Carnation Creek and Queen Charlotte Islands

Fish/Forestry Workshop: Applying 20 Years

of Coast Research to Management Solutions

Ministry of ForestsResearch Program

Dan L. Hogan, Peter J. Tschaplinski,

and Stephen Chatwin

(editors)

Canadian Cataloguing in Publication DataCarnation Creek and Queen Charlotte Island

Fish/Forestry Workshop (1994 : Queen CharlotteCity, B.C.)

Carnation Creek and Queen Charlotte IslandFish/Forestry Workshop : applying 20 years of coastresearch to management solutions

(Land management handbook ; 41)

ISBN 0-7726-3510-2

1. Fish habitat improvement – British Columbia –Carnation Creek Region – Congresses. 2. Habitat(Ecology) – British Columbia – Carnation CreekRegion – Management – Congresses. 3. Forestmanagement – Environmental aspects – BritishColumbia – Carnation Creek Region – Congresses.I. Hogan, Daniel Lewis, 1954– . II. Tschaplinski,Peter John, 1953– . III. Chatwin, Stephen C.IV. British Columbia. Ministry of Forests. ResearchBranch. V. Series.

SH173.C36 1998 639.9’77’097112 C98-960079-3

CitationHogan, D.L., P.J. Tschaplinski, and S. Chatwin (Editors). 1998. B.C. Min. For., Res. Br., Victoria, B.C.Land Manage. Handb. No. 41.

Prepared byD.L. Hogan,P.J. Tschaplinski andS. Chatwin (editors)forB.C. Ministry of ForestsResearch Branch31 Bastion SquareVictoria, BC

© 1998 Province of British Columbia

Copies of this and other Ministry of Foreststitles are available from:Crown Publications Inc.521 Fort StreetVictoria, BC

Ministry of ForestsPublication Internet Catalogue: www.for.gov.bc.ca/hfd

LIST OF CONTRIBUTORS

Name Address

J.M.E. Balke Lacon Road, Denman Island, BC

William J. Beese MacMillan Bloedel Limited, Front Street,Nanaimo, BC

Stephen A. Bird Pacific Watershed Research Association,– West th Avenue, Vancouver, BC

M.J. Bovis Department of Geography, University of BritishColumbia, Vancouver, BC

Tom G. Brown Department of Fisheries and Oceans, Pacific BiologicalStation, Nanaimo, BC

Michael Brownlee Integrated Resources Branch, B.C. Ministry of Forests,Victoria, BC

Anthony L. Cheong B.C. Ministry of Environment, Lands and Parks,Fisheries Branch, nd Floor, Blanshard Street,Victoria, BC

Michael Church Department of Geography, University of BritishColumbia, Vancouver, BC

S.J. Crockford Pacific Identifications, Nelthorpe Street,Victoria, BC

James E. Doyle Mt. Baker Snoqualmie N.F., U.S. Forest Service, th Avenue West, Mountlake Terrace, WA

R.J. Fannin University of British Columbia, Faculty of Forestry,Vancouver, BC

Darren Ham University of British Columbia, Vancouver, BC

Gordon F. Hartman Fisheries Research and Education Services, Rose Ann Drive, Nanaimo, BC

Judith K. Haschenburger Department of Geography, University of BritishColumbia, Vancouver, BC

Eugene D. Hetherington E.D. Hetherington and Associates Ltd., DunnettCrescent, Victoria, BC (formerly ResearchHydrologist with the Canadian Forest Service,Pacific Forestry Centre, Victoria, BC)

Dan Hogan B.C. Ministry of Forests, Research Branch, PO Box

Stn Prov Govt, Victoria, BC

L.B. Holtby Department of Fisheries and Oceans, Pacific BiologicalStation, Nanaimo, BC Canada

Josh Korman West th Avenue, Vancouver, BC

Ray Krag Group Supervisor, Harvest Engineering,Forest Engineering Research, Institute of Canada, East Mall, Vancouver, BC

iii

iv

Werner Kurz ESSA Technologies Ltd., ‒ West th Avenue,Vancouver, BC

C. Peter Lewis B.C. Ministry of Environment, Lands and Parks,Victoria, BC

J. Stevenson Macdonald Fisheries and Oceans Canada, Department ofResources and Environmental Sciences, Simon FraserUniversity, Burnaby, BC

D. Marmorek ESSA Technologies Ltd., ‒ West th Avenue,Vancouver, BC

T.H. Millard B.C. Ministry of Forests, Labieux Road,Nanaimo, BC

Greta Movassaghi Mt. Baker Snoqualmie N.F., U.S. Forest Service, th Avenue West, Mountlake Terrace, WA

Roger Nichols Mt. Baker Snoqualmie N.F., U.S. Forest Service, th Avenue West, Mountlake Terrace, WA

M.E. Oden Madrone Consultants Ltd., Herd Road, RR#,Duncan, BC

Ian Parnell ESSA Technologies Ltd., ‒ West th Avenue,Vancouver, BC

Stephen Rice University of British Columbia, Department ofGeography, Vancouver, BC

T.P. Rollerson B.C. Ministry of Forests, Labieux Road,Nanaimo, BC

Jim W. Schwab B.C. Forest Service, Forest Sciences Section,PO Box , Smithers, BC

J. Charles Scrivener Consultant Biologist, Rutherford Road,Nanaimo, BC

G. Suther Ecofocus Environmental Consultants, LaconRoad, Denman Island, BC

B. Thomson B.C. Ministry of Environment, Lands and Parks,–nd Street, Surrey, BC

Derek B. Tripp Tripp Biological Consultants Ltd., Extension Road,Nanaimo, BC

Peter J. Tschaplinski B.C. Ministry of Forests, Research Branch, PO Box

Stn Prov Govt, Victoria, BC

Tim Webb ESSA Technologies Ltd., ‒ West th Avenue,Vancouver, BC

David Wilford B.C. Ministry of Forests, Prince Rupert Forest Region,Forest Sciences Section, Bag ,Smithers, BC

Robert P. Willington Integrated Resource Analysis Section, TimberWestLimited, PO Box , Crofton, BC

M.P. Wise International Forest Products, Dunsmuir Street,Vancouver, BC

Michael Z’Graggen East th Avenue, Vancouver, BC

CONTENTS

List of Contributors ....................................................................................................................................................... iii

Introductory Comments for FFIP/Carnation Creek WorkshopDavid Wilford ............................................................................................................................................................... 1

Introduction to Day 1: Focus on ResearchMichael Brownlee ......................................................................................................................................................... 3

Introduction: Workshop Outline and Experimental DesignC. Peter Lewis ................................................................................................................................................................ 5

The Landscape of the Pacific NorthwestMichael Church ............................................................................................................................................................ 13

An Introduction to the Ecological Complexity of Salmonid Life History Strategies andof Forest Harvesting Impacts in Coastal British Columbia

J. Charles Scrivener, Peter J. Tschaplinski, and J. Stevenson Macdonald .............................................................. 23

Focus on Forestry-fisheries Problems: Lessons Learned from Reviewing Applicationsof the Coastal Fisheries-Forestry Guidelines

D. Tripp and D. Hogan ................................................................................................................................................ 29

Watershed HydrologyEugene D. Hetherington .............................................................................................................................................. 33

Landslides on the Queen Charlotte Islands: Processes, Rates, and Climatic EventsJim W. Schwab ............................................................................................................................................................... 41

Gully Processes in Coastal British Columbia: The Role of Woody DebrisM.J. Bovis, T.H. Millard, and M.E. Oden .................................................................................................................. 49

Stream Channel Morphology and Recovery ProcessesD. L. Hogan, S. A. Bird, and S. Rice ............................................................................................................................ 77

Evolution of Fish Habitat Structure and Diversity at Log Jamsin Logged and Unlogged Streams Subject to Mass Wasting

Derek Tripp ................................................................................................................................................................... 97

Channel Scour and Fill in Coastal StreamsJudith K. Haschenburger ............................................................................................................................................. 109

Fine Sediments in Small Streams in Coastal British Columbia: A Review of Research ProgressMichael Church ............................................................................................................................................................ 119

Changes of Spawning Gravel Characteristics after Forest Harvesting in Queen Charlotte Islands andCarnation Creek Watersheds and the Apparent Impacts on Incubating Salmonid Eggs

J. Charles Scrivener and Derek B. Tripp .................................................................................................................... 135

Overwintering Habitats and Survival of Juvenile Salmonids in Coastal Streams of British ColumbiaGordon F. Hartman, Derek B. Tripp, and Tom G. Brown ...................................................................................... 141

v

vi

Long-term Patterns in the Abundance of Carnation Creek Salmon, and the Effects of Logging,Climate Variation, and Fishing on Adult Returns

Peter J. Tschaplinski, J. Charles Scrivener, and L.B. Holtby .................................................................................... 155

Watershed Hydrology: Forest Management ImplicationsRobert P. Willington ..................................................................................................................................................... 181

