta 1978).

From the above, it is clear that watershed lithology plays a big

role in how forestry activities affect SSC levels and sedimentyield. In an earlier study of storm-based sediment yield in theTsitika River watershed, Hudson and Sterling (1998) found thatCatherine Creek yielded about 5.5 times as much sediment asRussell Creek over the same time period. Catherine Creek wasundeveloped at the time, and 25% of Russell Creek was roadedand harvested. The difference was attributed to the dominantbedrock lithology. Basaltic bedrock areas in the Tsitika Riverwatershed produced over 6 times more debris torrents per unitarea than did the granitic bedrock areas (Sterling 1997).

In the absence of slash burning, forest roads remain the pri-mary harvesting-related source of fine sediments in streams.Efforts have been made to quantify the relative importance ofroad components. Reid et al. (1981) found that 60% of road-related sediment production was derived from landslides, and20% was derived from running surfaces. Sediment yield fromroad surfaces is related to road use, with active haul roads yield-ing one and two orders of magnitude more sediment than mod-erately used and abandoned roads, respectively (Reid and Dunne1984). Nistor and Rood (1999) found that sediment yield fromroads was related to road gradient, with over 90% of road-re-lated sediment produced by the running surface.

The effects of road-related sediment yield on the sedimentbudget of fish-bearing streams has also been studied, becausemost sediment produced by roads enters small ephemeralstreams. Silt and clay-sized fractions were transported efficientlyeven at low flows, whereas sand-sized and coarser particles werestored in ephemeral stream channels such that up to 45% ofthe sediment was actually delivered to the receiving stream(Duncan et al. 1987). Distance of sediment travel in headwaterchannels was related to attributes of the channels, includingcatchment area, channel gradient, the presence of log jams, areaof sediment sources and length of roads draining into the site(Megahan and Ketcheson 1996).

Earlier sediment-budget studies relied on developing relation-ships between SSC of point samples and stream discharge inorder to calculate SSC and sediment yield during storms. Sev-eral authors have alluded to the difficulty of this approach, be-cause different relationships exist for rising and falling stages,and from storm to storm (Brown and Krygier 1971; Beschta1978). Other investigators have used bulk methods�such assettling basins (e.g., Megahan et al. 1995)�of determining sedi-ment yield; the drawbacks are that long time periods are re-quired to conduct the measurement, and that channel cannotbe assessed because it is cut off from the source.

As early as the 1970s, in-stream turbidity meters were tested toovercome the problems associated with the hysteretic behaviourof suspended sediment in relation to streamflow (Walling andWebb 1982). The advent of optical backscatter (OBS) technol-ogy (Downing 1983; Downing and Beach 1989; Ludwig andHanes 1990) has since resulted in much greater precision inmonitoring SSC for sediment budget studies (Hudson and Ster-ling 1998; Jordan and Commandeur 1998; Hudson 2001).

3



Sediment-Budget Research in theTsitika River Watershed

In an effort to understand sediment-production processes, theTsitika River Sediment Monitoring Program was initiated in1991. This program as originally implemented consisted of twocomponents:

1. Continuous monitoring of streamflow, turbidity, suspended-sediment concentration, and grain-size distribution at three sites:the mainstem of Tsitika River, and at the mouths of Russelland Catherine Creeks, two sub-basins of the Tsitika River.

2. Manual measurement of discharge and sampling of waterfor the determination of suspended-sediment concentrationat several pre-selected sites during storms; these sites weresupposed to represent the individual sediment sources thatwould supply sediment to the main creeks, and were dis-persed throughout the Tsitika River watershed.

The purpose of having the two components was to determinethe relative contributions of different types of sediment sourcesto the overall sediment production in the main creeks duringstorms. Storm sampling was carried out for three years in con-junction with automatic monitoring. The early stage of the pro-gram met with some success, but it became apparent that thescale of the project was too large. Because the Tsitika Riverwatershed is 370 km2 in area, it was not possible to sample insufficient detail, either temporally or spatially, to construct areproducible sediment budget at that scale. It was concludedthat a sub-basin of the Tsitika River would be a more appropri-ate scale for this type of research. Russell Creek was selectedfor sediment-budget investigations for the following reasons:

1. It was already gauged.

2. It has a history of harvesting, as well as ongoing forestryactivities, including active harvesting and a mainline haul road.

3. It contains both granitic and basaltic bedrock types, as well astributary creeks with contrasting watershed morphologies thatwere thought to influence sediment-production processes.

Nistor (1996) studied fine sediment transfer in gullies in RussellCreek. The study focussed on fluvial sediment transfer as op-posed to episodic sediment production due to debris flows and/or landslides. Thus, Nistor�s results are highly relevant to thecurrent study. His findings include the following:

� Sediment transfer in gullies occurred almost entirely dur-ing large rainfall and snowmelt events.

� Sediment was entrained from sediment sources on the chan-nel margins above a flow threshold that is characteristic toeach gully.

� Fluvial sediment transfer in gullies occurred as a series ofdiscrete sediment pulses, as opposed to a steady backgroundrate. These pulses occurred on the rising limb of the stormhydrographs, and were brief compared to the duration ofthe runoff event. Thus the normal fluvial sediment produc-tion is in itself episodic in nature.

Based on these results, Nistor recommends that for the deter-mination of sediment budgets, high frequency sampling, be-ginning at the onset of runoff events, is essential.

The sediment-budget research program in Russell Creek as itcurrently exists is based on five automatic stream-monitoringsites where stage and turbidity are measured continuously, andten representative sediment source sites where manual samplingand flow measurements are carried out during storms. The pur-pose of this paper is to report on the results of sediment-bud-get investigations during storms, based on an analysis of sedi-ment yields at the manual and automatic sites. The sedimentbudget includes an analysis of the relative importance of natu-ral and man-made sediment sources. Sediment yield is definedas the total mass of sediment carried past a point by a streamover specified time interval. This study focuses on the suspendedsediment yield, which is the total mass of sediment carried inthe water column, as opposed to bed yield, which is transportedalong the channel bed by rolling or saltation.

In principle, the sediment-budget approach is an accountingprocedure in which sediment yield during storms is measuredin one or more mainstem stream channels, and also at selectedsediment source sites. The mainstem sites are monitored auto-matically, whereas the individual sources are monitored manu-ally.

The sediment source sites that are selected for monitoring rep-resent specific types of sediment sources, such as gullies, roadsources, and landslides. The most significant elements of for-est roads are those that are directly connected to the streamchannel network. These road elements are associated withstream crossings. The sediment yield from connected road ele-ments can be measured by collecting water samples in gulliesabove and below the road crossings. The sediment yield de-rived from the crossings can be separated from the yield de-rived from the gullies, by subtracting the sediment yield abovethe crossings from that below.

This sampling strategy directly measures the sediment yield fromthe crossings but does not provide a breakdown of the yieldaccording to the components that make up the crossing; theseinclude the cut slopes, fill slopes, ditches, and running surface.

The measured source yields are used to develop relationshipsbetween sediment-production characteristics and measured at-tributes of the sediment sources. These relationships are thenused to reconstruct the sediment budget during the storm; thesource yields are added up and the result is compared to themeasured sediment yield at the mainstem stream that drainsthe watershed area that contains those sediment sources. If thetotal yield from sediment sources is close to the measured streamyield, then the relative contributions of the different types ofsediment sources to the sediment budget can be determined.

The sediment-budget approach has been successfully appliedto four storms of varying magnitudes in Russell Creek. Thepurpose of this paper is report on the results of those investi-gations.

4



STUDY AREA

Physical Characteristics

Russell Creek (Figure 1) drains an area of 30.9 km2. A largebatholith of granitic Island Intrusions underlies the lower two-thirds of the watershed. The upper one-third of the watershedis underlain by the extrusive or basaltic Karmutsen Formation.These basaltic rocks are typically weak and easily erodible, andthe igneous rocks are typically hard and resistant. There is aband across the middle of the watershed where surficial mate-rial derived from the Karmutsen formation is draped over theigneous bedrock. The lower watershed is relatively gently sloped,the basin is bowl shaped and has two bowl-shaped, third-ordertributaries. There is a broad valley flat that buffers the middlereaches of the channel from coarse sediments generated bydebris flows and landslides. Thus, sediment sources in the middlepart of the watershed are not directly connected to the main

channel. In contrast, the upper part of the watershed that isunderlain by the Karmutsen formation tends to be steep withexposed basaltic bluffs and incised stream channels. In theKarmutsen areas the sediment sources tend to be directly con-nected to the channel system.

Automatic Gauging Sites

Morphological characteristics of Russell Creek and its sub-ba-sins are given in Table 1. The watershed is divided into sub-basins by the automatic stream gauge sites (Figure 1). Streamgauges are located at Russell Creek mainstem at the bridge (Fig-ure 2), Stephanie Creek at the bridge (Figure 3), upper StephanieCreek, Russell Creek above Stephanie confluence (�theConfluence�, Figure 4), and upper Russell at TS120. This net-work of gauging sites creates a nested set of sub-basins of de-creasing size. Sediment-budget investigations are carried outwithin areas as defined by the gauging sites.

Figure 1. Map of Russell Creek: for readability, only the features discussed in the text are plotted on the map.

5



Sediment Source Inventory and theSelection of Sampling Sites

Maynard (1991) carried out a complete inventory of sedimentsources in Russell Creek and elsewhere in the Tsitika RiverWatershed. In the summer of 1996, the inventory was updatedand ground-truthed (Paulig 1996). All sediment sources inRussell Creek were documented and described by a set of quali-tative and quantitative attributes. Qualitative attributes consistedprimarily of the type, erosional activity and connectivity of thefeature, as well as the geological formation and whether or notthe feature was harvested. The main features of interest werethe ones that could potentially be directly connected to the chan-nel system, including landslides, gullies, road crossings, and erod-ing ditches. A sediment source feature was considered signifi-cant if it was actively eroding and connected to a gully, tribu-tary, or mainstem channel. Quantitative attributes included areaof exposed sediment sources in the case of landslides and roadcrossings; channel gradients, gully catchment areas; lengths oferoding ditch; and some semi-quantitative descriptors of gully

Table 1. Morphological factors of Russell Creek and sub-basins.

srotcaF einahpetSevoballessuRecneulfnoCeht

llessuRniaM

mk(aerA 2) 16.8 30.61 88.03

)m(egnar.velE 0861-074 5171-024 5171-562

eganiarDytisned

82.2 62.2 32.2

noitacrufriBoitar

85.4 14.3 88.2

lennahcnaeMrewol,tneidarg

evobalennahc)%(eguag

0.51 1.3 0.5

eliforplennahC)%(eguagta

5.8 6.3 1.3

epytlennahC ,loop-edacsaCwolebdesicni

reppueinahpetS

levargloop-elffiRybdereffubdeb

talfyellav

loop-elffiR,deblevarg

rewolsehcaerdesicni

epolsdnal.gvA 35.0 25.0 74.0

noitavele-aerAevruc

xevnoC evacnoC evacnoC

rotcafmroF 65.0 43.0 55.0 Figure 2. Russell Main channel.

Figure 3. Stephanie Creek.

6



stability, degree of revegetation, etc. The significant sedimentsources are described in Appendix A (Table A1). The sourcesare organized by type and by the sub-basin in which they fall.To conserve space, only the directly connected and active sedi-ment sources are listed.

A subset of the sediment sources was selected for manual sam-pling. The purpose of the manual sampling was to collect dataupon which to build the sediment-budget model. Therefore,the sample sites should be representative of sediment produc-tion in Russell Creek. To achieve this, cluster analysis was doneon the sediment inventory attributes (Paulig 1996; Hudson 1999).This analysis was done on gullies, road crossings, and ditch sitesseparately. Representative crossings and ditches were paired withrepresentative gullies. The cluster analysis resulted in the selec-tion of ten manual sampling sites (in Figure 1, Site 48 is notshown because it is very close to the upper Stephanie gauge).The relevant attributes of those sites are summarized in Table2. The manual sites were grouped into two areas within RussellCreek, so the sampling could be done efficiently during stormsby two operators, each following a continuous loop.

METHODS

Instrumentation

The Russell Main stream gauging site was installed in Novem-ber 1991, and operated by Water Survey of Canada. In 1996 theoriginal equipment was upgraded to fully digital operation. Stagewas measured and recorded using a nitrogen bubbling systemconnected to a Water-log H-350 transducer, and recorded by ahigh-resolution Unidata 7000 series �Macro� data logger. Dis-charge was measured using the standard current metering tech-nique, and rating curves were established to convert the mea-sured stages into streamflow. Turbidity was measured using aD&A Instruments optical backscatter (OBS) probe calibratedin the range of 0-500 NTU and monitored by the data logger.An ISCO 3700C automatic sampler was used to collect watersamples. The turbidity probe was installed in an ABS housing,which was installed on a steel mounting plate along with thesampler intake tube such that the sensor and intake were at aheight of 20 cm above the base of the plate. In this way, theOBS probe �sees� the same water volume from which thesample is withdrawn. This assembly was placed in a pool in thestream channel. The sampler, data logger, stage recorder, powersupplies, etc. were housed in a shelter on the stream bank. Iden-tical instrumentation was used at the other sites, except thatstage was measured with a Keller 210-S submersible transducer,and discharge was measured with a combination of currentmetering and salt dilution gauging. The data loggers were pro-grammed to monitor mean stage, and minimum, mean, andmaximum turbidity on a 5-min log interval. At upper StephanieCreek, two ISCO samplers were used simultaneously above andbelow the landslide (Site 206) to characterize the sediment pro-duction from a landslide connected to the channel. The watersamples were collected from the ISCO samplers regularly andanalyzed for sediment concentration.