Gully Assessment MethodsD.L. Hogan and T.H. Millard ...................................................................................................................................... 183

Classification and Assessment of Small Coastal Stream ChannelsD.L. Hogan and S.A. Bird ............................................................................................................................................ 189

Productivities, Costs, and Site and Stand Impacts of Helicopter-logging in Clearcuts, Patch Cuts,and Single-tree Selection Cuts: Rennell Sound Trials

Ray Krag ......................................................................................................................................................................... 201

Ten Years of Watershed Restoration in Deer Creek, Northwest Cascades of Washington StateJames E. Doyle, Greta Movassaghi, and Roger Nichols ........................................................................................... 215

The Fish/Forestry Interaction Program Simulation Model (FFIPS)D. Marmorek, Ian Parnell, Tim Webb, Michael Z’Graggen, Werner Kurz, and Josh Korman ........................... 231

Problems, Prescriptions, and Compliance with the Coastal Fisheries-Forestry Guidelinesin a Random Sample of Cutblocks in Coastal British Columbia

Derek Tripp ................................................................................................................................................................... 245

POSTERS

The Spatial Variation and Routine Sampling of Spawning Gravels in Small Coastal StreamsStephen Rice .................................................................................................................................................................. 257

Debris Avalanches-flows on British Columbia’s North CoastJim W. Schwab ............................................................................................................................................................... 259

Landslide Runout Behaviour in the Queen Charlotte IslandsR.J. Fannin, M.P. Wise, and T.P. Rollerson ................................................................................................................ 261

Landslide Reforestation and Erosion Control in the Queen Charlotte IslandsWilliam J. Beese ............................................................................................................................................................. 263

River Otter Predation on Juvenile Salmonids in Winter: Preliminary Report ofRiver Otter Scat Collection and Diet Analysis

J.M.E. Balke, P.J. Tschaplinski, S.J. Crockford, and G. Suther ................................................................................. 265

Applications of Photography in Geomorphology: Size Scales and Appropriate PlatformsDarren Ham and Dan Hogan .....................................................................................................................................267

Terrain Attribute Study: Slope Failure Frequencies Following Logging in Coastal British ColumbiaB. Thomson ................................................................................................................................................................... 271

Quantifying Basin Comparisons in the Queen Charlotte IslandsAnthony L. Cheong ...................................................................................................................................................... 273

Riparian Area Response to the Development of a Lateral Sediment WedgeStephen A. Bird ............................................................................................................................................................. 275

49

Introduction

The Gully Management Problem in Coastal BritishColumbia Large tracts of forested terrain in coastalBritish Columbia are dissected by gullies. Thesesteep channels are typically less than 1 km in lengthand 3–30 m in depth, and have a V-shaped cross-sectional form. Because of their steepness and insta-bility, gullies are important sources of both sedimentand large woody debris (LWD) for downstreamareas in coastal British Columbia (Chatwin et al.1994). Much of this material is delivered by debrisflows, triggered by relatively small debris slides onsteep, unstable gully walls.

It is now widely acknowledged that loggingactivities have greatly increased the delivery rate ofboth sediment and LWD from gullies, principally byincreasing the frequency and magnitude of gullydebris flows. The result in many cases has beenunacceptably high debris loadings to downstreamareas (Wilford and Schwab 1983; Rood 1984;Roberts and Church 1986). The need for betterforest management practices in gullies prompted theB.C. Ministry of Forests to develop the GullyAssessment Procedures (Hogan and Millard, thisvolume). The procedures are designed to assess thelikelihood of debris slides, debris flows, and fluvialtransport of sediment and woody debris occurring,and recommend the best practicable strategies tominimize gully instability. It is generally recom-mended that excess woody debris produced bylogging operations be removed if the potential foreither water transport of debris or debris flow isdeemed significant. Indeed, post-harvest clearance ofLWD is now common practice in coastal gullies.However, despite the acknowledged importance ofwoody debris in the sediment dynamics of gullies,there are very few quantitative data on the real-timein-gully interactions between sediment and coarsewoody debris. This study reports data on the effects

of woody debris in both logged and unlogged gullychannels, including observations on the effects ofwoody debris removal from gully channels followingtimber harvest.

Overview of Processes in Forested Gullies Gullieshave long been recognized as distinct landformsduring routine terrain mapping in British Columbia(Howes and Kenk 1988), but it is only in the past15 years that their geomorphic significance havebeen fully appreciated (Wilford and Schwab 1983;Rood 1984, 1990; Krag et al. 1986; Howes 1987;Buchanan and Savigny 1990; Millard 1993; Chatwinet al. 1994; Oden 1994). Some of this awarenessstems from work conducted earlier in forestedterrain in other parts of the Western Cordillera(e.g., Swanston and Swanson 1976; Dietrich andDunne 1978; Swanson et al. 1982).

Gullies combine features of both hillslopes andstream channels, and therefore a wide variety ofprocesses tend to occur within them. Hillslopeprocesses include debris slides and debris flows, creepand ravel, and significant fluvial transport of sedi-ment and LWD. All of these processes are affected bytimber harvesting to varying degrees. The rate ofsupply of LWD is usually greatly increased by thebreakage of trees during tree falling, and by trimmingand log-bucking. This debris slides down steep gullysidewalls and becomes concentrated along the gullychannel (Fig. 1). Coupled with increased LWDproduction is a greater sediment supply by debrisslides from steep gully walls. These small landslidesincrease in frequency following harvest because ofyarding disturbance to sidewalls and through rootdecay leading to loss of soil strength over time (Sidleet al. 1985). Debris-slide scars, in turn, promoteaccelerated fine sediment production by surfaceerosion and ravelling. In summary, an accelerationin the supply rate of both LWD and sediment tendsto occur following harvest.

Gully Processes in Coastal British Columbia: The Role of Woody Debris

M.J. B, T.H. M, M.E. O

50

Typical logged gullies in lower MacmillanCreek, north Moresby Island. Gullies areclassed as “slash-full,” but have not yetproduced debris flows.

Much of this increased debris load is stored alongthe gully channel, since water flow depths in gulliesare probably less than 0.2 m. Large pieces of woodydebris retard the flow and tend to promote sedimentdeposition. Although some reworking of the LWDand sediment load stored along a gully channel canoccur by fluvial transport during periods of highrunoff, only a debris flow is capable of removing allof the stored material as a single catastrophic event.Most gully debris flows are triggered by debris slidesthat start on steep gully walls during winter rain-storms (Krag et al. 1986; Fannin and Rollerson 1993).Debris flows usually run the full length of a gully,since channels are steep (15–30°) as well as confined.The total volume of material moved by a debris flowusually depends more on entrainment of materialfrom a gully channel than on the volume of theoriginal debris slide triggering the flow. This fact wasfirst widely publicized by Swanston and Swanson(1976) in the context of forested gullies. Because ofsediment entrainment along the gully channel, asmall debris slide of about 100 m3 can produce adebris flow totalling several thousand cubic metres(Fannin and Rollerson 1993). Entrainment of thesurcharge debris load produced by logging operationsaccounts for the large volume of many debris flowsthat scour the clearcut sections of gullies.

After the passage of a debris flow, a gully channelis usually scoured to bedrock or to less erodiblematerial such as basal till, and the process of debris

recharge must begin anew. Sediment supply to agully is usually large directly following a debris flow,since the sidewalls are scoured and undercut by thepassage of the flow, which increases the area of baresoil. Since the gully is also temporarily devoid ofwoody debris, finer sediment derived from erosionof scoured gully walls is readily removed by fluvialtransport. Eventually, the input of fresh LWD createsobstructions along the channel, which become newsediment storage sites. Woody debris is therefore animportant regulator of sediment movement andstorage, and is one of the main factors controllingthe magnitude of future debris flows. For thisreason, the control of LWD in gullies, both duringand after timber harvest, is an important issuehaving long-term implications for sedimentmanagement in coastal British Columbia. Gullyprocesses were the central focus of this study.

Study Objectives To evaluate the possible benefitsof LWD removal from gullies following timberharvest, data on the rates of sediment production,storage, and delivery were gathered from gullies inboth logged and unlogged terrain. This study reportsresults from two separate investigations conductedin the period 1990–1993.

The process study was designed to measure real-time differences in sediment production, storage,and delivery in both logged and unlogged gullies.Gullies were studied in two geologically distinctenvironments—the Queen Charlotte Islands and thesouthern Coast Mountains—to assess the influenceof logging and terrain factors in gully processes. Themain objectives of the process study were to:1. compare the temporal patterns of sediment

storage and discharge in gullies fully loaded withlogging debris, with those in gullies cleared oflogging debris following harvest;

2. make recommendations concerning various stra-tegies for LWD management in logged gullies; and

3. investigate variations in sediment output fromgullies over a range of geologic conditions.

Examples of the main types of gullies investigatedin the process study are illustrated in Figure 2.