At the manual measurement sites, stream water was collectedmanually with a DH-48 sampler. Staff gauges were installedand were read each time samples were collected. Stream dis-charge at these sites was measured using salt dilution, and rat-ing curves were developed. In the summer of 1998, additionalautomatic monitoring sites, consisting of stage and rainfall only,were installed at Sites 3 and 27 (Figure 1). These sites were usedto derive a continuous record of streamflow at the other manualsites for storms that occurred after that time, based on themanual readings. Rainfall and total precipitation were measuredat sites near Russell Main, upper Russell, and upper Stephanie.Rainfall was measured with tipping-bucket rain gauges, and to-tal precipitation was measured with 16-inch (40-cm diameter)PVC standpipe gauges, monitored continuously.

Data Interpretation

The process of sediment production in high-energy coastalstreams is extremely complex. Consequently, interpretation ofsuspended sediment and turbidity data is also very complex,and has been described in detail by Hudson (2001). Relation-ships between turbidity and suspended sediment depend on

Figure 4. The Confluence.

7



Table 2a. Attributes of road-crossing sites selected for monitoring.

Table 2b. Attributes of gullies selected for monitoring.

ecruoS.on

gnissorCepyt

-buSnisab a

detaicossAyllug

ylluGtneidarg

tagnissorc

)%(

fohtgneLdetcennoc

daor)m(

fohtgneLdetcennoc

hctid)m(

ytilibatSfo

gnissorcS b

saeraecruostnemideS

etiS.vele)m(

gnissorCepolsllif

m( 2)

gnissorCepolstuc

m( 2)

hctiDtucepols

m( 2)

2 gol NC 16G 03 063 531 4 492 003 054 066

5 epip NC 26G 51 002 05 1 0 23 0 005

72 gol SC 47G 03 08 02 3 53 69 04 065

43 epip SC 181SL 05 005 041 2 51 0 024 006

01 epip RU 46G 72 58 55 3 465 0 23 045

21 gol RU 56G 51 57 01 1 3 02 0 015

54 gol S 57G 52 021 021 4 192 0 069 067

64 tuohsaw S 67G 52 001 001 5 0 0 008 067

74 gol S 67G 52 004 002 2 61 84 006 067a .einahpetS=S.021STevoba,llessuRreppU=RU.epolshtuos,ecneulfnoC=SC.epolshtron,ecneulfnoC=NC b erusaemevitatilauqasiS

elbatsnutsom=5dna)noisorefoecnediveon,tneidargwol(elbatstsom=1erehw,sepolsllif-dna-tucgnissorcehtfoytilibatsllarevoehtfo.gnissorcehtdnuoradnani,noisorellirgnidulcni,noisorefoseergedevah4ot1seirogetaC.)gnissorcehtfoeruliaflatotfoecnedive(

ylluG.on

daoRsgnissorc

-buSnisab

deBytilibats

1S a

knaBytilibats

2S a

detnerroTsv

detnerrotnuT a htgneL

)m(

latoTaera

mk( 2)

deguaGaera

mk( 2)tneidarG

)%(

cigoloeGnoitamrof

G b

svdetsevraHdetsevrahnu

H c

16)reppu(

4,3,2 NC 1 0 0 0021 485.0 235.0 75 1 1

16)rewol(

NC 1 1 1 008 02 1 1

26 6,5 NC 0 0 0 0041 062.0 632.0 74 1 1

46 01,9 NC 0 0 0 0002 087.0 467.0 35 1 1

56 21,11 NC 0 1 0 0071 420.1 410.1 12 1 1

47 ,72,6203,92

SC 1 1 1 0001 825.0 612.0 54 0 2

57 54 S 0 0 0 0052 003.1 272.1 52 0 0

67 74,64 S 1 1 1 0022 959.0 049.0 53 1 0a .elbatsnu=1dnaelbats=0erehw,ytilibatsyllugebircsedotselbairavyrogetaceraTdna,2S,1SselbairavehT b citlasab=1dnacitinarg=0

.kcordebcitinargrevodepardlairetamlaicifruscitlasabro c .detsevrahyletelpmoc=2.detsevrahyllaitrap=1.detsevrahnu=0

8



grain-size distribution. The grain-size distribution of the sus-pended sediment in streams depends on sediment supply andhydrologic conditions. Consequently at any given site there ismore than one relationship between turbidity and suspendedsediment. This results in various suspended-sediment regimes.

A sediment-production regime is defined as a specific range ofconditions that controls the relationship between SSC and tur-bidity at a point on a stream. The relationship that is usually ineffect is called the normal regime (Figure 5). The relationshipsare curves because, as in-stream flow conditions and turbiditychange, the grain-size distribution of the suspended sedimentalso changes. At low turbidity the normal regime consists en-tirely of fines (i.e., silt and clay). At higher flows (and hence, athigher turbidity) the stream is capable of carrying sand as wellas fines, and the proportion of sand in the suspended load in-creases with increasing turbidity. During a large storm, sedi-ment production will switch to a coarser regime. At Russell Creekthis transition occurs in response to a specific rainfall intensitythreshold. Under the coarse regime, the stream carries either acoarser grade or a higher proportion of sand for a given turbid-ity, resulting in a steeper SSC vs turbidity curve. After largestorms, the sediment production switches to a finer regime thatconsists entirely of fines. This transition occurs because thesupply of transportable sand is more limited than the supply offines, and large storms have a tendency to flush most of theavailable sand from the watershed. Subsequent to this, the finerregime will remain in effect until erosional processes have madeavailable a fresh supply of transportable sand, and there arestorms capable of transporting it.

At Russell and Stephanie Creeks these regime transitions canbe predicted from rainfall intensity and flow conditions. SSC vsturbidity relationships can be used to determine the relativeproportion of sand and fines in the suspended load (Hudson2001). These methods are integral to the calculation of SSCand sediment yield at the mainstem gauging sites.

For each storm, suspended sediment yield was calculated foreach site that was monitored. Sediment yield is calculated as theproduct of the suspended sediment concentration (SSC) andstreamflow or discharge (Q), as follows:

1000

..)(

ILQSSCkgYield

××= (1)

where

SSC is in mg/L,Q is in m3/s, andL.I. is the log interval in seconds.

This gives the total mass of suspended sediment transportedover a single log interval. In most cases, the interval is 5 min-utes. Summing the 5-min yields over the duration of a stormgives the total storm sediment yield.

For the automatic gauging sites, SSC was calculated using thedetailed methods described by Hudson (2001). For the manu-ally sampled sediment source sites, streamflow and SSChydrographs had to be reconstructed from point data. Thestreamflow records were reconstructed by developing relation-ships between the staff gauge readings and continuous stagerecords from a nearby reference site (either a mainstem site, orSites 3 or 27). The reference site was chosen based on maximumcorrelation among the sites for which data were available, recog-nizing that there may be different relationships for rising andfalling stage. This synthesized stage data were then convertedto discharge using the appropriate rating curve. As an indepen-dent check on the synthesized flow data, flow volume ratios wereplotted against catchment area ratios between the manual sites andthe reference sites on which the synthesis was based. An ex-ample of the development of a storm hydrograph from pointdata and a summary of volume ratios appears in Appendix B.

The SSC data for the manual sites were reconstructed usingmethods similar to the reconstruction of the flow data. In thiscase, for a given site, relationships were developed between theSSC derived from sample data and either streamflow at the samesite, or turbidity at a reference site, whichever produced thebetter results. Again, different relationships were developed forrising and falling SSC.

RESULTS

The grain-size distribution of suspended sediment in streamflowat mainstem sites is highly site-specific. A comparison of nor-mal regime SSC vs turbidity relationships at Russell Main, theConfluence, and Stephanie Creek demonstrates the effects ofchannel characteristics on grain-size distribution (Figure 5).

Figure 5. Comparison of normal regime SSC vs turbidityrelationships at Russell Main, Stephanie, and theConfluence.

0 40 80 120M ean T urb id ity (N TU )

0

100

200

300

400

500

Nor

ma

l Re

gim

e S

SC

(m

g/L

)

N o rm a l R egim eStephanie

R ussellC on fluence

Lab C a librationsF inesVery F ine S and

9



Factors that affect grain-size distribution of the suspended sedi-ment load include channel gradient and connectivity of sedi-ment sources. A key feature of the graph is the slopes of thefield calibrations relative to the lab calibration lines. The slopeof the normal regime curve for the Confluence is less than thatof the lab calibration for fines throughout its range. This indi-cates that the normal regime�s suspended sediment load at theConfluence is composed entirely of silt and clay. This is partlydue to the channel gradient, and also due to the fact that thechannel above the Confluence is buffered by a long section ofvalley flat. The reduced gradient encountered by the gullies asthey reach the valley flat causes the sand fraction carried insuspension to be deposited before reaching the main channelof Russell Creek. In contrast, the normal regime�s sedimenttransport at Russell and Stephanie Creeks includes increasingsand content as turbidity rises. At Stephanie Creek, the normalregime includes a proportion of sand that varies from 0% at 10NTU turbidity to 45% at 100 NTU; at Russell Creek, the pro-portion of sand varies from 0% at 40 NTU turbidity to 37% at100 NTU, peaking at 70% at 800 NTU (Hudson 2001). Thelower reaches of Russell Creek are incised, and are affected bya series of small sidewall failures. Russell Creek is also affectedby inflows from Stephanie Creek, which has a higher channelgradient and contains sediment sources with a high degree ofconnectivity. These facts all have a significant bearing on sedi-ment-budget accounting. The fact that the automatic gaugingsites break up the system into smaller measurement reaches is veryuseful for calculating the sediment budget within each section.

It proved very difficult to obtain complete data upon which tobase a sediment budget. One cause of the difficulty was thetiming of events. Even the largest sediment-production eventsat Russell Creek have a time base of less than 24 h (Hudsonand Sterling 1998). Other factors included difficulty of access

due to snow, equipment malfunction, and inability to predictlarge events with any accuracy. In addition, many events occurover night, making it impractical and dangerous to do the sedi-ment source sampling. However, to date, complete informa-tion has been collected for two large events on 01 October1997 and 12 November 1998, as well as smaller events on 24January 1998, 14 November 1998 and 20 November 1998. Inaddition to this, Gullies G75 and G76 were monitored auto-matically at Sites 45 and 46 during a storm in April 1996 thatprovided important information on sediment production dur-ing extreme conditions. Meteorological characteristics of thestorms are summarized in Table 3.

Sediment-Production Characteristics of SelectedSources in Russell Creek

The suspended-sediment yields measured during the storms atmainstem and sediment source sites are summarized in Table 4.

The storm that occurred on 12�13 November 1998 resulted ina peak flow of 17.1 m3/s on Russell Creek, with a peak SSC ofover 50 mg/L (Figure 6). SSC derived from turbidity readings isin very close agreement with observed SSC. This is not a largeevent�the peak flow has been exceeded an average of 4.5times/y since monitoring began in the fall of 1991, giving it areturn period of <0.25 y. It therefore represents conditions thatoccur frequently at Russell Creek.

At Site 2, on the north slope of the Confluence, a large pulseof sediment was generated from the road crossing, whereasSSC levels generated by the gully above the crossing were rela-tively low (Figure 7a). Site 2 is the upper crossing of gully G61,of which the lower part is torrented (Figure 7b). The gully aboveSite 2 has an unstable bed, but is scoured of fine sediment (i.e.,sediment that is capable of being transported in suspension).

Table 3. Characteristics of storms for which sediment budgets are calculated.

tratsmrotSetad

(eguagniarniaMllessuR .vele )m572 (etislacigoloroetem.rCeinahpetS .vele )m038

stnemmoC

mrotSllafniar

Rs

)mm(

mrotsmumixaMseitisnetni

mrotSllafniar

Rs

)mm(

yliaDwons)mm(

mrotsmumixaMseitisnetni

ruoh-21R 21

)h21/mm(

ruoh-42R 42

)h42/mm(

ruoh-21R 21

)h21/mm(

ruoh-42R 42

)h42/mm(

40 rpA 69 831 06 711 291 3.45 45 601 niar,wonSwonsno

79tcO10 44 72 34 45 5.8 83 55

89voN21 86 45 86 85 0 85 95

89voN41 75 51 72 95 0 51 12 ,wonsnoniaRtneveyad-3

89voN02 82 72 14 36 5.5 24 65 gnizeerFm047level

10



Table 4. Measured storm sediment yields at gauging sites and sediment source sites. a

etiS

6991lirpA6-4 7991rebotcO10 8991rebmevoN21

latoT)gk(

evobAerutaef

)gk(

erutaeFdleiy)gk(

latoT)gk(

evobAerutaef

)gk(

erutaeFdleiy)gk(

latoT)gk(

evobAerutaef

)gk(

erutaeFdleiy)gk(

niaMllessuR 792 1.632 .a.n .a.n 11 9.033 .a.n .a.n 81 1.064 .a.n .a.n

ecneulfnoC 22 9.941 .a.n .a.n 6.6681 .a.n .a.n 0.1262 .a.n .a.n

llessuRreppU .m.n .a.n .a.n .m.n .a.n .a.n 0.909 .a.n .a.n

einahpetS 7.972171 .a.n .a.n 8.4278 .a.n .a.n 4.88821 .a.n .a.n

602SL/.hpetSreppU 61 0.192 21 7.113 3 3.979 6.425 0.623 6.891 0.4321 7.1351 7.792