A second component, the synoptic study, wasdesigned to complement the process study by inves-tigating differences in debris recharge rates betweenlogged and unlogged gullies over periods of several

51

Photographs of representative gullies in the three main treatment groups:(a) Logged slash-full gully M2 (left) and logged torrented gully M1 (right), in upper Macmillan Creek. Note the

contrast in woody debris loadings and the highly scoured sidewalls of the recently torrented gully. (b) Logged slash-clear gully C10, Coquitlam basin. Note that channel is scoured clear of fine sediment following

removal of woody debris.(c) Unlogged, old-growth gully in Gregory Creek, Rennell Sound area showing typical large woody debris

accumulation.

decades. Gullies were studied in the Rennell Soundarea of southwestern Graham Island, QueenCharlotte Islands. The main objectives of thesynoptic study were to:1. estimate the volumes of debris stored in logged

and unlogged gullies in the period since the lastdocumented debris flow;

2. estimate from (1) the recharge rates of clastic andwoody debris in logged and unlogged gullies overperiods of several decades; and

3. consider the effects of converting old-growth tosecond-growth forests on debris recharge ratesin gullies.

Study Areas

Process Study Field Areas Study areas were selectedin the Queen Charlotte Islands and the southernCoast Mountains to ensure that research findingscould be extended to a large area of coastal BritishColumbia. Macmillan and Deena creek basins on

2a. 2b. 2c.

north Moresby Island were considered typical of theweak sedimentary and volcanic rocks that underlielarge areas of gullied terrain on the Queen CharlotteIslands (Fig. 3). Macmillan Creek is a 6.2-km2 basincontaining more than 20 steep gullies dischargingdirectly into the main stem channel (Fig. 4).Approximately two-thirds of the basin was loggedduring the 1970s and, since then, at least foursignificant debris flows have occurred, all associatedwith gullies. The basin is underlain below about300 m by shales and friable sandstones of theCretaceous Haida Formation, and above that levelby conglomerates and sandstones of the CretaceousHonna Formation (Sutherland-Brown 1968).Exposures along gully channels indicate that muchof the basin is mantled with compact basal till,capped with about 1–2 m of unconsolidatedcolluvium derived from till and bedrock weathering.Given the lack of a floodplain along MacmillanCreek, debris fans are generally absent from themouths of the gullies.

52

generally unknown thickness mantles most slopes.Climatically, the Deena Creek study area is probablysimilar to Macmillan Creek in that the general aspectof both areas is northerly and most of the instru-mented gullies are located between 350 and 450 mabove sea level.

Coquitlam basin, located 15 km northeast ofVancouver, is typical of the till-mantled, crystallineintrusive-rock terrain of the southern CoastMountains (Fig. 6). The monitored gullies arelocated on the northwest side of upper Cedar Creekbasin, underlain by coarse-grained intrusives(gabbro to quartz-diorite) generally massive andresistant to erosion (Roddick 1965). Gullies oftenfollow zones of finely fractured rock, some of whichmay be fault lines. Much of the topography ismantled with basal till, 1–5 m thick, though manyknob-shaped bosses of crystalline rock crop out inthe upper and mid-slope sections. Colluvial material

Six gullies were selected for monitoring on thesouth flank of middle Deena Creek: two unloggedgullies (D3, D4) and two slash-full gullies (D1, D2)in upper Shomar Creek, and two slash-clear gullies(D5, D6) in the headwaters of a nearby unnamedbasin (Fig. 5). Most of the Deena map area waslogged in the late 1980s and early 1990s, and duringthis period at least two large debris flows originatedin clearcut gullies (Fig. 5). The terrain is similar tothat in Macmillan basin and is characterized by steepgullies discharging directly into main-stem streams.However, there are geologic differences between thetwo basins; the northeastern half of the Figure 5 maparea is underlain by andesitic volcanic units of theJurassic Yakoun Formation (Sutherland-Brown 1968),which weathers to a more clay-rich debris than thatfound in Macmillan basin. The southwestern half ofthe area is underlain by carbonates and argillites ofthe Jurassic-to-Triassic Kunga Group. A clayey till of

Locations of study areas within Queen Charlotte Islands.

0 20

kilometres

Deena Creek area

Sandspit

QueenCharlotteCity

Macmillan Creek area

Rennell Sound area

Graham Island

Moresby Island

53

Macmillan basin study area, north Moresby Island, Queen Charlotte Islands.

54

Deena Creek study area.

55

usually mantles bedrock and till to a depth of lessthan 2 m. Large colluvial fans have accumulated atthe mouths of most of the gullies. Much of the areaabove 850 m is old-growth forest and was high-leadlogged during the late 1980s and early 1990s. Severaldebris flows occurred in Cedar Creek basin duringthe large rainstorms in November 1990. Two of theflows, one of which initiated in an unlogged area,ran the full length of two instrumented gullies anddestroyed the gully monitoring equipment.

Table 1 summarizes the attributes of allinstrumented gullies in Macmillan, Deena, andCoquitlam basins. Despite the variety of terrain andgeologic conditions in the three study areas, the totalsample of gullies is relatively homogeneous withrespect to morphology. Most gully gradients areclose to 30°, and with the exception of gullies C5and C8 in Coquitlam basin, most are 3–4 m indepth. Gully lengths, and therefore drainage areas,are more variable.

Coquitlam basin study area.

Synoptic Study Field Area Twenty-nine gullies wereinvestigated for the synoptic study in four basinsdraining to Rennell Sound, namely Bonanza Creek,Gregory Creek, Riley Creek, and Shelly Creek (Fig. 7).Most of the area is underlain by weak volcanic andsedimentary rocks of the Jurassic Yakoun Formation.Also present are carbonates and argillites of thelower Mesozoic Kunga Group (Hesthammer et al.1991). All of the formations weather rapidly andhave a low resistance to surface erosion and massmovement (Alley and Thompson 1978; Wilford andSchwab 1983). Although the area has been glaciated,most of the steeper slopes now lack glacial depositsand the dominant surficial material found croppingout on gully walls is colluvium. Debris torrents fromgullies have reworked these colluvial blankets toproduce thick colluvial aprons and fans in most ofthe basins.

56

Process study: gully characteristics

Channel Channel Gully Gully DrainageTreatment length slope depth width area

Basin Gully typea (m) (deg) (m) (m) (ha)

Macmillan M1 LT 150 36 3.5 10 0.20

M3 LT 240 20 3.0 10 0.24

M9 LT 120 25 2.5 9 0.10

M12 LT 170 30 3.5 11 0.19

M2 SF 260 31 4.0 13 0.34

M8 SF 180 19 3.0 12 0.21

M10 SF 170 26 2.0 15 0.25

M11 SF 120 27 3.0 12 0.15

M4 U 100 33 3.0 10 0.10

M5 U 100 31 2.5 12 0.12

M6 U 80 31 2.0 5 0.04

M7 U 80 30 2.5 11 0.09

Deena D1 SF 140 29 3.5 6 0.22

D2 SF 90 29 3.0 6 0.13

D3 U 110 30 5.0 17 0.19

D4 U 120 28 4.0 12 0.15

D5 SC 60 18 3.5 10 0.09

D6 SC 80 28 4.5 10 0.06

Coquitlam C3 LT 230 20 4.5 18 0.49

C5 LT 170 24 10.5 27 0.55

C6 LT 230 32 3.5 13 0.35

C4 SF 140 29 4.0 21 0.42

C8 SF 190 34 1.0 12 0.29

C1 U 120 28 2.0 14 0.17

C2 U 105 27 4.5 17 0.20

C7 U 140 26 4.5 14 0.19

C10 SC 45 30 2.5 10 0.07

C11 SC 80 30 4.0 13 0.12

Mean: 130 28 3.5 13 0.19

Stand. dev: 55 4 2.0 5 0.10

a LT = logged-torrented; SF = logged, slash-full; U = unlogged; SC = logged slash-clear.

57

Rennell Sound study area, showing locations of gullies sampled in the synoptic study.

58

The physical characteristics of gullies in thesynoptic study are summarized in Table 2. Apartfrom the fact that gullies investigated in the synopticstudy are longer than those in the process study, thetwo groups of gullies are similar in terms of theirattributes. These similarities allow comparisons tobe made between the results of the two studiesbelow, under “Discussion.”

The Process Study

Experimental Design The process study wasconducted in the period 1990–1993 to measure theseasonal and year-to-year differences in sedimentproduction, storage and output across four groupsof gullies (Fig. 2):a) logged, slash-full (SF): no logging debris

removed, no debris flows since harvestb) logged, slash-cleared (SC): no debris flows since

harvest; logging debris removed by conventionalyarding, followed by hand cleaning

c) logged, “torrented” (LT): logged, formerly slash-full gullies with at least one debris flow sinceharvest

d) unlogged (U): old-growth gullies with naturaldebris loads and no recent debris flows.

Gully groups (a) and (b) allowed the effects oflogging debris removal to be assessed directly. Group(c) allowed post-torrent sediment yield to be com-pared with that of SC gullies. Group (d) provided acontrol comparison for Group (a).

The original experimental objective was to have abalanced, replicated design involving three replicategullies in each of the four treatment groups (a)–(d),giving a total of 12 gullies in each study area. Thisdesign was subsequently modified since availablefunds permitted the clearing of LWD and loggingslash from only two “treated” Group (b) gullies ineach study area. Macmillan basin (Figs. 3–4) afford-ed an accessible area containing unlogged, slash-full,and many logged-torrented gullies. In this basin,four gullies in each of Groups (a), (c) and (d) weremonitored. However, since Macmillan basin hadbeen logged some 15 years prior to the start of thisstudy, it was not a suitable basin for slash-cleargullies, given that the term “treated gully” in thisstudy refers to debris removal concurrent with orshortly after the yarding of timber. Conversion ofthe slash-full gullies to slash-clear status in

Macmillan basin many years after logging wouldhave entailed the release of considerable volumes ofsediment stored behind the slash barriers.Accordingly, treated gully sites were sought in areaswhere timber harvesting was in progress on NorthMoresby Island during 1990/91.

Two gullies were selected for logging debrisremoval in the southeastern part of Deena Creekbasin (Fig. 5). Machine cleaning (with an American100 grapple yarder), supplemented by hand removalof LWD pieces down to 5 cm diameter, was com-pleted in September 1991. Because of geologic andterrain differences between Macmillan and Deenabasins, two additional slash-full and unlogged gullieswere selected in Deena basin, in close proximity tothe slash-cleared gullies, to provide control for thetwo slash-cleared gullies. The total number of gulliesin Deena basin was then six, yielding a MoresbyIsland total of 18 instrumented gullies.

In Coquitlam basin (Fig. 6), gullies in each ofgroups (a), (c), and (d) were instrumented duringsummer 1990. In summer 1991, two SC gullies wereskyline-logged, then cleared of logging debris withthe yarder and hand cleaning. This yielded a total of11 instrumented gullies in Coquitlam basin. Destruc-tion of two gully installations in Coquitlam basin bydebris flows during the November 1990 rainstormsforced their abandonment in Coquitlam basin.Consequently, the sample size of gullies suitable fortreatment-group comparisons in all field areascombined, over the period 1990–1993, was unavoid-ably reduced from 29 to 26. However, the debris flowevents provided a basis for comparing chronic,fluvial rates of sediment delivery with the morecatastrophic debris flow process (see “Discussion”).

Field Installations and Measurements: ProcessStudy Each gully in the process study wasinstrumented to provide information on all threecomponents of the gully sediment budget: sedimentinput, changes in sediment storage, and sedimentoutput. A quarterly program of site monitoring wasmaintained at most sites during the period March1990 through November 1993. Sediment input wasassessed from arrays of erosion pins (25-cm spikes)inserted at right angles to gully sidewall slopes atlocations where surface erosion seemed to be mostactive. Each pin array consisted of 15–20 pins, withnumbered tags, installed over areas of several squaremetres to provide a local average erosion rate for

59

Synoptic study: gully characteristics

Channel Channel Gully Gully DrainageTreatment length slope depth width area

Basin Gully typea (m) (deg) (m) (m) (ha)

Bonanza B1 LT 900 24 5.0 9 0.61

B2 LT 230 19 6.0 9 0.17

B3 LT 780 24 3.5 8 0.43

B4 LT 280 20 4.5 9 0.17

Gregory G1 U 220 29 6.0 14 0.20

G2 U 340 23 6.0 12 0.29

G3 U 230 17 6.0 12 0.19

G4 U 240 17 6.0 12 0.20

G5 U 250 23 6.0 16 0.24

G6 U 510 17 9.0 15 0.61

G9 U 410 21 5.0 11 0.31

G10 U 180 25 7.0 15 0.18

G11 U 280 28 7.5 17 0.32

G12 U 50 35 6.0 14 0.05

Riley R2 LT 160 30 3.0 6 0.06

R3 LT 260 32 5.5 11 0.19

R4 U 350 24 5.0 12 0.28

R5 U 300 29 4.0 7 0.16

R7 LT 340 27 2.0 5 0.11

R8 LT 320 29 5.0 11 0.23

R9 LT 330 27 3.0 7 0.15

R10 LT 310 27 3.5 8 0.10

R12 LT 390 21 3.0 8 0.20

R13 LT 250 27 1.0 2 0.04

R14 LT 150 29 2.0 6 0.06

Shellyb S1u U 120 38 9.5 14 0.14

S1l LT 180 32 3.5 8 0.10

Shields Bay SB1 LT 130 27 2.0 5 0.05

SB2 LT 190 25 2.0 6 0.07

Mean: 300 26 4.5 10 0.20

Stan. Dev: 180 5 2.0 4 0.15

a LT = logged-torrented; SF = logged, slash-full; U = unlogged; SC = logged slash-clear.b Lower case u indicates unlogged section of gully; lower case l indicates logged section of gully.

60

each eroding soil patch. Each gully contained severalsuch arrays. In addition, periodic inventories ofsmall slides and slumps were conducted to supple-ment the erosion pin results. To convert erosionrates derived from pin surveys were converted tovolumes, pin array areas were multiplied and thensupplemented by the estimated volumes of smallslumps. Division by the total area of each pin arrayyielded an annual rate of equivalent surface loweringin mm/yr.

Sediment storage changes along gully channelswere estimated from measured cross-sections at siteswhere net accumulation or net removal of materialappeared to be taking place. Given the morpholog-ical complexity of most of the gully channels, it wasconsidered impractical to apply the measuredcross-sectional area changes to a representativelength of gully channel to calculate volume changesfor a particular gully reach. These were normalizedby the width of the given cross-section to yieldvalues in m3/m per year.

Sediment output from each gully was determinedfrom the volume of sediment and woody debrisstored behind a porous geocloth screen securedacross a gully channel (Fig. 8). (In Coquitlam basin,somewhat larger sediment traps were established insome gullies by using sediment catchment basinsexcavated on the upslope side of logging roads.)Reinforcing rods driven into the gully channel up-slope of the geocloth screen were used to measurethe changing level of accumulated material over agiven time interval. To convert this depth of

deposition to a volume, the depth values wereapplied to representative areas (of the order 0.5 m2)around each deposition bar. The total volume ofmaterial trapped behind each screen was thennormalized by the gully drainage area as m3/ha.

In the more actively eroding gullies, periodicexcavation of sediment was required to preventovertopping of the screen in the ensuing measure-ment period. Gully M1, a logged-torrented examplein Macmillan basin, proved to be anomalous in itsextremely high rate of sediment delivery, whichresulted in complete burial of the geocloth screen onseveral occasions. From 1992 until the end of thestudy, net sediment output from this gully wasestimated by resurveying several channel cross-sections downstream of the original geocloth screen.

Inspection of geocloth screens during high flowperiods indicated some loss of sand, silt, and clayaround the margins of certain of the screens (forexample, those in gullies M3 and M8 in Macmillanbasin). As the geotextile aged, its permeabilitydecreased as a result of fine sediment plugging thepores. In effect, the geotextile became a small dam,and sediment settled out in the pool upstream. Attimes of high flow, water was observed flowing overthe screen. Results from Coquitlam basin suggestthat about 30% of all suspended fine sediment (siltand clay) was lost from the trap because of finesediment remaining in suspension as water passedover the trap.

A morphological description of each gully wascompleted using compass, hip-chain and tapemeasure, supplemented with photogrammetry andelectronic distance measurement. Gullies longerthan about 250 m were generally avoided, since thesediment screens installed across each gully wouldlikely have been overwhelmed by either water orsediment discharges.

The Synoptic Study

Experimental Design The second phase of theproject, termed the synoptic study, was conducted inthe period 1992–1994 and focussed on patterns ofsediment build-up in gullies over time spans ofseveral decades. This approach was designed tocomplement the relatively short time-frame of theprocess study. Gullies were investigated in theRennell Sound area of southern Graham Island(Fig. 7) with the objective of comparing rates of

View of lower M1 gully, a torrented channel inupper Macmillan Creek basin, showing typicalgeocloth screen installation. Channel is 6 mwide at the screen.

61

debris recharge between logged and unloggedgullies. The calculation of debris recharge ratesrequired knowledge of the elapsed time since the lastdebris flow in a given gully. Good dating control wasa major factor favouring the Rennell Sound area,where there existed a high frequency of recent debrisflows in both logged and unlogged areas, as well asprevious investigations of debris flow ages (Wilfordand Schwab 1983; J. Schwab, unpublished data).Twenty-nine gullies were used in the synoptic study(Fig. 7). Although the study areas used in the processand synoptic studies do not overlap, the similarity ingeologic and terrain conditions between northMoresby Island and southwestern Graham Islandallows the longer-term synoptic study to comple-ment the process study.