2etiS/16G .m.n .m.n .m.n 8.18 5.62 3.55 0.932 7.93 3.991

5etiS/26G .m.n .m.n .m.n 7.31 8.01 8.2 5.81 5.11 0.7

01etiS/46G .m.n .m.n .m.n 3.972 4.95 9.912 6.804 9.75 7.053

21etiS/56G .m.n .m.n .m.n 0.112 9.771 1.33 9.364 3.483 6.97

72etiS/47G .m.n .m.n .m.n .m.n .m.n .m.n 3.824 0.463 3.46

43etiS/181SL .m.n .m.n .m.n 9.92 0.51 9.41 6.82 7.04 1.21-

54etiS/57G 5.9978 5.7471 0.2507 8.661 7.341 1.32 5.181 9.321 6.75

64etiS/67G 8.38832 8.38832 0.0 1.2282 9.5372 2.68 8.0862 5.0962 7.9-

74etiS/b67G 9.473 7.68 2.882 2.51 5.3 7.11 1.96 7.61 4.25

8991rebmevoN41 8991rebmevoN02 8991yraunaJ52/42

niaMllessuR 3.6082 .a.n .a.n 5.3684 .a.n .a.n

ecneulfnoC 8.7421 .a.n .a.n 0.299 .a.n .a.n

llessuRreppU 0.672 .a.n .a.n 2.455 .a.n .a.n

einahpetS 7.519 .a.n .a.n 6.1482 .a.n .a.n

602SL/.hpetSreppU 3.42 00.0 3.42 8.2 0.0 8.2

2etiS/16G .m.n .m.n .m.n 4.8 0.0 4.8

5etiS/26G .m.n .m.n .m.n 0.2 1.1 9.0

01etiS/46G .m.n .m.n .m.n 7.32 0.0 7.32 0.21 0.0 0.21

21etiS/56G .m.n .m.n .m.n 4.621 2.721 8.0- 3.44 6.8 7.53

72etiS/47G 4.5 4.5 0.0 .m.n .m.n .m.n

43etiS/181SL 0.0 0.0 0.0 0.0 snoitidnocwons

54etiS/57G 0.0 0.0 0.0 0.0 0<TniM 0.0

64etiS/67G 8.22 8.22 0.0 0.0 0<TniM 0.0

74etiS/b67G 0.0 0.0 0.0 0.0 0<TniM 0.0

a .elbacilppaton=.a.n.derusaemton=.m.n.detamitseerascilatiniseulaV

11



In contrast, at Site 12 in gully G65 (Figure 1), sediment produc-tion from the gully is higher, but the crossing produces almostno sediment (Figure 8a). This is evident from the SSC concen-trations above and below the crossing, which are very close to-gether. This gully has a stable bed, and unstable sides (Figure 8b).

On the south slope of the Confluence, there are two promi-nent features that were selected for sampling (Figure 9a); gullyG74, the unstable gully on the left side of the slope, and thelandslide scar (LS181) to the right of gully G74. The photo-graph was taken in the summer of 1994. Gully G74 experi-enced a debris flow in October 1994. Photographs taken in thespring of 2000 show evidence of ongoing instability in gullyG74, while the landslide scar is more difficult to pick out be-cause of re-growth of vegetation (Figure 9b). At Site 27 ongully G74, most of the sediment is derived from the gully itself,with a minor contribution from the crossing (Figure 10). AtSite 34, at the base of landslide LS181, sediment-productionlevels are relatively low (Figure 11), with the crossing actuallyaccumulating sediment (i.e., the SSC above the crossing has ahigher peak than that below). This occurs because there is asmall sump at the upstream end of the culvert. Water ponds inthe sump before flowing through the culvert, allowing the sandto settle out. Because water samples were collected in the ditch

Figure 6. Streamflow and SSC, Russell Creek, 12 November1998. The blue line is streamflow, the red line is SSCcalculated from turbidity, and the black diamonds areobserved SSC from samples.

Figure 7a. Streamflow and SSC at Site 2 above (red dashedline) and below (brown solid line) the crossing. The diamondsand plus signs represent SSC from samples. Figure 7b. Gully G61. Site 2 is the upper crossing.

0

20

40

60

80

100

Sus

pend

ed S

edim

ent C

once

ntra

tion

(SS

C, m

g/L)

0 .0

4 .0

8 .0

12 .0

16 .0

20 .0

Str

eam

flow

(Q

, m3/s

)

S SC Q

00:00 12 :00 00 :00 12 :00 00 :00N ov 12 1998 N ov 13 1998

N ov 12 1998 N ov 13 199800:00 12:00 00:00 12:00

0.0

0 .2

0 .4

0 .6

0 .8

Str

eam

Dis

cha

rge

(Q, m

3/s

)

0

40

80

120

160

200

Sus

pend

ed S

edi

men

t Con

cen

trat

ion

(S

SC

, mg/

L) P eak S S C = 22 00 m g/L

12



Figure 8a. Streamflow and SSC at Site 12 above (reddashed line) and below (brown solid line) the crossing. Thediamonds and plus signs represent SSC from samples.

Figure 8b. Gully G65 looking upstream from Site 12. Mostof the sediment is derived from the gully, due to its unstablesides.

Figure 9a. Russell Confluence, north slope, as of summer 1994, prior to debris flow in gully G74 (on left of slope). Thelandslide scar LS181 (center of photo) is also clearly visible, and was identified as a potential sediment source.

N ov 12 1998 N ov 13 199800:00 12:00 00:00 12:00

0

40

80

120

160

200

Su

spe

nded

Se

dim

ent C

onc

en

tra

tion

(S

SC

, m

g/L

)

0 .0

0 .2

0 .4

0 .6

0 .8

Str

eam

Dis

cha

rge

(Q

, m

3 /s)

13



Figure 9b. Russell Confluence, north slope, as of spring 2000.Note ongoing instability in gully G74, whereas the landslidescar LS181 is becoming re-vegetated, and is hard to see.

Figure 11. Streamflow and SSC at Site 34 above (reddashed line) and below (brown solid line) the crossing. Thediamonds and plus signs represent SSC from samples.

Figure 10. Streamflow and SSC at Site 27 above (red dashedline) and below (brown solid line) the crossing. The diamondsand plus signs represent SSC from samples.

above the crossing, those samples contained the sand, silt, andclay that had eroded from the ditch and other sources, whereasthe water sampled below the culvert likely contained only siltand clay.In Stephanie Creek, the two large gullies (gullies G75 and G76)show markedly different sediment-production behaviour (Fig-ure 12a). Gully G75 is a stable underlain by granitic bedrockand produces very little sediment despite its size, whereas gully

G76 lies within the Karmutsen formation and experiences fre-quent debris flows. Consequently, it is one of the largest sedi-ment producers in Russell Creek. Sediment is entrained fromthe debris flow levees that line the gully sides whenever flowrises in response to storms (Figure 12b).

At upper Stephanie Creek, the sediment yield from a small land-slide (LS206, Figure 13a) was measured by subtracting the sedi-ment yield derived from the automatic gauging site at upperStephanie from the yield derived from manual sampling abovethe landslide (Figure 13b). In this case, the continuous recordfor upper Stephanie above landslide LS206 was based on a rela-tionship between sampled SSC and turbidity as measured at thegauging site. Directly connected landslides on the sidewalls ofincised channels constitute an important type of sediment sourcein Stephanie and lower Russell Creeks. Landslide LS206 is theuppermost feature of this type in Stephanie Creek. Sedimentproduction above that point is derived from gullies and roadsources. For the 12 November storm, the yield from landslideLS206 was relatively small compared to the yield from upperStephanie. However, the variability in SSC below the featurecompared to that above is a direct result of the proximity ofthe sampler to the feature, and of the process by which thelandslide delivers sediment to the channel. The rapid fluctua-tions in SSC below landslide LS206 (Figure 13b, brown line)indicate that sediment produced from the landslide scar is de-livered to the channel by means of a series of small slumps.The sediment production from sources above landslide LS206

N ov 12 1998 N ov 13 199800 :00 1 2 :00 00 :00 12 :00

0 .0

0 .2

0 .4

0 .6

0 .8

Str

eam

Dis

char

ge (

Q, m

3/s

)

0

40

80

12 0

16 0

20 0

Su

spen

ded

Sed

imen

t Co

ncen

trat

ion

(SS

C, m

g/L

)

N ov 12 1998 N ov 13 199800 :00 12 :00 00 :00 12 :00

0 .0

0 .2

0 .4

0 .6

0 .8

Str

ea

m D

isch

arg

e (

Q, m

3/s

)

0

40

80

120

160

200

Su

spe

nd

ed

Se

dim

en

t C

onc

entr

atio

n (

SS

C, m

g/L

)

14



is integrated over time by the flow, because those sources aredispersed throughout the upper watershed, resulting in a rela-tively smooth SSC hydrograph (Figure 13b, red dashed line).

The sediment yield data from Site 45 are a special case. In No-vember 1995, a debris flow occurred in gully G76. The culvertwas plugged, which caused the flow to run down the road gradeto enter gully G75 above the crossing. The washout caused alarge amount of erosion to both the road surface and the ditchwhere it flows into the gully on the upstream side of Site 45(Figure 14a). During the storm of 4�6 April 1996, ongoing roadand ditch erosion likely contributed most of the high sedimentyield derived from the crossing (Figure 14b, Table 4). In thesummer of 1996, the road was deactivated and the crossing atSite 46 was restored. The deactivation resulted in reduced sedi-ment production from the crossing during subsequent storm events.

These measured yields were used to build the sediment-budgetmodel. That model consists of three components: a road-cross-ing model, a gully model, and a landslide model.

Road-Crossing Model

To develop a model of sediment yield from road crossings,stepwise regression analysis was used to select the best set ofpredictors (Draper and Smith 1981). All the attributes listed inTable 2a were considered. For a given storm, the best predictorof sediment yield was the total area of cut-and-fill slopes adja-

Figure 12b. Gully G76 looking upstream from Site 46. Thegully is lined with debris flow deposits such as these alongits whole length.

Figure 12a. Streamflow and SSC in gullies G75 (blue andred dashed lines) and G76 (solid blue and red lines). In bothcases, the SSC is from above the crossing.

cent to the crossing and connected to the stream channel (Fig-ure 15). There is clearly an interaction with storm rainfall, be-cause the slope of the yield vs area relationship is greater forthe larger storm. Therefore, to build a general model of sedi-ment yield from crossings, total storm rainfall and the maxi-mum 12-h and 24-h intensity parameters were introduced intothe analysis, as well as interaction variables between the rainfallvariables and the crossing attributes. Stepwise regression analy-sis was applied to the lumped data set. Data from Site 45 had tobe excluded because �unusual� yields from that site�both inthe post-washout and post-deactivation phases�influenced theregression analysis. The analysis identified a model consistingof the total area of cut-and-fill slopes adjacent to the crossingand connected to the stream channel (A, m2) and the interac-tion between A and storm rainfall (R

S, mm). The model of to-

tal storm sediment yield from crossings (YC, kg) is:

AARY SC 29.00137.0 -×= (2)

(R2 = 93.4%, s.e. = 29.75)

N ov 12 1998 N ov 13 199800 :00 12 :00 00 :00 12 :00

0 .0

0 .4

0 .8

1 .2

Str

eam

Dis

cha

rge

(Q

, m3 /

s)

0

40

80

120

160

200

Sus

pen

ded

Se

dim

ent

Co

nce

ntr

atio

n (S

SC

, mg

/L)

G u lly 75

G ully 76

SSC G 76

SSCG 75

15



Figure 13b. Streamflow and SSC at upper Stephanie aboveand below Site 206, 12 November. Note the variability ofthe SSC below the landslide (solid brown line) as comparedto that above (red dashed line).

Figure 13a. Site 206, the landslide at upper Stephaniemonitored for sediment yield.

Figure 14b. Streamflow in gully G75, and SSC above (reddashed line) and below (solid brown line) Site 45 in its post-washout state, April 1996 storm.

Figure 14a. Ditch above Site 45 after the washout ofNovember 1995 and prior to road deactivation.

If sediment yield data from Site 45 are re-introduced into theanalysis, then the resulting equation for storm sediment yieldfrom crossings becomes:

AARY SC 82.5131.0 -×= (2a)

(R2 = 98.2%, s.e. = 621.3)

Because the model coefficients of Equation 2a are an order ofmagnitude greater than those of Equation 2, the resultant cal-culated yields would be an order of magnitude greater than thechronic sediment yield that occurs from crossings under nor-mal conditions. This equation might represent unusual post-failure sediment production, however it is based on data fromone storm event at one site.

Measured and calculated yields are given in Table 5, along withstorm rainfall. Note that R

S is calculated according to the eleva-

tion of the site, using rainfall � elevation relationships that arespecific to each storm.

April 5 1996 April 6 199612:00 00:00 12:00 00:00 12:00 00:00

0.0

0.4

0.8

1.2

1.6

2.0

Str

eam

flow

(Q

, m3 /

s)

0

50

100

150

200

250

Su

spen

ded

Se

dim

ent C

once

ntr

atio

n (

SS

C, m

g/L

)

N ov 12 1998 N ov 13 199800:00 12:00 00:00 12:00

0.0

0 .4

0 .8

1 .2

1 .6

Str

eam

Dis

cha

rge

(Q, m

3/s

)

0

40

80

120

160

200

Sus

pend

ed S

edim

ent

Con

cent

ratio

n (S

SC

, mg/

L)

16



There is a bias in the model that becomes apparent if measuredyield is plotted against calculated yield (Figure 16). The biasoccurs because ditch length and area of ditch cut slopes are notsignificant terms in the model. However, it is known that ditcherosion contributes sediment to stream crossings, and that thesediment eroded off road surfaces is also delivered to crossingsby way of ditches (e.g., Reid and Dunne 1984). For this reason,the model is likely to underestimate sediment yield. Therefore,a one-tailed test can be used to account for road surface andditch erosion (Table 5, Figure 16). In Figure 16, the solid linerepresents a relationship between measured and calculated yield.The dashed line is equal to the solid line minus 2 s.e.1 All thedata points that fall within that band represent erosion fromonly the cut-and-fill slopes adjacent to the crossing. Thus, anydata points that lie above the dashed line represent sedimentproduction from the crossing, and also include ditch and/orroad surface erosion. In Table 5, these values are calculated anddivided by the length of connected ditch emptying into thecrossing, resulting in a logarithmic model of road erosion as afunction of storm rainfall:

18.3)ln(83.0 -×= SB RE (3)

(R2 = 98.2%, s.e. = 0.059)

8.19)ln(18.6 -×= SM RE (4)

(R2 = 75.4%, s.e. = 1.509)

whereE

B and E

M are branch road and mainline road erosion re-

spectively, in kg/m of ditch connected to the stream chan-nel at the crossing.