Measurement Procedures: Synoptic Study Deter-mination of debris recharge rates in logged andunlogged gullies was based on the volume of LWDand sediment stored along the gully channel and thetime elapsed since the last debris flow event. Animportant assumption was that all of the pre-existing debris fill along a gully had been scouredout by the last debris flow. Examination of recentlytorrented gully channels confirmed the validity ofthis assumption (Fig. 9). Debris fill volumes werecalculated from measured cross-sections at 25-mintervals along the gully channel. Although thesurface area of various debris prisms along a gullycould be accurately surveyed, there was less certaintyas to depth. In most cases, the cross-sectional shapeof a debris prism was well approximated by either atriangle or a trapezium. The longitudinal shape wasidealized as a uniformly tapering triangle. Thevalidity of these assumptions was tested byexcavation through the fill material to bedrock in anumber of gullies. Based on 20 such excavations innine gullies, the observed average cross-sectionalarea was 0.87 m2 (standard deviation 0.02 m2), andthe estimated mean area was 0.71 m2 (standarddeviation 0.07 m2). These tests show that themethods used to estimate the cross-sectional areas ofdebris prisms yielded fairly reliable results.

The total debris volume stored in a gully wasobtained by summing the individual debris prismvolumes, then normalized by gully area to yieldrecharge as m3/ ha as well as by gully length (m3/m).Annual average yields (m3/ha per year) were obtain-ed by dividing area-normalized volumes by elapsed

time since the last debris flow. In some cases, forest-industry and B.C. Ministry of Forests personnelwere able to recall from memory or written recordsboth the year and month of occurrence of debrisflows. In addition, unpublished investigations byJ. Schwab (B.C. Ministry of Forests, Smithers, BC)revealed the occurrence of major cycles of debrisflows in the years 1891, 1917, 1935, and 1978(J. Schwab, pers. comm., 1992; Schwab, this volume).Both dendrochronology and air photograph analyseswere then used to corroborate this evidence, and tosearch for additional debris flow dates.

A chronological sequence of air photographs overthe period 1935 to 1989 was used to constrain thetime span within which known events could haveoccurred. This record of vertical air photographswas supplemented in summer 1992 with aerialoblique images of selected gullies in Bonanza,Gregory, and Riley creek basins. Tree-ring datingconfirmed many of the dates previously noted by

Torrented gully scoured to bedrock after recentdebris flow in Bonanza Creek basin, RennellSound.

62

Cumulative mean sediment output forMacmillan basin gully groups, based onmaterial trapped by geocloth screens.

Schwab and narrowed the time intervals obtainedfrom the air-photograph record. Most debris flowswere dated either by determining the age of alderstands in areas previously scoured by debris flow, orby dating impact scars on older trees growing closeto the margins of a debris flow channel.

Results

Process Study Results

Sediment Output from Gullies The sediment out-puts measured from gullies over a three-year periodare shown as cumulative curves in Figures 10–12. Allvalues are normalized as m3/ha by dividing by gullycontributing area (Table 1). Variations in the slopesof the cumulative curves show that most gully typeshave a peak of sediment production and delivery inthe autumn and winter quarters when maximumprecipitation and runoff occur. Minimum valuestend to occur during the summer period of leastprecipitation. Significant variations in total sedimentoutput are evident across the gully types in eachstudy area. Over the entire period of record,logged-torrented gullies showed the highest output

40

Normalized output (m3/ha)

Dec

–93

35

30

25

20

15

10

5

0

Aug

–93

Ap

r–93

Dec

–92

Aug

–92

Ap

r–92

Dec

–91

Aug

–91

Ap

r–91

Dec

–90

Aug

–90

Ap

r–90

Torrented

Unlogged

Slash full

Cumulative mean sediment output forCoquitlam basin gully groups, 1990–1993,based on material trapped by geocloth screens.

35

Normalized sediment output (m3/ha)

Dec

–93

30

25

20

15

10

5

0

Aug

–93

Ap

r–93

Dec

–92

Aug

–92

Ap

r–92

Dec

–91

Aug

–91

Ap

r–91

Dec

–90

Aug

–90

Ap

r–90

Slash full

Unlogged

Torrented

Slash clear

Cumulative mean sediment output for Deenabasin gullies, 1990–1993, based on materialtrapped by geocloth screens.

6

Normalized sediment output (m3/ha)

Dec

–93

5

3

2

1

0

Aug

–93

Ap

r–93

Dec

–92

Aug

–92

Ap

r–92

Dec

–91

Aug

–91

Ap

r–91

Dec

–90

Aug

–90

Ap

r–90

Slash full

Unlogged

Slash clear

4

63

values in all three study areas; however, both Figures11 and 12 show that, in the period 1991–1993 whenslash-clear gullies operated, they had the highestsediment output. Unlogged gullies are third in rankorder of sediment production and delivery, withlogged slash-full consistently showing the leastoutput of sediment in all three study areas.

The similarity in rank ordering of gully sedimentresponses within different geographic areas suggeststhat geologic differences between the study basinshave a lesser effect on gully sediment output than dodifferences in either land-treatment effect (i.e., loggedversus unlogged) or mass movement history(i.e., torrented versus non-torrented). High sedimentoutputs from torrented gullies are explained in partby a lack of LWD—and hence sediment storagesites—and in part by the high rates of sedimentproduction from recently scoured gully walls. Inslash-cleared gullies, high sediment output wasapparently related more to channel erosion than tosidewall erosion. Removal of logging debrisdisturbed the gully channel sufficiently to releasesediment that otherwise would have been storedbehind LWD. Low sediment outputs from slash-fullgullies are explained by the high trap efficiency ofthick LWD fills. Moderate sediment outputs fromunlogged gullies are attributable to the relativestability of naturally vegetated gully walls, and to thedevelopment of fluvial sediment pathways throughthe decaying LWD fills.

Sediment Storage in Gullies Changes in debrisstorage in gullies were derived from repeatedmeasurements of 15–20 points at each of severalchannel cross-sections in gullies. As noted earlier, itwas not possible to report actual volume changes ingully reaches because of uncertainty in extrapolatingchanges upstream and downstream from themeasured cross-section. Results reported in Tables 3and 4 show normalized changes (m3/m of cross-section width) in Queen Charlotte gullies over theperiod of record. Negative values indicate neterosion, positive values net deposition. Table 5 givesaverage values by both geographic area and gullytreatment group. In most cases, negligible changeswere recorded in slash-full and unlogged gullies;however, both logged-torrented and slash-cleargullies registered significant sediment losses. Theseresults emphasize the importance of LWD inregulating sediment storage. In particular, the lack

of LWD in torrented gullies in Macmillan Creekreflects the fact that most LT gullies in this basin hadbeen scoured by debris flows in the 5–8 years priorto the inception of the process study. Significantsediment recharge had not yet occurred, since mostchannels were totally devoid of LWD at the start ofthe study. In the slash-clear (SC) gullies in DeenaCreek (Table 4), significant erosion of channel bedmaterials was evident following machine and hand-cleaning operations.

In Coquitlam basin, over a similar period ofrecord (Table 6), negligible changes were also notedin both unlogged and slash-full gullies, consistentwith the Queen Charlotte results. The response ofCoquitlam slash-clear gullies was also similar to thosein Deena Creek. Gullies C10 and C11 in Coquitlamshowed very rapid scour of their channels down tobedrock or an armoured bouldery surface within afew months of the completion of machine and handcleaning (as clearly seen in Figure 11). However, asignificant difference in Table 6, relative to the QueenCharlotte results in Tables 3 and 4, is an appreciablenet gain of material in two of the logged-torrentedgullies in Coquitlam basin. This may reflect themuch coarser calibre of colluvial materials inCoquitlam basin, as well as the existence of largegranitic boulders which provided fine-sedimentstorage sites along many channels. By contrast, theless competent bedrock materials in the QueenCharlotte gullies disintegrated almost completely togravel or finer sizes. Queen Charlotte colluvium andtill materials generally lacked large boulders, andthus were more readily eroded by fluvial transport.

Sediment Input to Gullies The available data onsediment inputs (Table 7) are based on repeatedmeasurement of erosion pin arrays establishedwithin obviously eroding patches of exposedmineral soil along gully sidewalls. Each of the datapoints in the table represents an average value fromhundreds of individual pin measurements withineach gully. The data are somewhat more difficult tointerpret than the channel storage data justdescribed because of concerns about a lack of arealcoverage, since it was neither possible nor desirablefor reasons of site disturbance to monitor everypatch of eroding soil within a gully. Given estimatederrors of at least ± 2 mm per year, all values inTable 7 have been rounded to the nearest 5 mm. Thevalues are average rates of surface lowering in the

64

Cross-sectional area changes in Macmillan basin gullies, 1991–1993

Cross- Width Cross-section Inclusive Normalized Mean GullyGully section section (m) change (m2/m) dates change (m2/m) (m2/m) typea

M1 XS-1 6.02 -1.55 4/91–11/93 -0.26

XS-2 5.54 -2.36 -0.43

XS-4 6 -0.49 11/91–11/93 -0.08

XS-5 .87 1.02 0.17

XS-6 8.10 -5.17 -0.64 -0.25 LT

M3 XS-1 5.91 0.10 4/91–11/93 0.02

XS-2 4.60 0.27 8/91–11/93 0.06

XS-3 11.30 0.07 0.01 0.03 LT

M4 XS-1 2.97 0.68 8/91–11/93 0.23

XS-2 4.82 -0.56 -0.12

XS-3 3.90 -0.28 -0.07 0.01 U

XS-1 3.80 0.16 8/91–11/93 0.04

XS-2 3.16 0.01 0.00

XS-3 5.80 -0.41 -0.07 -0.01 U

M6 XS-1 3.03 0.14 8/91–11/93 0.05

XS-2 3.00 -0.01 -0.00 0.02 U

M7 XS-1 3.84 -0.22 8/91–11/93 -0.06

XS-2 2.94 0.07 0.02 -0.02 SF

M8 XS-1 7.15 -0.30 8/91–11/93 -0.04

XS-2 6.82 -0.09 -0.01

XS-3 5.70 -0.07 -0.01

XS-4 6.60 -0.04 -0.01 -0.02 LT

M9 XS-1 2.60 -0.02 8/91–11/93 -0.01

XS-2 3.80 0.11 0.03

XS-3 3.45 -0.18 -0.05 -0.01 LT

M10 XS-1 4.24 -0.05 8/91–11/93 -0.01

XS-2 2.72 0.06 0.02

XS-3 3.57 0.23 0.06 0.02 SF

M11 XS-1 3.52 0.04 8/91–11/93 0.01

XS-2 5.78 -0.40 -0.07 -0.03 SF

M12 XS-1 3.20 0.01 8/91–11/93 0.00

XS-2 3.77 0.36 0.10

XS-3 7.83 0.32 0.04 0.05 LT

a LT = logged-torrented; SF = logged, slash-full; U = unlogged; SC = logged slash-clear.

65

Cross-sectional area changes in Deena basin gullies, 1991–1993

Cross- Width Cross-section Inclusive Normalized Mean GullyGully section section (m) change (m3/m) dates change (m3/m) (m3/m) typea

D1 XS-1 4.95 -0.02 8/91–11/93 -0.00

XS-2 8.30 0.10 0.01

XS-3 3.38 0.11 0.03 0.01 SF

D2 XS-2 3.47 0.07 8/91–11/93 0.02

XS-2 2.91 -0.03 -0.01

XS-3 8.00 0.41 0.05 0.02 SF

D3 XS-1 1.80 0.11 8/91–11/93 0.06

XS-2 4.42 -0.08 -0.02 0.02 U

D4 XS-1 2.35 -0.14 4/91–11/92 -0.06

XS-2 7.60 0.04 6/91–11/92 0.01

XS-3 6.26 -0.08 8/91–11/92 -0.01 -0.02 U

D5 XS-1 3.35 0.02 11/91–11/93 0.01

XS-2 4.01 -0.10 -0.03

XS-3 3.39 -0.18 -0.05 -0.02 SC

D6 XS-1 2.63 -0.15 11/91–11/93 -0.06

XS-2 3.15 -0.19 -0.06 -0.06 SC

a LT = logged-torrented; SF = logged, slash-full; U = unlogged; SC = logged slash-clear.

Cross-sectional area changes in Queen Charlotte gullies, 1991–1993

Gully treatment groupsa

LT SF U SC Basin meanBasin (m3/m) (m3/m) (m3/m) (m3/m) (m3/m)

Macmillan -0.062 -0.007 0.003 -0.022

Deena 0.017 -0.001 -0.041 -0.008

Regional means -0.062 0.005 0.001 -0.041

a LT = logged-torrented; SF = logged, slash-full; U = unlogged; SC = logged slash-clear.

66

Cross-sectional area changes in Coquitlam gullies, 1991–1993

Width of Cross-section Normalized SiteGully Cross- section area change Inclusive cross-section area mean Gullysite section (m) (m2) dates change (m3/m) values (m3) treatmenta

C2 XS-2 5.7 0.00 10/91–5/93 0.00 U

XS-3 4.7 -0.32 -0.07

XS-4 4.2 0.04 0.01

XS-5 5.8 0.23 0.04 -0.00

C3 XS-1 6.3 0.09 10/91–5/93 0.01 LT

XS-2 7.6 0.11 0.01

XS-3 10.1 -0.01 -0.00

XS-4 7.0 0.30 0.04

XS-5 8.0 -0.02 -0.00 0.01

C5 XS-2 10.0 0.21 10/91–5/93 0.02 LT

XS-3 7.8 0.15 11/91–5/93 0.02

XS-4 7.0 0.14 0.02

XS-5 8.0 0.20 0.03

XS-6 5.7 0.03 0.01 0.02

C6 XS-1 5.3 -0.18 10/91–5/93 -0.03 LT

XS-2 3.8 -0.14 -0.04

XS-3 4.7 0.06 0.01 -0.02

C10 XS-1 3.4 -0.32 10/91–5/93 -0.09 SC

XS-2 3.3 -0.43 -0.13

XS-3 3.7 0.11 0.03

XS-4 4.5 -0.51 -0.11 -0.08

C11 XS-1 3.7 -0.09 10/91–5/93 -0.02 SC

XS-2 4.0 -0.29 -0.07

XS-3 5.8 -0.20 -0.03

XS-4 5.3 -0.27 -0.05 -0.04

a LT = logged-torrented; SF = logged, slash-full; U = unlogged; SC = logged slash-clear.

67

soil patches monitored; they do not reflect averageerosion rates over the entire gully sidewalls, but givesome idea of typical rates of change within the mostrapidly eroding sidewall areas.

The highest group means in Table 7 occur in thelogged-torrented (LT) category, followed in rankorder by slash-full (SF), slash-clear (SC), andunlogged (U) gullies. Since many SF gullies werelogged at least 10 years prior to the inception of thisstudy, the implication is that healing of bare soilareas caused by yarding disturbance takes asignificant time to complete. Slash-clear gullies haverelatively low erosion values, which may reflect therelatively careful logging practices used in thesegullies. The highest erosion pin values in logged-torrented gullies in both study areas reflect therelatively recent scour of sidewalls below the rootdepth of most shrubs. Significant erosion byrainsplash, overland flow, dry ravelling, and frostaction occurs before the gully walls eventuallybecome revegetated.

Sediment Transport During the November 23, 1990,Debris Flows in Coquitlam Basin During theautumn and winter of 1990/91, several large stormsoccurred in the southern Coast Mountains, causingnumerous debris flows. The storm of November 23,1990, caused a total of 19 debris flows in Coquitlambasin alone, nine of which occurred in Cedar Creekbasin. Most of these debris flows initiated in old-growth forest areas, then scoured the clearcut sectionsof the affected gullies. Two of the monitored gulliesexperienced debris flows (gullies C1 and C6; Fig. 6),causing complete destruction of the gully-monitoringinstallations. The C1 event initiated in old-growth

forest and therefore may be regarded as a naturalevent. The C6 event initiated within the clearcutportion of the gully. Both debris flows were surveyedin early December 1990, shortly after their occur-rence, to assess the total volume of material involved(Table 8). These surveys indicated about 1.2 m3/m ofLWD derived from logging in gully C6. This debrisand its associated sediment are estimated to haveincreased the magnitude of this event byapproximately 20%.

The occurrence of the events within monitoredgullies allows a comparison of the quantities ofmaterial moved by chronic, non-catastrophic fluvialevents and episodic mass-transport events. Normal-ized by gully length, the C1 debris flow yielded3.5 m3/m of scoured channel length and the C6 flowyielded 5.9 m3/m, for an average of 4.7 m3/m fromboth flows (Table 8). Using somewhat larger gullieson the Queen Charlotte Islands, Fannin andRollerson (1993) calculated typical debris yields of5–10 m3/m. Thus, the events in C1 and C6 areprobably typical of debris flows in the smaller-sized

Average erosion pin response by gully treatment group, 1990–1993

Study area Treatment groupa Average erosion rate (mm/yr)

Coquitlam basin LT 15SF 10U 5

Macmillan andDeena basins LT 25

SF 20U 10SC 10

a SC value refers to period March 1992–November 1993 only.