Branch road erosion consists primarily of ditch erosion, whereasmainline erosion also includes contributions of fine sedimentfrom the road running surface.

There is a threshold of 25 mm of storm rainfall required be-fore mainline road surface will occur. For branch erosion, thethreshold is higher, at about 42 mm (Figure 17). The reason forthe difference in the threshold is that on mainline haul roads,fine sediment is continually being created by truck traffic. Thissediment is more easily eroded than ditch sediments or sedi-ment on the running surfaces of occasional-use roads. Thismodel describes chronic sediment production from road sur-faces. Like the road-crossing model, it does not account forepisodic sediment production such as the April 1996 event atSite 45, where the sediment produced by ditch erosion was over50 times the erosion rate that would be predicted by the modelfor normal branch roads (Figure 17).

The running surfaces of mainline roads are known to be thelargest producer of fine sediments (Reid and Dunne 1984). Fre-quent-use mainline road surfaces produce one order of magni-tude more sediment than moderate use or temporarily unusedmainline roads. Reid et al. (1981) broke down sediment yieldfrom mainline road erosion into categories according to roaduse. Annual yield during one measurement year was reportedas 73 kg/y/m of road length during heavy use periods and 9.75kg/y/m during temporary non-use periods. These results werederived from reconstructing unit hydrographs and SSC ratingcurves for culvert flows, from rainfall records. Measured rain-fall for the year for which the yield was calculated was 3468mm. Mainline roads were reported as sustaining heavy use 48%of the time. Temporary non-use periods included weekendsand overnight during the week. This results in a weighted aver-age yield of 160.4 kg/m/y, and includes sediment yield fromroad surfaces, cut slopes, and ditches.

While the methods used by Reid were different from the meth-ods described in this report, both methods are based on stormrainfall. Therefore, Reid�s results can be used to verify Equa-tion 3. This is important because Equation 3 is based on datafrom only one site with 10 m of connected ditch, and thereforea small error in measuring any of the quantities can lead to alarge error in applying the equation.

A total of 310 measured storm runoff events have been identi-fied for Russell Creek between November 1991 and July 1999,

Figure 15. Sediment yield at road crossings vs area of cut-and-fill slopes for selected storms.

1 Normally, in a two-tailed test the regression line would be drawn with a +1 s.e.band, creating an error band of 2 s.e. in width on either side of the regressionline. Because this is a one-tailed test, all the error is loaded onto the negative sideof the regression line. In this way, we are treating overprediction as due to randomerror, and underprediction as due to a systematic error, resulting from the factthat road surfaces and ditches are not represented in the crossing yield model.

0 200 400 600A rea o f C ut and F ill S lopes (m 2)

0

100

200

300

400

Sto

rm S

edi

men

t Y

ield

(kg

)

0 1 00 2 00 3 00

0

2 000

4 000

6 000

8 000

192 m m Storm62 m m Storm50 m m Storm

P ost-W a sh outS ite 45

R oad D eactiva tionS ite 45

17



Table 5. Measured and calculated crossing yields. a

etiS aerAm( 2)

niaR)mm(

derusaeMY,dleiy

)gk(

detaluclaCclacY,dleiy

)gk(noisoredaoR

)gk(

hctidm/noisoredaoR

E,hcnarBB

)m/gk(E,niaM

M

)m/gk(

2etiS 492 3.15 3.55 4.121 0.0

5etiS 23 9.74 8.2 7.11 0.0

01etiS 465 7.84 9.912 8.212 1.7 821.0

21etiS 32 1.84 1.33 5.8 6.42 064.2

43etiS 51 0.05 9.41 9.5 0.9 060.0

54etiS 192 5.35 6.74 8.821 0.0

64etiS 0 5.35 0.0 0.0 0.0

74etiS 46 9.35 2.91 7.82 0.0

2etiS 492 0.67 3.991 6.212 0.0

5etiS 23 3.06 0.7 9.8 0.0

01etiS 465 4.56 7.053 8.333 9.61 703.0

21etiS 32 6.16 6.97 4.4 2.57 615.7

72etiS 131 0.95 3.46 6.95 7.4 532.0

43etiS 51 0.95 7.04 5.0- 2.14 492.0

54etiS 192 0.95 6.75 5.241 0.0

64etiS 0 0.95 7.9- 3.8- 0.0

74etiS 46 0.95 4.25 9.42 5.72 831.0

2etiS 492 5.53 4.8 7.75 0.0

5etiS 23 0.13 9.0 3.4 0.0

01etiS 465 1.23 7.32 7.48 0.0

21etiS 32 3.13 8.0- 2.3 0.0

01etiS 465 6.42 0.21 4.62 0.0

21etiS 32 0.22 0.0 0.0 0.0 000.0

21etiS 32 0.44 7.53 2.7 5.82 158.2

54etiS 192 2.291 0.2507 9.186 2.0736

64etiS 0 2.291 0.0 0.0 0.0

74etiS 46 2.291 2.882 0.051 2.831 322.1

a .dnab.e.s1rewolafossecxenidleiyynaebotnoisoredaorgnimussayb,tsetdeliat-enoagnisudenimretederanoisorehctiddnadaoR

18



of which 122 were over the threshold of 25 mm of rainfall.Storm rainfall contributing sediment from mainline roads rangedfrom 25 to 133 mm, and averaged 49 mm. Total annual rainfallfor the period averaged 1826 mm. Application of Equation 3to the storm record resulted in a mean annual yield from main-line roads of 54.6 kg/m of connected ditch. If the storm rain-fall for the 310 events is multiplied by a factor of 1.9 to matchthe total annual rainfall for Reid�s study, application of Equa-tion 3 to this amplified storm data set results in an average an-nual yield of 156.7 kg/m. This matches Reid�s results for aver-age mainline road use.

Gully Model

Many of the gullies that were monitored have crossings abovethe monitoring site; therefore, the first step in developing thegully model was to remove the sediment produced by thoseadditional crossings from the measured gully yields. This re-sulted in a data set that is summarized in Table 6.

For each gully, it was found that there is a relationship betweenstorm sediment yield and storm rainfall (R

S) that takes the form

220 SG RbbY += (5)

whereY

G is the gully yield in kg, and

b0 and b

2 are coefficients (Figure 18).

Note that this is a second order polynomial without the linearterm. The coefficients of this equation are given in Table 6 foreach gully. Note also that the April 1996 storm is an importantdata point in this relationship that was measured directly onlyon gullies G75 and G76. The regressions of yield vs rainfallwere first performed on the data for all gullies excluding theApril 1996 storm. Then the regressions were re-done for gul-lies G75 and G76 with the April 1996 yields included. Based ona linear relationship between the two sets of coefficients forgullies G75 and G76, yields from the other gullies for the April1996 storm were estimated.

A general model of gully yield was developed to predict thecoefficients of Equation 5 from gully attributes listed in Table2b. Also included were interaction variables. As with the cross-ing model, the gully yield model predictors were identified us-ing stepwise regression, resulting in the following equations:

20737.010062.0773.0134.00407.02 SSTATAb GG ×+×-×+×-×=

(6)

23.8414.167.8513021243 220 SSTbTbb ×+×-×-×-×=

(7)

The relationship between the coefficients and gully attributes(gully stability S1, S2, and T, and area A

G, m2) is shown in Figure

19. The terms S1, S2, and T are category variables that repre-sent the stability of the channel bed (S1) and channel banks

Figure 17. A model of sediment yield under “normal”conditions from ditches and from running surfaces ofmainline roads. Note that the ditch yield from Site 45 in post-washout phase is about an order of magnitude greater thanthe model predicts.

Figure 16. Measured vs calculated sediment yield fromcrossings (solid line). A one-tailed test suggests that thedata points within the -2 s.e. band (dashed line) representyield from cut-and-fill slopes, whereas points above the solidline include ditch and road erosion.

0 1 00 2 00 3 00 4 00C a lcu la ted Y ie ld (kg)

0

1 00

2 00

3 00

4 00

5 00

Mea

sure

d Y

ield

(kg

)

R o a d S urfa ce an d D itch E ros io n

R o a d D e activa tio n

20 60 100 140 180S torm R a in fa ll (m m )

0.0

0.4

0.8

1.2

Bra

nch

Ditc

h E

rosi

on

(kg/

m)

0

20

40

60

80

Ma

inlin

e E

rosi

on (

kg/m

)

20 60 100 140 180

P ost-w ashout E ros ionS ite 45

19



Table 6. Gully yield data set for modeling. a

Figure 19. The gully yield model consists of a method ofpredicting the equation coefficients of the curves in Figure18, based on gully attributes.Figure 18. Gully yield as a function of storm rainfall.

0 40 80 1 20 1 60 2 00S torm R a in fa ll (m m )

0

10 00

20 00

30 00

Sto

rm S

ed

ime

nt

Yie

ld G

65

/75

(kg

)

0

50 00

10 000

15 000

20 000

25 000

Sto

rm S

ed

ime

nt

Yie

ld G

76

(kg

)

0

20 0

40 0

60 0

80 0

Se

dim

en

t Y

ield

(kg

)

G u lly 76 (to rrented)G ully 65 (unstab le banks)G ully 75 (stab le)

G 64 (stab le)G 61 a t S ite 2 (scoured)

0 0.4 0.8 1.2 1.6G ully Catchm ent A rea (km 2)

0

0.4

0.8

1.2

1.6

B2 C

oef

ficie

nt

of

Yie

ld v

s.

Ra

infa

ll E

qua

tion

Torrented G ulliesG ullies w ithUnstab le BanksStab le G ullies

0 0.2 0.4 0.6 0.8B 2 Coef f ic ient

-160

-120

-80

-40

0

B0 C

oe

ffic

ien

t

.onylluGselbairavnoitacifissalcytilibatS

aeradeguaGmk( 2)

stneiciffeocRsvYtnemmoC

1S 2S T b0 b2

16 1 0 0 235.0 6.63- 5510.0 rewolnoitisoped,2etiStaderuocS

26 0 0 0 632.0 20-E16.1- 60-E07.3 elbatS

46 0 0 0 467.0 1.44- 0330.0 elbatS

56 0 1 0 410.1 4.56- 0511.0 sknabelbatsnU

47 1 1 1 612.0 3.42- 7901.0 secruostnemidesevitca,detnerroT

57 0 0 0 272.1 5.76- 4250.0 elbatS

67 1 1 1 049.0 0.95- 0996.0 secruostnemidesevitca,detnerroT

niarmrotS)mm(

dleiymrotS)gk(

niarmrotS)mm(

dleiymrotS)gk(

niarmrotS)mm(

dleiymrotS)gk(

niarmrotS)mm(

dleiymrotS)gk(

16 6.571 0.054 3.15 5.62 9.08 6.81 5.53 0.0

26 0.061 201.0 9.74 0.0 3.06 0.0 0.13 0.0

46 9.361 6.448 7.84 4.95 4.56 9.75 1.23 0.0

56 0.161 0.8192 1.84 8.871 6.16 9.103 3.13 2.721

47 9.561 9.3992 0.95 7.953 0.71 4.5

57 4.581 5.7471 5.35 7.341 0.95 9.321 0.71 0.0

67 4.581 5.07932 5.35 4.9372 0.95 2.7072 0.71 8.22

a .detcartbussinoitubirtnocdetaler-daoR.setubirttayllugtnacifingisehtylnosedulcnI

20



(S2) of the gully in question, and whether or not the gully istorrented. Each term is 1 if unstable and 0 if stable; togetherthey form a single parameter to describe gully stability. Thus,Equations 5�7 form a 3-parameter model built on 27 data points.The overall R2 is 99.9% with a weighted RMS error of 4.8%.

Landslide Model

To determine the chronic sediment yield from landslides, it wasassumed that a landslide would deliver a volume of sediment tothe channel that is proportional to its surface area, and to rain-fall intensity (Rood 1984). The landslide model is based on themeasured yields from landslide LS206 (Table 7, Figure 20), withthe exception of the April 1996 event, for which landslide yieldwas not measured directly, and 20 November 1998, during whichlandslide LS206 was under snow. As an alternate method, theyield from lower Russell (below the Confluence) was divided bythe total landslide area in lower Russell (2330 m2) to estimatelandslide yield for those storms. The yield from lower Russell isestimated as the yields from Russell Creek, minus the sum ofyields from the Confluence, Stephanie Creek, and additionalyields from two gullies in lower Russell. Because the Confluencegauging site was not operating at that time, a method was neededto estimate the yield at the Confluence for the April 1996 event.If all measured storm yields from the Confluence are plottedagainst those of Russell Creek (Figure 21), the resulting linearrelationship has an R2 of 95.0% and a s.e. of 506 kg. This uncer-tainty in the Confluence yield leads to an error in the estimatedlandslide yield of +0.2 kg/m2.