Volumes of November 1990 debris flows,Coquitlam basin

Initial Transport Deposited Totalfailure zone volume volume volume

Gully volume (m3) (m3) (m3) (m3)

C1 400 1340 -170 1560

C6 220 900 -230 890

Mean volume: 1230

68

gullies considered in this study. The average volumefrom the two debris flows was 1230 m3 (Table 8). Bycomparison, the average fluvial sediment yield cal-culated from Coquitlam basin sediment traps ingullies was only 2.7 m3/yr. (This figure ignores slash-full gullies which registered virtually zero outputover the period of study.) Assuming a recurrenceinterval of 50 years for such debris flows (as suggest-ed by the work of Schwab, this volume), episodicsediment export by debris flows over a 50-year periodwould be 9 times greater than the cumulative fluvialtransport of sediment (i.e., 1230 m3 versus 135 m3).However, many old-growth gullies contain apparentlyundamaged trees 300–400 years old, growing close tothe margins of gully channels. In these cases, debrisflow frequency may be of the order 500–1000 yearsand fluvial sediment output would equal or exceedthe debris flow sediment output.

Synoptic Study Results As noted previously, theprincipal objective of the synoptic study was toestimate the rates of sediment and woody debrisrecharge following torrenting in a representativesample of 13 unlogged and 16 logged gullies in theRennell Sound area of southwestern Graham Island(Fig. 7). Gully fill volumes were estimated from acombination of profile surveying and outrightexcavation of fills in 15 of the 29 gullies, thennormalized by gully drainage area to yield m3/ha.Annual average debris recharge rates (m3/ha peryear) were obtained by dividing each normalizedstorage volume by elapsed time since the last debrisflow (Table 9). During the gully surveys, separateinventories were kept of both coarse woody debris(CWD) and clastic sediment. All data are rounded tothe nearest five units.

Recharge Rates of LWD and Clastic Sediment FromTable 9 it is evident that after a debris torrent hasoccurred, LWD is delivered to unlogged, old-growthgullies at a rate more than twice that of clearcutgullies. It is worth re-emphasizing that this assumesthat the original surcharge load of LWD fromlogging has already been removed by a post-harvestdebris flow, which was the case in all examplesreviewed here. The LWD supply rate in old-growthgullies is related to the higher potential for wind-throw of branches and entire trees in mature toover-mature forests. The inference is that, untilsecond-growth forest reaches maturity, lower rates

of LWD input should be expected relative to those ofold-growth gullies. It is not known whether clearcutgullies are likely to recover to the background ratesof LWD supply found in old-growth gullies over the80- to 100-year rotation period typical of coastalharvesting. The supply rate of clastic sediment toclearcut gullies is roughly double that observed inold-growth gullies (Table 9). However, the clearcutstandard deviations are much higher, indicatingmuch more variability in this gully group. Thereason for this is that surface erosion rates typicallyproduce a right-skew distribution in which themean and variance are correlated. The higher LWDsupply in unlogged gullies offsets the lesser rate ofsediment supply, to yield a total debris supply ratethat is some 75% of the clearcut figure (Table 9,mean values).

These simple comparisons of LWD and sedimentrecharge rates ignore the differences in mean debrisrecharge times between the two sub-samples ofgullies. For example, in logged gullies, the meanelapsed time since the last debris flow was 7 years,and the earliest documented event was 14 years oldat the time of the 1992 surveys (i.e., 1978). In un-logged gullies, the mean elapsed time is 21 years andthe earliest event is 75 years old (i.e., 1917). Sincethe earliest documented debris flow in the loggedsample is only 14 years old, recharge rates were re-compared over this common time scale. For clasticsediment, the mean rate of recharge in clearcutgullies is almost double that recorded in old-growthgullies over the 14-year period of common record.However, the high variance in both samples pre-cludes a statistically significant distinction. Given thelower rates of LWD input to clearcut gullies, thehigher rate of sediment recharge in this gully groupimplies much higher rates of sediment delivery. Thisis explained by the greater instability of gullysidewalls and bank tops where vegetation has beenremoved, and by higher sediment production fromhillslopes disturbed by logging operations (Schwab1983; Rood 1984; Sidle et al. 1985).

The changing rates of sediment recharge over theentire 75-year period of record are shown inFigure 13. The data indicate a decline in the rate ofrecharge over time within both gully groups, withthe suggestion of a steeper rate of decrease inrecharge rate in clearcut gullies. This is consistentwith the progressive stabilization of gully walls withtime, causing a decrease in sediment supply.

69

Recharge rates of LWD and sediment in Rennell Sound gullies

LWD Sediment Total Total Time sincerecharge recharge recharge recharge last debris flow

Gully (m3ha-1yr-1) (m3ha-1yr-1) (m3ha-1yr-1) (m3m-1yr-1) (yr)

Unlogged

G1 10 135 145 0.14 14

G2 10 140 150 0.13 14

G3 30 120 150 0.13 14

G4 15 130 145 0.13 14

G5 <5 20 20 0.03 75

G6 <5 20 20 0.03 57

G9 20 40 60 0.06 14

G10 <5 40 40 0.13 14

G11 30 95 125 0.16 14

G12 <5 50 50 0.08 14

R4 <5 90 90 0.11 6

R5 <5 70 70 0.04 10

S1u <5 30 30 0.05 8

Means: 10 75 85 0.09 21

Standard Dev: 10 45 50 0.05 20

Clearcut

B1 25 530 555 0.42 1

B2 25 195 220 0.19 8

B3 0 75 75 0.05 10

B4 <5 75 75 0.06 6

R2 0 195 195 0.15 3

R3 <5 60 60 0.06 11

R7 <5 210 210 0.11 2

R8 <5 25 30 0.03 7

R9 <5 60 65 0.04 14

R10 <5 165 170 0.18 7

R12 10 300 310 0.28 1

R13 <5 90 90 0.05 8

R14 <5 40 40 0.03 14

S1l <5 135 140 0.14 8

SB1 0 65 65 0.04 8

SB2 <5 15 15 0.01 8

Means: 5 140 145 0.12 7

Standard Dev: 10 130 135 0.11 4

70

Although there are only two data points controllingthe long-term part of the recharge curve, one canspeculate that most gullies might be fully rechargedwith material within the time frame of a coastalforest-harvest rotation.

Discussion

Comparison of Process Study and Synoptic StudyResults Both studies reported here were designed toprovide quantitative data on the effects of LWD onsediment production, storage, and delivery in coastalgullies. The process study provided a detailed assess-ment of the effects of varying concentrations ofLWD, including deliberate removal of LWD follow-ing harvest, over a three-year period. The influenceof LWD on the sediment regimen of gullies is seenin the sediment output data (Figs. 10–12). Thesedata integrate the net effects of both sediment inputand storage across entire gully groups. Of particularnote are the consistently very low sediment outputsfrom slash-full gullies in comparison with all othergully groups, whereas erosion pin data (Table 7)suggest that slash-full and logged-torrented gullieshave similar sediment input values, well above thoseof unlogged gullies. Together these data on sediment

input and output imply higher sediment storage inslash-full gullies, a tendency partly confirmed by thetreatment group means in Tables 4, 5 and 6.

The synoptic study provides up to a 75-yearhistory of debris recharge in previously torrentedgullies and therefore gives a more reliable picture oflong-term sediment storage. In many cases, thevolumes of channel fill were corroborated by actualexcavation of gully fill material. As the regressionlines in Figure 13 show, the mean fill rate of LWDplus sediment in both logged and unlogged gullies isapproximately 0.08 m3/m per year. Over a 2.5-yearperiod comparable to the duration of the processstudy, the average recharge would be 0.20 m3/m ofchannel length, which reduces to 0.04 m3/m ofchannel width when an average channel width of5 m is used. This value is of the same order ofmagnitude as many of the positive values reportedin Tables 4 and 6, indicating a broad similarity ofbehaviour across gullies in both the process andsynoptic studies.

It is instructive to compare the above values ofpost-torrent recharge with other estimates fromsimilar environments. Fannin and Rollerson (1993)estimated average gully debris yields by dividing thetotal volume of recent debris flows by the length ofscoured channel, obtaining a mean value of 6.6 m3/m.In our study, we have calculated 3.5 m3/m and5.9 m3/m, respectively, for the C1 and C6 debrisflows of November 1990 in Coquitlam basin. Thesevalues are, of course, significantly higher than ourshort-term recharge rate of 0.08 m3/m computedfrom the synoptic study, since much of the surchargeload of LWD and sediment derived from logginghad already been scoured from the clearcut sampleof gullies by recent debris flows. When our short-term recharge rate of 0.08 m3/m is extrapolated, theFannin-Rollerson yield represents about 80 years ofrecharge, and the Coquitlam about 40–70 years ofrecharge. Since debris flows may be more frequentthan this in many gullies, it is likely that the surchargeload of LWD from logging operations accounts forthe high debris yields obtained from the debris flowvolume calculations. Estimates of LWD surchargefrom logging operations range from 0.4 m3/m byFroelich (1973) to as much as 1.2 m3/m by Millard(1993) in non-torrented slash-full gullies inCoquitlam basin. Many slash-full gullies probablyhave LWD loads even higher than this. Thus, logging

Rate of sediment recharge in logged andunlogged Rennell Sound gullies over theperiod, 1917–1992. Rates are based on thevolumes of clastic material stored in gullies,normalized by gully contributing area and timesince last debris flow.