Figure 20. Landslide yield per unit area, as a function of themaximum 24-h storm rainfall intensity. The graph includesyields calculated from LS 206 (upper Stephanie) and fromlower Russell landslides.

etaD Rs

)mm(R 42

)mm(etisecnerefeR

dleiyeinahpetSreppU

dleiyedilsdnaLm/gk( 2)

602evobA)gk(

602woleB)gk(

69rpA40 331 611 1R 06.14

79peS03 24 34 hpetS 7.036 3.838 82.1

79tcO10 55 55 hpetS 0.623 6.425 32.1

79tcO51 04 05 hpetS 0.191 2.333 88.0

79tcO61 02 05 hpetS 1.563 0.545 11.1

79tcO03 301 46 hpetS 2.272 5.246 92.2

79voN20 75 24 hpetS 3.6741 4.6661 71.1

79voN50 05 44 hpetS 9.426 9.047 27.0

89voN21 85 95 hpetS 0.4321 7.1351 48.1

89voN41 44 12 hpetS 0.0 3.42 51.0

89voN02 82 24 1R 95.0

Table 7. Table of measured landslide yields and storm rainfall characteristics.

0 40 80 120M axim um 24-hour R ainfa ll R 24 (m m /24hr)

0.1

1

10

100

Land

slid

e Y

ield

YLS

(kg

/m2 )

U pper S tephanie (LS206)Low er R ussell

21



Russell. For the Confluence, the budget applies to the area be-tween the Confluence gauging site and the upper Russell gaug-ing site. Similarly, at Stephanie Creek, upper Stephanie is con-sidered as one unit and the budget applied to the area that drainsto the channel below the upper Stephanie gauge (Figure 1).Because upper Russell began operation in the middle of Octo-ber, suspended sediment yield had to be estimated for the stormof 1 October 1997. A relationship was developed between themeasured sediment yields at upper Russell and the Confluence(Figure 21). The resulting linear relationship has an R2 of 90.0%and a s.e. of 141 kg.

The sediment-budgeting procedure for Stephanie Creek andlower Russell is relatively straightforward, because the sedimentsources in those sub-basins are directly connected to the chan-nel network. However, this is not the case for the Confluence,because the valley flat absorbs the sand component of the sus-pended load before it reaches the main channel. Therefore, itwas necessary to multiply the source yields by the proportionof fines found in the sediment sources.

The mean grain-size distribution of different source types wasreported by Maynard (1991) and later verified by Paulig (1996)by performing sieve analysis on samples collected from selectedsources (Table 8). Because only the sand and silt fractions aretransportable in suspension, the proportion of fines (Pƒ

T) in

the transportable load is equal to:

finessand

finesPfT %%

%

+

= (9)

Thus for the Confluence, the yields from road crossings andditches, mainline road surfaces, and landslides are multiplied by0.16, 0.22, and 0.24 respectively. To derive a multiplier for gullyyield is somewhat more complex, because the proportion ofsand transported by channels is related to hydrologic condi-tions and channel gradient (Hudson 2001).

The proportion of sand in the suspended load is in part relatedto sediment supply; however, an examination of selected eventswhere supply is not a limiting factor shows there are logarith-

The best predictor of landslide yield was the 24-h maximum rain-fall intensity (R

24), for an exponential relationship as follows:

240572.00567.0 ReYLS

×

= (8)

(R2 = 99.2%)

where

YLS

is landslide yield in units of kg/m2 of landslide surface area.

Application of the Sediment-Budget Model

As discussed above, the sediment-budget model is an account-ing procedure in which the yields from all significant sedimentsources in a sub-basin of Russell Creek are added up and com-pared to the measured mainstem yield for that sub-basin. Mea-sured source yields were used where available, and all other yieldsare calculated using the yield models described above. This re-sults in a sediment-budget model that is essentially rainfall based.Sediment budgets were calculated for the 1 October 1997 stormand for three storms in November 1998; as well, a �tentative�budget was calculated for the April 1996 storm.2

The sediment budget is applied to sub-basins of Russell Creek:Russell Creek above the Confluence, Stephanie Creek, and lower

Figure 21. Relationships between storm yields at RussellMain, the Confluence, and upper Russell.

2 Details available on request. Sediment-budget calculations are summarized inTable 10.

Table 8. Mean grain-size distribution of sediment sources,by type.

epytecruoS%

dnaS%

seniF

niseniF%elbatropsnart

noitcarf

llif-dna-tuCsehctid,epols

24 8 0.61

edargdaoR 53 01 2.22

eruliafepolS 54 31 4.22

wolfsirbeD 53 5.3 1.9

0 10000 20000 30000Y ie ld a t R usse ll M a in (kg)

0

1000

2000

3000

4000

5000

Yie

ld a

t the

Con

fluen

ce (

kg)

0 1000 2000 3000 4000 5000Y ie ld a t the C onf luence (kg)

0

200

400

600

800

1000

Yie

ld a

t Upp

er R

usse

ll (k

g)

Y U pR uss = 0 .213 Y C onfluence

Y Confluence = 0 .1 70 Y Russ

22



mic relationships between the proportion of sand in the sedi-ment yield and the channel gradient (Table 9a, Figure 22a). Thecoefficients of the relationships are related to the logarithm ofstorm rainfall (Figure 22b and 22c). This allows the proportionof sand transported by a channel to be calculated, based on thechannel gradient and the storm rainfall. The mean gradient ofgullies in the Confluence is 41%; thus there is a relationshipbetween the proportion of sand (hence the proportion of fines)in the sediment load carried by gullies, and storm rainfall (Table9b, Figure 22d). An equation to calculate the proportion of thegully load transported to the main channel in the Confluence is:

( )Sfines RP ln334.089.1 ×-= (11)

or, Equation 11 can be expressed as a percentage by multiply-ing the coefficients by 100.

The sediment-budget model was very successful (Table 10).For the November 1998 storms, error in the budget waswithin 10% for sub-basins and for Russell Creek as a whole.Error was somewhat higher for the 1 October 1997 storm, atbetween 9 and 13.5% for Russell Creek and sub-basins. Thereason for the higher error during these storms was probably

Table 9a. Calculated proportion of sand in sediment yield for selected storms at Stephanie and Russell Creeks. Anequation to predict the proportion of sand as a function of channel gradient for a given storm is given as:

IGradientCPsand +×= )ln( (10)(e.g., Figure 22a). The gradients of Russell Main, Russell at the Confluence, and Stephanie Creek are 5%, 3.1%, and 15%respectively. The equation coefficients C and I are related to storm rainfall.

Table 9b. Calculation of proportion of sediment production transported by gullies across the valley flat. The coefficients ofEquation 10 given in Table 9a are related to storm rainfall, also given in Table 9a (Figure 22b, c). These relationships can be used topredict coefficients C and I from storm rainfall for the storm for which one wants to know the proportion of sand or fines in thesediment load of a channel with a known gradient. Given that the mean gradient of Confluence gullies is 41.4%, Equation 10 canthen be used to calculate the proportion of sand for the storms given below. This allows Equation 11 to be derived, which is used todetermine the proportion of fines transported by Confluence gullies for a given storm (Figure 22d).

etadmrotSnisabSRegareva C I

yllugnaeMtneidarg

)%(dnasP)%(

senifP)%(

69rpA40 0.751 720.92 2099.72- 24.14 01.08 9.91

79tcO10 5.15 946.51 8644.51- 24.14 38.24 2.75

89voN21 5.66 717.81 3323.81- 24.14 73.15 6.84

89voN02 4.23 680.01 8132.01- 24.14 33.72 7.27

etadmrotS

einahpetS llessuR R(niaR s)llessuRta

niaM)mm(

svdnasP)tneidarGlennahC(nI

dleiySS)gk(

dnasP)%(

dleiySS)gk(

dnasP)%(

C I

69rpA40 171 7.972 0.24 792 1.632 0.32 331 410.52 667.32-

69rpA70 64 7.701 0.41 22 7.790 0.61 llessuRotevitalerdetimilylppuseinahpetS

79peS03 31 1.127 0.92 61 6.787 0.61 55 352.71 633.61-

79tcO10 8 4.369 0.8 11 0.133 0.31 llessuRotevitalerdetimilylppuseinahpetS

79tcO30 4 1.297 0.4 9.287 0.0 42 927.2 3326.3-

79tcO51 5 0.764 0.71 5.1515 0.31 34 610.01 1191.9-

79tcO62 2 7.480 0.01 3 2.067 0.8 43 655.5 290.4-

23



that SSC at the main gauging sites was derived entirely fromSSC vs turbidity relationships. Calculations could not be veri-fied because the automatic samplers were not working duringthat event.

DISCUSSION

Results show that landslides directly connected to the mainstemchannels are the biggest contributors of suspended sediment.In the four storms analyzed, the contribution from landslideswas between 50 and 90% of the sediment yield. Gullies weregenerally less important than landslides as sediment sources,with the exception of gully G76. For example, in the 12 No-vember 1998 storm, gully G76 accounted for 63% of the totalyield from gullies. The biggest producer in Russell Creek is land-slide LS199 (Figure 23), which happens to be in the unharvestedpart of Stephanie Creek. During the 12 November storm, gullyG76 and landslide LS199 together accounted for 44% of thetotal yield from Stephanie Creek and 35% of the yield fromRussell Creek as a whole. Both these features are unharvested.On the average, Stephanie Creek accounts for about 60% ofthe sediment yield of Russell Creek3 , even though it accountsfor only 28% of the area. The dominance of Stephanie Creek,and in particular, gully G76 and landslide LS199, over the sedi-

Figure 22. Determination of Equation 11. 22a, top left:determination of Equation 10, example given for 9/30/97storm. 22b and c, top right and bottom left: prediction of thecoefficients of Equation 10 based on storm rainfall. 22d,bottom right: a relationship to determine the proportion offines in the total suspended sediment yield carried by Con-fluence gullies as a function of storm rainfall.

Table 10. Summary of sediment-budget results for fourstorms. All sediment yields are in kg of suspended sedimentper storm. Total sediment budget for Russell Creek includesall sources from Stephanie, Russell above the Confluence,and lower Russell sub-basins.

3 This is based on an average of 24 events where yield at both Stephanie Creekand Russell Main were measured, but excludes 4 events where Stephanie Creekwas under snow.

nisab-buS

mm2.05:llafniar.gvanisaB7991tcO1

latoT)gk(

reppU)gk(

seilluG)gk(

sdaoR)gk(

sedilsdnaL)gk(

einahpetS

derusaeMdleiy

5278 623

detsevraH 91 334 132

detsevrahnU 9713 .a.n 8075

latoT 8913 334 9395

secruoslatoT 6989

)%(rorrE 4.31

ecneulfnoC

derusaeMdleiy

7681 293


detsevrahnU .a.n .a.n 32

latoT 529 672 73

secruoslatoT 0361

)%(rorrE 7.21-

keerCllessuR,tegdublatoT

derusaeMdleiy

9.03311



latoT 3214 907 5557

secruoslatoT 78321

)%(rorrE 3.9

snoitroporP

detsevraH 6.7 7.9

detsevrahnU 7.52 3.15

latoT 3.33 7.5 0.16

0 4 8 12 16C hannel G rad ien t G c (% )

0

10

20

30

40

Psa

nd

(%)

0 40 80 120S torm R a in fa ll R s (m m )

0

5

10

15

20

25

30

C c

oe

ffic

ien

t

0 5 10 15 20 25 30C coe f f ic ien t

-25

-20

-15

-10

-5

0

I co

eff

icie

nt

0 40 80 120 160S torm R a in fa ll R s (m m )

0

20

40

60

80

Pfin

es c

arr

ied

by

gu

llie

s(%

)

P sand = 17.25 ln (G c) - 16 .34 C = 13 .69 ln(R s) - 40 .90

I = -0 .94 C + 0 .05

P fines = 189 - 33 .4 ln (R s)

24



Table 10, continued: Table 10, continued:

nisab-buS

mm4.06:llafniar.gvanisaB8991voN21

latoT)gk(

reppU)gk(

seilluG)gk(

sdaoR)gk(

sedilsdnaL)gk(

einahpetS

derusaeMdleiy

88821 4321



latoT 4192 936 8238

latoTsecruos

51131

)%(rorrE 8.1

ecneulfnoC

derusaeMdleiy

1262 909



latoT 5311 924 38

latoTsecruos

6552

)%(rorrE 5.2-


derusaeMdleiy

1.06481

detsevraH 1611 4801 8082


latoT 8724 4801 23311

latoTsecruos

49661

)%(rorrE 6.9-

snoitroporP

detsevraH 0.7 8.61


latoT 6.52 5.6 9.76

nisab-buS


latoT)gk(

reppU)gk(

seilluG)gk(

sdaoR)gk(

sedilsdnaL)gk(

einahpetS

derusaeMdleiy

619 0

detsevraH 01 0 85


latoT 19 0 509

latoTsecruos

599

)%(rorrE 7.8

ecneulfnoC

derusaeMdleiy

8421 672

detsevraH 149 861 3


latoT 149 861 8

latoTsecruos

4931

)%(rorrE 7.11


derusaeMdleiy

6082



latoT 6911 281 3131

latoTsecruos

1962

)%(rorrE 1.4-

snoitroporP

detsevraH 3.53 2.11


latoT 4.44 8.6 8.84

25



ment budget is undoubtedly due to the underlying lithology.

Unharvested sediment sources are more important than har-vested sources to the sediment budget in Russell Creek. Har-vested sources include the gullies and landslides that are in ar-eas that have been harvested, and do not include road-relatedsources. The contribution from harvested sources is between17 and 47% of the sediment budget, with an average of 30%(Table 11a). Because about 30% of the watershed area has beenharvested, this might suggest that harvesting has had no effecton sediment production from gullies and landslides in this wa-tershed. This is based on the assumption that sediment pro-duction is uniformly distributed across the landscape. How-ever, because the sediment budget is dominated by two unhar-vested features, this assumption is questionable.

If the relative contribution of harvested and unharvested soucesare recalculated with contributions from gully G75 and land-slide LS199 excluded, the resultant contributions are perhapsless biased (Table 11b). In this case, the harvested contributionranges from about 30�60%, with an average of 47%. In lowerRussell, the sediment yield is almost entirely due to landslides,all of which are harvested. However, connected landslides tendto be concentrated in the incised channel reaches. Consideringthat all gullies in the Confluence (i.e., the area between RussellCreek above Stephanie confluence and Russell Creek at TS120)have been at least partly harvested, it is difficult to draw con-clusions about the effect of harvesting on gully yield. The in-stability in gullies G61, G65, and G74 could be a result of har-vesting. The Confluence area (between the Confluence andupper Russell stream gauges) is 19.5% of the area of RussellCreek. The average contribution from Confluence gullies to thetotal yield of Russell Creek is in the range of 14 to 23%, de-pending on whether buffering is considered and on whether ornot the contributions from roads and from gully G76 and land-slide LS199 are included (Table 11c). In any case, the percentcontribution to Russell Creek yield from gullies in theConfluence is proportional to the area of the Confluence, ex-

Table 10, continued:

Figure 23. Landslide LS 199, Stephanie Creek, producesmore sediment than any other source in Russell Creek.

nisab-buS


latoT)gk(

reppU)gk(

seilluG)gk(

sdaoR)gk(

sedilsdnaL)gk(

einahpetS

derusaeMdleiy

2482 0

detsevraH 0 0 532


latoT 13 0 4572

latoTsecruos

5872

)%(rorrE 0.2-

ecneulfnoC

derusaeMdleiy

299 455

detsevraH 524 95 5


latoT 524 95 21

latoTsecruos

9401

)%(rorrE 7.5


derusaeMdleiy

5.3684



latoT 654 36 2214

latoTsecruos

1464

)%(rorrE 6.4-

snoitroporP



latoT 654 36 2214

26



pressed as a percentage of the total area of Russell Creek. Basedon these figures, it is not possible to conclude that harvestinghas increased sediment yield from gullies or from landslides. Ifthere is an effect, it is minor compared to the dominant effectof lithology.

There is no relationship between the relative contribution fromharvested sources and storm size. Likely, this is because thedistribution of sediment production by source type is relatedto precipitation distribution, which is highly variable. Also, thepresence of snow affects sediment production. However, thereis a distinct relationship between the contribution from roadsources and storm size (Table 11, Figure 24). The contributionfrom roads at the source varied from 7% for a small (40 mmrainfall) storm to 23% for a large (140 mm rainfall) storm. If

Table 11a. Summary of relative contributions of harvested sources and roads to the overall sediment yield of RussellCreek, and the effect of the valley flat on those yields. Buffered refers to the sediment yield as delivered to channels, andunbuffered refers to sediment yield at the source; both are calculated excluding contributions from upper Russell andupper Stephanie.

Table 11b. As above, but excluding gully G76 and landslide LS191.

Table 11c. Summary of contributions from Confluencegullies to the sediment yield of Russell Creek at RussellMain. a

etaD

,.gvanisaBniarmrots

)mm(

secruoslla,dleiylatoT%

noitcuder.F.Vyb

morfnoitubirtnoCaeradetsevrah sdaormorfnoitubirtnoC

dereffuB)gk(

dereffubnU)gk(

dereffuB)%(

dereffubnU)%(

dereffuB)%(

dereffubnU)%(

69rpA40 9.141 42 5109 563 3.983 8.13 .a.n 5.03 .a.n 0.32

tcO10 79 2.05 21 783 51 331 1.81 3.71 4.22 7.5 0.51

voN21 89 5.06 61 496 22 051 6.42 8.32 8.33 5.6 5.51

89voN41 7.45 2 196 4 323 8.73 5.64 7.35 8.6 4.22

89voN02 5.04 4 146 5 603 5.21 8.13 7.43 4.1 4.7

etaD

,.gvanisaBniarmrots

)mm(

secruoslla,dleiylatoT%

noitcuder.F.Vyb

morfnoitubirtnoCaeradetsevrah sdaormorfnoitubirtnoC

dereffuB)gk(

dereffubnU)gk(

dereffuB)%(

dereffubnU)%(

dereffuB)%(

dereffubnU)%(

69rpA40 9.141 181 200 792 673 1.93 .a.n 9.73 .a.n 7.82

tcO10 79 2.05 7 164 01 702 9.62 7.82 2.33 5.9 7.12

voN21 89 5.06 01 288 61 833 4.33 6.35 8.33 0.01 0.12

89voN41 7.45 2 133 3 369 2.14 7.35 5.85 8.7 4.42

89voN02 5.04 2 698 3 165 7.81 15 8.15 2.2 1.11

latotfo%dleiy

,dleiyfo%gnidulcxenoitubirtnocsdaormorf

,dleiyfo%gnidulcxenoitubirtnoc

&67Gfo991SL

dereffuB 41 51 02

dereffubnU 61 71 32

a .stneve4revodegarevasnoitubirtnoctnecreperaelbatehtnisrebmuN

27



delivery are analyzed by comparing sediment yield at the sourceto that delivered to the channel (Table 11). The valley flat re-duces the overall sediment yield to Russell Creek as measuredat Russell Main by a factor that ranges from 12% to about 40%(Table 11a), depending on storm size and whether or not thecontributions from G75 and landslide LS191 are included (Table11b). Generally, the larger the storm, the greater the effect ofthe buffering: this is because larger storms are capable of en-training more sand, all of which is absorbed by the valley flat.The valley flat also reduces the contribution from road-relatedsources to a maximum of 10%, because most sources of thattype are concentrated in the Confluence.

If the effect of the valley flat is analyzed for the Confluencealone, then the buffering capacity is 50% for storms in the rangeof 50�60 mm rainfall. An interesting effect of the bufferinginvolves gully G61 as it crosses the mainline road. The lowerpart of the gully is torrented and unstable (Table 2), thus a largesediment yield from that feature is expected. A debris flow oc-curred in that gully in the fall of 1995, but the deposition ranout on the mainline road. Because the gully is scoured at Sites 2and 3, the gully is usually flowing at those sites. At low to me-dium flow, the flow becomes sub-surface through the debrisflow deposit. There is a culvert at the road crossing, but a rain-fall event of at least 30 mm is needed to induce flow throughthe culvert. This results in a reduction in effective rainfall atthat site, with a subsequent reduction in sediment yield in thesediment-budget calculations.

Forest Management Implications

The results of these investigations suggest that watershed mor-phology has a greater influence on the sediment budget ofRussell Creek than forest-harvesting activities. The bufferingcapacity of the valley flat is sufficient to absorb an amount ofsediment equivalent to the contribution from road-relatedsources as measured at the source (Table 11). The sedimentbudget of Stephanie Creek, which is not buffered by a valleyflat, is dominated by two large sediment sources that are in theunharvested area. Even in the case of the April 1996 storm inthe aftermath of the road washout at Site 45, road-relatedsources contributed only 6.9% of the suspended sediment inStephanie Creek. In the Confluence, the sediment is mostlyderived from harvesting-related sources because all of the gul-lies in that part of the watershed have been harvested, at leastpartially. Sediment-yield data suggest that harvesting may havecaused a slight increase in yield from harvested sources, but thisis not conclusive. Furthermore, an earlier analysis of sedimentyields at Russell and Catherine Creeks revealed that basin li-thology was a more important factor in controlling sedimentyield than forest development (Hudson and Sterling 1998). Takentogether, these results suggest that forest management activi-ties have had only minor effects on the sediment budget ofRussell Creek.

Given the above conclusion, deactivation may have little effecton sediment production from roads, particularly in theConfluence, because the buffering capacity of the valley flat is

the contribution from gully G75 and landslide LS199 is excluded,the proportions are increased to range from 11 to 29%.

The storm that occurred on 14�15 November 1998 was differ-ent from the other storms in that rainfall intensity was low (Table3). Because of this, yield from landslides was less than fromother sources. However, the rainfall that contributed to the stormwas sustained over three days, such that the gullies producedsediment yields comparable to other events. In addition, upperStephanie was under snow at the beginning of the event, whichreduced the yield from gullies G75 and G76 (Table 4). Conse-quently, the Confluence produced more sediment than StephanieCreek. This resulted in an increase in the relative contribution fromharvested and road-related sources for that event.

Connectivity of sediment sources has a major effect on sedi-ment yield. As discussed above, Stephanie Creek usually con-tributes about 60% of the suspended sediment load of RussellCreek, except for events such as the 14 November event, whereStephanie Creek was under snow and lower parts of the water-shed were snow-free. On the average, 25% of the total yield ofRussell Creek is derived from landslides in the incised lowerreaches of Russell Creek below Stephanie confluence.4 The areaof lower Russell is 6.24 km2, or 20% of the total watershedarea. Stephanie Creek is dominated by the Karmutsen forma-tion, whereas lower Russell is underlain by the granitic IslandIntrusive formation. Because the areas of Stephanie Creek andlower Russell are comparable, this demonstrates the importanceof lithology compared to connectivity as factors controllingthe sediment budget.

The effects of the valley flat in the Confluence on sediment

Figure 24. Contribution of road-related sources to thesediment budget of Russell Creek. The roads become moresignificant contributors as storm rainfall increases, but are lesssignificant when the main channel is buffered by a valley flat.

4 Averaged over 27 storm events with yield data from Russell Main, StephanieCreek, and the Confluence.

0 40 80 120 160Storm R ainfa ll R S (m m )

0

5

10

15

20

25

Pro

port

ion

of S

edim

ent Y

ield

D

eriv

ed f

rom

Roa

ds

(%)

U nbufferedB uffered

28



sufficient to absorb about half of the sediment that is pro-duced. However, this may not always be the case: in water-sheds that lack that buffering capacity, road deactivation mightreduce sediment yield substantially. The sediment budget canalso be used to show the effect of road-related sources on sedi-ment yield in watersheds that are not buffered by a valley flatand to predict the potential effects of deactivation on roadcrossings, and of harvesting-induced debris flows in gullies.

In the Confluence, the most significant crossings in terms ofsediment production are the large ones such as those at Site 2(Figure 25) and Site 10. At those sites, the primary mechanismresponsible for sediment production seems to be rill erosionon the fill slopes. Rilling also occurred at Site 45; however, thedeactivation of the road appears to have succeeded in reducingsediment production from that crossing by diverting water awayfrom the fill slope (Figure 26 and Figure 15). Thus, two pos-sible solutions for reducing sediment yield from crossings are:either the effective crossing area can be reduced (i.e., reducethe area of exposed sediment in the cut-and-fill slopes), or wa-ter can be diverted away from the crossing. This can be achieved

by cross ditching on inactive roads, or by spacing culverts so asto divert water onto the forest floor. These solutions can be usedonly if water is not diverted onto potentially unstable slopes.

To learn more about the effectiveness of these options, severalscenarios were investigated using the sediment-budget modelfor the Confluence, for the 12 November 1998 storm (Table12). The sediment budget for the storm as it occurred indicatesthat 48% of the sediment at the source was derived from roads;as delivered to channels, that contribution was reduced to 26%by the valley flat. The reduction of connected ditch length to amaximum of 10 m was more effective at reducing sedimentyield from roads than reduction of the effective area of thesediment sources. However, if a combination of water man-agement and reduction of source areas to a maximum of 30m2/crossing are applied, then the sediment-budget model indi-cates that the relative contribution from road sources is reducedto 17% at the source, or 7% if buffered by a valley flat. Sourcearea reduction would likely take the form of armouring the largerfill slopes and vegetative stabilization of the cut slopes in cross-ings that have large areas of exposed sediment. Water manage-

Figure 25. Evidence of rill erosion on the fill slope at Site 2.

Figure 26. Site 45 after road deactivation. There is evidenceof past rill erosion, but the deactivation has diverted wateraway from the crossing.

29



ment would involve additional culvert placement or cross ditch-ing to reduce delivery of road surface and/or ditch-derived sedi-ment to the crossings.

CONCLUSIONS

A program of sediment-budget research commenced in 1994at Russell Creek, a sub-basin of the Tsitika River watershed onNortheastern Vancouver Island, British Columbia. Part of theprogram aimed to determine the relative contribution of dif-ferent types of sediment sources to the sediment load ofmainstem channel sites in Russell Creek and its main tributary,Stephanie Creek, which are two typical Coastal British Colum-bia streams.

This report presents a rainfall-based, sediment-budget modelthat predicts the sediment yield of Russell Creek and its sub-basins to within +10% for four storms with a range of stormrainfall volumes and intensities. The model is based on the at-tributes of measured sediment sources, including landslides,gullies, and road-related sources. The model did an excellentjob of determining the relative contributions of the different

types of sediment sources to the overall sediment yield fromthe watershed, and also explains the effect of basin morphol-ogy on the sediment budget. The main factors controlling thesediment budget of Russell Creek, in order of importance, arelithology and basin morphology, with forest management ac-tivities playing a minor role.

The model can be used to forecast the effectiveness of for-estry-road-deactivation scenarios on the relative contributionof sediment from forestry roads, and to assess the effect ofexisting sediment sources on the sediment budget of creeks.As such, it is a very valuable management tool, but it requiresfurther research to verify and/or refine it for a broader rangeof storm conditions, and to extend it to other watersheds withdifferent morphological characteristics.

REFERENCES

Beschta, R.L. 1978. �Long-Term Sediment ProductionFollowing Road Construction and Logging in the OregonCoast Range� in Water Resources Research 14(6):1011�1016.

Brown, G.W. and J.T. Krygier. 1971. �Clear-Cut Logging andSediment Production in the Oregon Coast Range� in WaterResources Research 7(5):1189�1198.

Downing, J. 1983. �An Optical Instrument for MonitoringSuspended Particulates in Ocean and Laboratory� inProceedings OCEANS �83. Report No. 83CH1972-9. Instituteof Electrical and Electronics Engineers.

Downing, J. and R. Beach. 1989. �Laboratory Apparatus forCalibrating Optical Suspended Solids Sensors� in MarineGeology 86:243-249.

Draper, N.R. and H. Smith. 1981. Applied Regression Analysis.John Wiley & Sons, Toronto.

Duncan, S.H.; R.F. Bilby; J.W. Ward; and J.T. Heffner. 1987.�Transport of Road-Surface Sediment through EphemeralStream Channels� in Water Resources Bulletin 23(1):113�119.

Frederiksen, R.L. 1970. Erosion and Sedimentation Following RoadConstruction and Timber Harvest on Unstable Soils in Three SmallWestern Oregon Watersheds. Research Paper PNW-104. PacificNorthwest Forest and Range Experiment Station, ForestService, USDA. Portland, Oregon. 15 pp.

Hudson, R.O. 1996. �Use of Automated TurbidityMonitoring and Suspended Sediment Sampling in SedimentBudget Research in the Tsitika River�, paper presented atAutomatic Water Quality Monitoring Workshop,Richmond, BC, February 12�13, 1996.

��. 1999. A Storm-Based Sediment Budget for RussellCreek. Final Report on FRBC Project PA96511-RE: TsitikaRiver Sediment Monitoring Program. Unpublished.Vancouver Forest Region, BC Ministry of Forests,Nanaimo, B.C.

��. 2001. Interpreting Turbidity and Suspended-SedimentMeasurements in High-Energy Streams in Coastal BritishColumbia. Forest Research Technical Report TR-008(Hydrology). Vancouver Forest Region, BC Ministry ofForests. Nanaimo. 14pp.

Table 12. Potential effects of several deactivation/watermanagement scenarios on the contribution of road sourcesto the sediment budget at the Confluence, based on the 12November 1998 storm. The unbuffered yields are the totalyields from road-related sources as measured at the source;the buffered yields are the total mass of road-derived sedi-ment delivered to the channel.

,ecneulfnoC89voN21

oiranecs

dereffubnU dereffuB

dleiY)gk(

% dleiY)gk(

%

gnitsixEnoitarugifnoc

6632 84 834 62

ecudeR.1tnemides

repaeraecruosotgnissorc

xam . m05fo 2

7551 93 113 02

ecudeR.2tnemides

repaeraecruosotgnissorc

m03fo.xam 2

6241 73 782 91

hctidecudeR.3rephtgnelotgnissorc

m01fo.xam

2331 63 632 61

3dna2.4rehtegot

944 71 58 7

30



Nistor, C. and K. Rood. 1999. �Fine-Sediment Productionfrom Gravel-Surfaced Roads in Seymour Watershed� pp.73�82 in Confronting Uncertainty: Managing Change in WaterResources and the Environment. Canadian Water ResourcesAssoc., Conference Proceedings, Richmond, BC, Oct. 27-29, 1999. P. Galvagno, editor.

Paulig, K. 1996. �Russell Creek�Set Up of a StormMonitoring Program� appendix in Hudson, R.O., 1999, AStorm-Based Sediment Budget for Russell Creek. Final Report onFRBC Project PA96511-RE: Tsitika River SedimentMonitoring Program. Unpublished. Vancouver ForestRegion, BC Ministry of Forests, Nanaimo, BC.

Reid, L.M. 1981. Sediment Production from Gravel Surfaced ForestRoads, Clearwater Basin, Washington. Final Report FRI-UW8108. Fisheries Research Institute, University ofWashington. 247pp.

Reid, L. and T. Dunne. 1984. �Sediment Production fromForest Road Surfaces� in Water Resources Research20(11):1753�1761.

Reid, L.M.; T. Dunne; and C.J. Cederholm. 1981. �Applicationof Sediment Budget Studies to the Evaluation of LoggingRoad Impact� in New Zealand Journal of Hydrology 20:49�62.

Rood, K.M. 1984. An Aerial Photograph Inventory of theFrequency and Yield of Mass Wasting on the Queen CharlotteIslands, British Columbia. Land Management Report 34. BCMinistry of Forests, Victoria, BC.

Sterling, S., 1997. Influence of Bedrock Type on the Magnitude,Frequency and Spatial Distribution of Debris Torrents on NorthernVancouver Island. M.Sc. Thesis, Department of Geography,University of British Columbia. Vancouver, BC.

Walling, D.E. and B.W. Webb. 1982. �Sediment Availabilityand the Prediction of Storm-Period Sediment Yields� pp.327�337 in Recent Developments in the Explanation andPrediction of Erosion and Sediment Yield. Proceedings ofExeter Symposium, July 1982. International Assoc. ofHydrological Sciences Publ. #137. Wallingford, Oxon., UK.

Hudson, R.O. and S. Sterling. 1998. �Sediment Production inTwo Northeast Vancouver Island Creeks� pp. 336�347 inMountains to Sea: Human Interaction with the Hydrological Cycle.Canadian Water Resources Association 51st AnnualConference Proceedings, June 1998, Victoria, BC.

Jordan, P. and P. Commandeur. 1998. �Sediment Research inthe West Arm Demonstration Forest, Nelson, BC� pp.348�363 in Mountains to Sea: Human Interaction with theHydrological Cycle. Canadian Water Resources Association51st Annual Conference Proceedings, June 1998, Victoria,BC.

Ludwig, K.A. and D. Hanes. 1990. �A Laboratory Evaluationof Optical Backscatterance Suspended Solids SensorsExposed to Sand-Mud Mixtures� in Marine Geology 94:173-179.

Maynard, D. 1991. Tsitika River Watershed Sediment SourceInventory. Unpublished internal report. Vancouver Region,BC Ministry of Forests. Nanaimo.

Megahan, W.F. and G.L. Ketcheson. 1996. �PredictingDownslope Travel of Granitic Sediments from ForestRoads in Idaho� in Water Resources Bulletin 32(2):371�381.

Megahan, W.F.; J.G. King; and K.A. Seyedbagheri. 1995.�Hydrologic and Erosional Responses of a GraniticWatershed to Helicopter Logging and Broadcast Burning�in Forest Science 41(4):777�795.

Megahan, W.F.; K.A. Seyedbagheri; T.L. Mosko; and G.L.Ketcheson. 1986. �Construction Phase Sediment Budgetfor Forest Roads on Granitic Slopes in Idaho� in Proceedings,Drainage Basin Sediment Delivery. R.F. Hadley, editor.International Assoc. of Hydrological Sciences Publ. #159.Wallingford, Oxon., UK.

Minitab Incorporated. 1995. Minitab Reference Manual Release10Xtra. State College, Pa.

Nistor, C.J. 1996. Temporal Patterns in the Normal-Regime Fine-Sediment Cascade in Russell Creek Basin, Vancouver Island. M.Sc.Thesis. Department of Geography, University of BritishColumbia. Vancouver, BC.

31



APPENDIX A: SEDIMENT SOURCE INVENTORY

Table A1. Gullies.

.onylluG

deBytilibats

1S a

knaBytilibats

2S a htgneL)m(

latoTaera

mk( 2)

deguaGaera

mk( 2)tneidarG

)%(

cigoloeGf noitamro

G b

svdetsevraHdetsevrahnu

H c

detnerroT.rrotnusv

T a

daoRsgnissorc nisab-buS d

)reppu(16 1 0 0021 485.0 235.0 75 1 1 0 4,3,2 C

)rewol(16 1 1 008 .a.n deguagton 02 1 1 1 enon C

26 0 0 0041 062.0 632.0 74 1 1 0 u6,6,5 C

36 0 0 0051 403.0 672.0 74 1 1 0 u8,8,7 C

lanoitidda 0 0 .a.n 045.0 .a.n .a.n 1 1 0 enon C

46 0 0 0002 087.0 467.0 35 1 1 0 01,9 C

56 0 1 0071 420.1 410.1 12 1 1 0 21,11 C

66 0 0 026 961.0 .a.n 16 1 1 0 31 RU

76 0 0 0431 016.0 .a.n 93 1 0 0 41 RU

86 0 0 008 192.0 .a.n 01 1 2 0 51 RU

96 1 1 048 603.0 .a.n 46 1 1 1 81 RU

07 0 0 0411 915.0 .a.n 04 1 1 0 12 RU

17 0 0 0861 216.0 .a.n 56 1 0 0 32 RU

27 0 0 047 101.0 .a.n 83 1 0 0 42 RU

37 0 0 047 101.0 .a.n 34 1 2 0 52 RU

47 1 1 0001 825.0 612.0 54 0 2 1 04,93,03-62 C

lanoitidda 1 0 .a.n 056.0 .a.n .a.n 0 2 0 63-13 C

57 0 0 0052 003.1 272.1 52 0 0 0 54 S

67 1 1 0022 959.0 049.0 53 1 0 1 74,64 S

77 0 0 045 401.0 401.0 84 1 2 0 75,65 S

87 0 0 0001 281.0 .a.n 45 1 1 0 55 SU

97 0 0 0021 812.0 .a.n 34 1 0 0 45 SU

08 0 0 063 660.0 .a.n 55 1 1 0 85 SU

18 0 0 047 202.0 .a.n 45 1 1 0 05,15 SU

28 0 0 047 202.0 .a.n 15 1 1 0 25 SU

38 0 0 006 461.0 .a.n 35 1 1 0 94,35 SU

48 0 0 0022 469.0 .a.n 56 0 0 0 enon RL

58 0 0 0821 083.0 .a.n 52 0 0 0 34 RL

68 0 0 0082 402.2 .a.n 92 0 0 0 44 RL

78 0 0 0011 002.0 .a.n 04 1 0 0 45 SU

a .elbatsnu=1.elbats=0 b .kcordebcitinargrevodepardlairetamlaicifruscitlasabrocitlasab=1.citinarg=0 c yllaitrap=1.detsevrahnu=0.detsevrahyletelpmoc=2.detsevrah d .einahpetSreppU=SU.einahpetS=S.llessuRrewoL=RL.llessuRreppU=RU.ecneulfnoC=C

33



Table A2. Crossing attributes.

continued

.onetiSgnissorC

epyt a

daoRtneidarg

GR b

maertStneidarg

GS)%(

htgneL.cossafo

daorL

)m(

cigoloeGnoitamrof

G c

ytilibatSgnissorc

S d

saeraecruoS

hctiDepolstuc

m( 2)

hctiDhtgnel

)m(.veleetiS

).l.s.am( e

epolslliFm( 2)

epolstuCm( 2)

1 1 1 6 006 0 1 gnissorcegdirb .m.n .m.n 562

34 3 3 03 044 0 1 51 0 .m.n 231

44 3 1 52 086 0 1 51 0 .m.n 751

ecneulfnoC

2 2 2 03 063 1 4 492 003 054 531 066

3 3 3 72 006 1 4 86 61 822 251 065

4 3 1 51 083 1 4 91 0 .m.n 011 005

5 3 1 51 002 1 1 0 23 0 05 005

6 3 1 72 021 1 1 021 24 0 46 065

reppu6 .a.n .a.n .a.n .a.n 1 .a.n 22 53 004 001 066

7 3 1 51 021 1 1 23 42 0 01 005

8 3 1 72 04 1 1 801 661 .m.n 0 065

reppu8 .a.n .a.n .a.n .a.n 1 .a.n 001 0 432 45 046

9 3 1 51 06 1 2 8 8 47 47 015

62 .a.n 1 45 .a.n 0 .a.n .m.n .m.n .m.n .m.n

72 2 1 03 08 0 3 53 69 04 02 065

82 5 1 55 06 0 4 05 8 58 71 065

92 3 1 81 082 0 1 51 0 .m.n 601 015

03 3 1 81 004 0 1 51 0 .m.n 721 084

13 3 1 02 007 0 1 51 0 .m.n 951 084

23 3 1 02 022 0 1 51 0 .m.n 29 015

33 4 1 13 001 0 3 03 91 .m.n 64 065

43 3 4 05 005 0 3 51 0 .m.n 041 006

53 5 3 45 07 0 5 08 02 0362 62 43evobA

63 4 3 45 08 0 3 03 07 113 13 43evobA

73 4 3 45 08 0 3 03 07 471 71 43evobA

83 4 3 45 08 0 3 03 07 471 71 72evobA

93 4 3 45 08 0 3 03 07 471 71 72evobA

04 4 3 45 08 0 3 03 07 471 71 72evobA

34



Table A2, continued:

.onetiSgnissorC

epyt a

daoRtneidarg

GR b

maertStneidarg

GS)%(

htgneL.cossafo

daorL

)m(

cigoloeGnoitamrof

G c

ytilibatSgnissorc

S d

saeraecruoS

hctiDepolstuc

m( 2)

hctiDhtgnel

)m(.veleetiS

).l.s.am( e

epolslliFm( 2)

epolstuCm( 2)

llessuRreppU f

01 3 2 72 58 1 2 465 0 23 55 045

11 3 2 72 06 1 2 441 6 4 2 045

21 2 1 51 027 1 2 3 02 0 01 015

einahpetS

14 5 4 02 002 0 4 801 661 .m.n 68 016

24 1 1 81 084 0 1 03 0 .m.n 731

54 2 3 52 021 0 1 192 0 069 021 067

64 5 4 52 001 1 5 0 0 008 001 067

74 2 5 52 004 1 2 61 84 0 002 067

84 2 5 81 002 1 1 03 0 .m.n 68 SU

94 4 1 81 04 1 3 03 0 .m.n 0 SU

05 4 1 81 001 1 3 03 0 .m.n 64 SU

15 4 3 81 06 1 3 03 0 .m.n 71 SU

25 4 3 81 001 1 3 03 0 .m.n 64 SU

35 4 3 81 003 1 3 03 0 .m.n 011 SU

45 3 1 53 05 1 3 03 0 .m.n 6 SU

55 4 1 43 001 1 3 03 0 .m.n 64 SU

65 4 3 54 061 1 2 03 0 .m.n 47 009

75 4 4 54 025 1 2 03 0 .m.n 241 0001

85 5 5 81 021 1 5 03 0 .m.n 75 SU

a .tuohsaw=5.hctidssorc=4.trevlucepip=3.trevlucgol=2.egdirb=1 b .%02=5.%9.91-51=4.%9.41-01=3.%9.9-5=2.%9.4-0=1c .kcordebcitinargrevodepardlairetamlaicifruscitlasabrocitlasab=1.citinarg=0 d ehtfoytilibatsllarevoehtfoerusaemevitatilauqasiS

ehtfoeruliaflatotfoecnedive(elbatsnutsom=5dna)noisorefoecnediveon,tneidargwol(elbatstsom=1erehw,sepolsllif-dna-tucgnissorc.gnissorcehtdnuoradnani,noisorellirgnidulcni,noisorefoseergedevah4ot1seirogetaC.)gnissorc e .deriuqerton=knalB f lanoitiddA

.detsiltonllessuRreppUnignissorc

35



Table A3. Landslides and other slope failures.

continued

ecruoS.on a epyT b ytivitcA c tcennoC d htgneL

)m(htdiW

)m(aerA

m( 2)tneidarG

)%(

cigololoeGnoitamrof

G e

detsevraH.vrahnusv

H f

llessuRrewoL

311 3L 2 6 01 4 04 001 0 0

411 3L 2 6 01 4 04 001 0 0

511 2L 2 6 51 01 051 57 0 0

611 1L 2 6 01 01 001 07 0 0

811 1L 2 6 01 5 05 001 0 0

021 2L 2 6 03 8 042 57 0 0

381 1L 2 6 5 4 02 57 0 0

481 1L 2 6 01 5 05 57 0 0

412 2L 2 6 52 8 002 57 0 0

711 3L 1 6 002 5 0001 051 0 0

911 2L 1 6 02 3 06 06 0 0

121 2L 1 6 01 5 05 08 0 0

512 2L 1 6 51 2 03 07 0 0

09 1L 2 7 51 4 06 001 0 1

19 1L 2 6 03 6 081 001 0 1

69 1L 2 6 01 01 001 001 0 1

79 1L 2 6 01 01 001 001 0 1

89 2L 2 6 01 51 051 001 0 1

101 1L 2 6 51 01 051 011 0 1

211 2L 2 6 001 7 007 001 0 1

401 1L 1 6 5 4 02 001 0 1

011 1L 1 6 51 4 06 06 0 1

001 1L 1 5 6 4 42 001 0 1

501 3L 1 5 06 5 003 011 0 1

221 1L 1 5 5 4 02 07 0 1

)+2ytivitca(evitcadnadetcennocaeralatoT 0332

detcennoctoN

59 1L 2 4 52 01 052 001 0 1

601 1L 1 4 53 01 053 001 0 1

39 1L 1 1 5 7 53 57 0 1

49 1L 1 1 01 01 001 07 0 1

301 1L 1 1 01 3 03 001 0 1

36




continued


)m(htdiW

)m(aerA

m( 2)tneidarG

)%(

cigololoeGnoitamrof

G e

detsevraH.vrahnusv

H f

einahpetS

781 1L 2 7 01 01 001 001 0 0

191 2L 3 7 002 5.1 003 05 1 0

291 2L 3 7 002 5.1 003 05 1 0

391 2L 3 7 002 5.1 003 05 1 0

491 2L 3 7 002 5.1 003 05 1 0

891 1L 2 7 02 51 003 001 1 0

991 1L 2 7 08 52 0002 001 1 0

002 1L 2 7 01 4 04 001 1 0

102 1L 2 7 05 51 057 001 1 0

202 1L 2 6 02 8 061 57 1 0

302 1L 2 6 51 01 051 57 1 0

402 1L 2 6 02 6 021 57 1 1

502 1L 2 7 51 5 57 001 1 1

112 1L 2 5 01 52 052 57 0 1

602 1L 3 7 81 9 261 001 1 1

detcennoctoN

091 3L 2 1 7 2 41 57 0 1

902 3L 2 3 042 002 00084 001 1 0

012 3L 2 1 003 06 00081 001 1 0

212 1L 7 3 007 01 0007 54 0 0

312 3L 1 1 006 041 00048 05 1 1

llessuRreppU

271 1L 1 7 081 06 00801 001 1 0

471 3L 1 7 021 061 00291 001 1 0

731 1L 2 5 51 5 57 001 1 0

831 1L 2 5 52 7 571 001 1 0

751 2L 3 5 51 5 57 001 1 1

061 1L 1 5 5 4 02 07 1 1

971 1L 1 5 02 3 06 57 1 0

37





)m(htdiW

)m(aerA

m( 2)tneidarG

)%(

cigololoeGnoitamrof

G e

detsevraH.vrahnusv

H f

ecneulfnoC

131 C 1 5 54 3 531 001 1 1

231 3L 1 5 51 3 54 57 1 1

331 3L 1 5 01 2 02 57 1 1

141 2L 6 5 572 3 528 53 1 1

241 C 2 6 021 3 063 001 1 1

341 C 2 6 56 3 591 001 1 1

281 1B 9 6 51 2 03 5 0 0

detcennoctoN

171 2L 6 3 009 3 0072 06 1 1

181 1L 7 2 044 02 0088 45 1 1

561 1L 1 1 003 03 0009 001 1 0

571 1L 1 1 022 02 0044 001 1 0

671 3L 1 1 061 061 00652 001 1 0

881 3L 2 1 01 31 031 57 0 1

a .ecruostnacifingisaderedisnocebot,2tsaeltafolevelytivitcanadna,5tsaeltafoyitilbareviledaevahtsumecruostnemidesAb .eruliafknablennahc=1B.knabtuc=C.gnilevar=3L.wolfsirbed=2L.edilssirbed=1L c =3.gnilevaretaredom=2.gnilevarwol=1

laretal=9.knislanoitisoped=8.peedmc001>,ruocs=7.peedmc05-52,ruocs=6.peedmc52-01,ruocs=5.gnilevarfoeergedhgih.noisore d niamro,yratubirt,yllugotytilibareviled=7ot5.)krowtenlennahcotdetcennocton,.e.i(nialpdoolfotetisnoytilibareviled=4ot1

.)detcennoc,.e.i(lennahc e .kcordebcitinargrevodepardlairetamlaicifruscitlasabrocitlasab=1.citinarg=0 f rocitlasab=1.citinarg=0.kcordebcitinargrevodepardlairetamlaicifruscitlasab

38



APPENDIX B: FLOW RATING AND VERIFICATION

Flow rating was done by a combination of salt dilution gaugingand current metering at mainstem gauging sites. At manuallysampled sediment source sites, salt dilution was used, but therating curves were extended at the high end by surveying thechannel, and calculating flows using Manning�s formula. Valuesfor Manning�s n were estimated from flow measurements atlower stages. The rating curves were used to derive streamflowsfrom stage readings. For example, the rating curve for gullyG75 at Site 45 is an exponential curve that is based on a combi-nation of salt dilution gauging, supplemented with survey dataat the upper end (Figure B1).

Flow records for each gully were reconstructed from point data,based on relationships with a reference site where stage wasmonitored continuously. For example, manually collected stagereadings at Site 45 during the 12 November 12 1998 storm cor-related best with continuous stage readings at Site 3, but weredifferent for rising and falling stage at Site 3 (Figure B2). Stagewas still rising at Site 45 while at Site 3, it was falling. There ispotential for substantial error in the reconstruction of flowrecords using these methods. In particular, the exact value andtime of the peak stage at Site 45 was not known. Calculation ofsuspended sediment yield depends on accurate SSC and flowdata. Thus, an independent method of verifying the flow datawas needed.

The rising and recession characteristics of gullies differ fromthose of the mainstem stream sites. Because gullies are steeperthan mainstem sites, are of lower stream order, and have smallerareas, they tend to be more flashy mainstem channels. For thesereasons, the ratio of instantaneous flow in a gully to that in themainstem stream is constantly changing. However, the ratio oftotal flow volume in a gully to total flow in the mainstem streamshould be related to the ratio of the area of the gully to thedrainage area at the mainstem gauge.

For each storm, the total volume of flow at each site was calcu-lated. For each gully, the ratio of the flow volume to the vol-ume of flow at the reference site used to reconstruct flow recordwas then calculated (Table B1). The ratio of the gully area tothe drainage area at the reference site was also calculated. Insome cases, adjustments had to be made in order to produceconsistent relationship between flow volume ratios and arearatios. Following the example from gully G75 at Site 45, adjust-ments were made to the timing and magnitude of peak flowsaround the transition between rising and falling stage (FigureB1) to bring the flow volume ratio in line with the flow ratio vsarea ratio relationship (Figure B3).

The resultant relationship between flow volume ratios and arearatios is consistent from storm to storm, and from site to site(Figure B3). Variability in the relationship is likely due to vari-ability in precipitation distribution.

Figure B1. Exponential rating curve for gully G75 at Site 45.

Figure B2. Relationships used to determine stage at Site45 for the 12 November 1998 storm, based on continuouslymonitored stage at Site 3.

0.20 0.30 0.40S tage (m )

0.00

0.20

0.40

0.60

0.80

1.00

1.20

1.40

Str

eam

flow

(Q

, m3/s

)

R ating C urveG ully 75 at S ite 45

0.00 0.10 0.20 0.30 0.40S tage at S ite 3 (m )

0.15

0.20

0.25

0.30

0.35

0.40

Sta

ge

at

Site

45

(m

)

R is ingF alling

Figure B3. Storm flow volume ratios vs area ratios, for gulliescompared to reference sites. The upper points are for upperStephanie Creek. The fact that the relationship between flowratios and area ratios is linear, and is consistent from stormto storm, indicates that the flow data are accurate.

0.0 0.1 0.2 0.3 0.4 0.5R atio o f G ully A rea to R eference G auge C atchm ent A rea

0.0

0.1

0.2

0.3

0.4

0.5

Ra

tio o

f G

ully

Flo

w V

olu

me

to

R

efe

ren

ce G

au

ge

Flo

w V

olu

me

O ctober 1, 1997N ovem ber 12, 1998N ovem ber 14, 1998N ovem ber 20, 1998

39



Table B1. Calculation of flow volume and area ratios.

etadmrotS emulovwolFm( 3)

aeradeguaGmk( 2)

soitaraeradnawolF

ecnerefeReguag a

keerCllessuRfooitaR eguagecnereferfooitaR

wolF aerA wolF aerA

7991tcO10

llessuR 5148201 088.03

ecneulfnoC 709437 820.61 517.0 915.0

einahpetS 5.746612 216.8 112.0 972.0

einahpetSreppU 048701 254.3 501.0 211.0 894.0 104.0 S

2etiS 21011 235.0 110.0 710.0 510.0 330.0 C

5etiS 6791 632.0 200.0 800.0 300.0 510.0 C

01etiS 5.83503 467.0 030.0 520.0 240.0 840.0 C

21etiS 88524 410.1 140.0 330.0 850.0 360.0 C

43etiS 877.1

54etiS 3.55472 272.1 720.0 140.0 721.0 841.0 SU

64etiS 5.70902 049.0 020.0 030.0 790.0 901.0 SU

89voN21

llessuR 0.9294421 88.03

ecneulfnoC 7.261666 820.61 535.0 915.0 535.0 915.0

einahpetS 0.857963 216.8 792.0 972.0

einahpetSreppU 4.291911 254.3 690.0 211.0 223.0 104.0 S

2etiS 1.20761 235.0 310.0 710.0 520.0 330.0 C

5etiS 6.6143 632.0 300.0 800.0 500.0 510.0 C

01etiS 1.85872 467.0 220.0 520.0 240.0 840.0 C

21etiS 1.01904 410.1 330.0 330.0 160.0 360.0 C

BylluG 7.04071 612.0 410.0 700.0 620.0 310.0 C

43etiS 4.5506 802.0 900.0 310.0 C

54etiS 8.50915 272.1 240.0 140.0 041.0 841.0 S

64etiS 7.72914 049.0 430.0 030.0 311.0 901.0 S

89voN41

llessuR 684365 88.03

ecneulfnoC 868333 820.61 395.0 915.0 395.0 915.0

einahpetS 1.912941 216.8 562.0 972.0

einahpetSreppU 0.72654 254.3 180.0 211.0 603.0 104.0 S

BylluG 6.6753 612.0 600.0 700.0 420.0 310.0 C

64etiS 4.72061 049.0 820.0 030.0 701.0 901.0 S

89voN02

llessuR 6411551 88.03

ecneulfnoC 884588 820.61 175.0 915.0 175.0 915.0

2etiS 1.50221 235.0 800.0 710.0 410.0 330.0 C

5etiS 2.4252 632.0 200.0 800.0 300.0 510.0 C

01etiS 4.70413 467.0 020.0 520.0 530.0 840.0 C

21etiS 3.59634 410.1 820.0 330.0 940.0 360.0 Ca .einahpetSreppU=SU.einahpetS=S.ecneulfnoC=C

40

forest research technical report - british columbia...forest research vancouver forest region 2100...

Documents