1000

Sediment recharge rate (m3/ha/yr)

100

101 10 100

Elapsed time since last debris flow (years)

R = 385 t-0.80

ClearcutOld growth

71

operations apparently increase debris flow magni-tude by 6–18% if slash is left in gullies, and probablyby as much as 30% if the sediment surcharge load isincluded. The volume of woody debris in a gully istherefore as much an issue as the volume of sedi-ment stored within the woody debris, and in manycases is volumetrically greater.

Implications for Clearing Woody Debris fromGullies It is well known that logging results in asignificant increase in the volume of woody debris ina channel, and that over time sediment is trapped bythis debris. The sediment outputs in Figures 10–12suggest that removal of LWD from slash-clear gulliescauses an increase in the level of chronic fluvialsediment output compared with that in loggedgullies which retain their logging debris. This isattributed both to channel disturbance during gullycleaning, and to removal of LWD which providessediment storage sites. In the event of future debrisflows, slash-clear gullies should produce lesser-magnitude events, since higher chronic outputs ofsediment imply lesser volumes of long-termsediment storage. Woody debris is therefore a criticalfactor in gully behaviour, but affects fluvial anddebris flow processes differently. Water transport ofwoody debris depends on sufficient discharge tofloat and move the debris. Generally, the larger thechannel, the larger the size of woody debris moved.Water-transported debris tends to develop aclumped distribution along the length of thechannel, frequently resulting in small wedges ofdebris and sediment and the development of astepped channel profile. In severe cases, the channelis filled with debris and sediment, and erosion of thebanks or sidewalls may result.

Of greater consequence is the occurrence of adebris flow, since the entire debris fill in the gully isusually scoured away and the surcharge loads ofwoody debris and sediment augment debris flowmagnitudes. The decision concerning removal ofwoody debris from logged gullies therefore requiresan assessment of the probability of debris flowoccurrence. This is one of the main functions of theGully Assessment Procedures within the ForestPractices Code. As detailed in those guidelines,several factors must be considered, including thesteepness and drainage area of a gully channel, the

steepness and surface area of the sidewalls andheadwall, the thickness of colluvium on sidewalls,the degree to which sidewalls were disturbed duringharvest and the sensitivity of main-stem channelsdownstream of the gully. A major consideration iscost, since cleaning debris from gullies is expensiveand can only be considered where the benefits justifythe expenditure. Worker safety is also an importantissue. If the likelihood of a debris flow is high, thenwoody debris should be cleaned. In some cases,cleaning debris from gullies may not be financiallyfeasible and may preclude logging of a gully orrequire modified logging methods.

As noted in this study, removal of woody debrisalso has physical consequences. Gully channelscleared of debris in this study responded by erodingtheir channel beds, and it is unlikely that thiserosion would have occurred had woody debrisremained in the channel. Thus, an increase in finesediment output is to be expected when LWD isremoved. Provided that the long-term, chronicoutput of fine sediment is not deleterious to thereceiving channel, there may be a land-managementadvantage to clearing logging debris from gullies ifthe objective is to reduce the magnitude of futuredebris flows. Conversely, if a gully is cleaned tominimize debris flow volume and a debris flow doesnot occur, an environmental cost is incurred fromchronic sediment output from the gully, in additionto the financial costs incurred.

Recommendations for Clearing Woody Debris Thefollowing recommendations are provided as guide-lines for cleaning woody debris. Although the resultspresented earlier in this study apply to some of therecommended treatments, it is important to empha-size that there are as yet no data available to quantifythe effects of many of these recommendations onsediment storage and delivery in gullies. In any event,each gully would need to be assessed for the likeli-hood of debris flow occurrence, as well as for channelcapability to transport woody debris and sedimentby fluvial action. If woody debris is left in a gullyafter harvest, it is likely to become incorporated intothe channel structure over time. Prescriptions forwoody debris cleaning should therefore reflect thecurrent condition of both the woody debris as wellas the channel, as outlined below.

72

Recently Logged Gullies (Less than 1 Year AfterHarvest) with Moderate to High Water Flows and aLow Probability of Debris Flows The objectives inthis case are to prevent excess transport of woodydebris downstream, or to prevent woody debrisfrom diverting flows towards erodible channel banksand sidewalls. Two cases should be considered:• Woody Debris that may be Moved by Water Flows

This situation requires that small woody debris beremoved first and large debris left behind. Thegreater the water discharge of the gully, the largerthe maximum size of woody debris to beremoved. At present, there are few quantitativeguidelines for woody debris flotation andtransport in gullies.

• Large Woody Debris Directing Water Flow TowardsErodible Banks or Sidewalls If large woody debrisremaining in the channel will block the channeland cause bank or sidewall erosion, then some ofthe large debris should be removed as well.Occasional large pieces of debris should be left inthe channel, particularly if removal of the debriswould significantly disturb the channel, orwould disturb natural woody debris present priorto harvest.

Recently Logged Gullies with High Debris FlowPotential and Low to Moderate Water TransportPotential The objective in this case is to minimizethe additional volume of debris and sediment whichcould be incorporated into a post-harvest debrisflow. Since there is not sufficient water power tomove most of the debris, much will stay in placeuntil a debris flow occurs. Channels with high debrisflow potential and more than 2 m3 of woody debrisper metre of channel length should be cleaned toremove most of the woody debris. The 2 m3/mguideline is derived from the study by Fannin andRollerson (1993). Small gullies that likely have lowwater flows typically generate 1–5 m3 of sedimentand debris per metre length of channel in a debrisflow. Channels with greater than 2 m3 of loggingdebris will yield an approximate doubling of debrisflow volume per metre of channel length in thelogged area should the gully fail. However, since thelogged, low-flow portion of the gully generates onlya fraction of the volume in a debris flow, theincrease in the final volume of the debris flow islikely to be less than double.

Recently Logged Gullies with Moderate to HighWater Transport Potential and Moderate to HighDebris Flow Potential For gullies having significantwater transport and debris flow potential, woodydebris should be managed to consider both hazards.Small woody debris should be removed. The totalvolume of woody debris should be assessed and, ifgreater than 2 m3/m, then the total volume shouldbe reduced to minimize debris flow volume.

Recently Logged, Steep Channels with High WaterTransport Potential Casual observation of somegullies shows that woody debris loads may occasion-ally be mobilized causing debris jams or, in the eventof flotation of large volumes of LWD, perhaps evendebris flows. Debris flows mobilized in this mannerare known to have occurred in upper Gordon Riverbasin, southwestern Vancouver Island. To minimizethis type of activity, removal of all logging-relatedwoody debris (except for the occasional very largepiece) should be considered in gullies with thefollowing characteristics:• steep channels, typically greater than 60% grade

with sufficient water discharge to mobilize largeamounts of woody debris by flotation; these arelikely to be channels wider than 5 m, with peakflows deeper than 0.5 m; and

• channels developed in bedrock or hard glacial till,where woody debris has little chance of beingbuttressed against flotation by channel deposits.

Gullies Logged more than 5 Years Ago with Moderateto High Water Transport Potential Gullies that havenot already produced a debris flow within severalyears of harvest are likely to have incorporated someof the woody debris into the channel deposits bypartial burial, and some smaller woody debris willhave been transported farther down the gully or intolower gradient streams. The objective here should beto minimize storage of sediment and woody debrisin the gully reach, since this could lead to channelwidening and increased erosion of sidewalls. Theefficacy of LWD cleaning in this situation decreasesas elapsed time since harvest increases, because ever-increasing proportion of the woody debris becomesburied by sediment and incorporated into thechannel structure. All loose, transportable woodydebris should be removed, but buried or partlyburied material should be left undisturbed unless the

73

part of a log still exposed can be cut and removed. Iferosion of gully sidewalls is occurring, considerationshould be given to removing woody debris that isdirecting water flows against the sidewalls, or to plac-ing woody debris in a way that protects the sidewalls.

Gullies Logged more than 5 Years Ago with High toModerate Debris Flow Potential, but Low WaterTransport Potential Woody debris in these types ofgullies tends to remain in place over time, unless adebris flow occurs. There is no significant hazard ofthe woody debris initiating a debris flow or causingerosion. The objective in these gullies is to avoid asignificant increase in the total volume of a debrisflow should one occur. Estimates of the amount ofstored woody debris should be made along theentire logged gully reaches, and compared to thetotal volume of material likely to be generated fromsource area to deposition zone (consult Fannin andRollerson 1993). If the total volume of a debris flowwould increase by more than 50%, then the woodydebris should be cleaned. If the gully was loggedmore than 10 years ago, the volume increase shouldbe greater than 100% to warrant cleaning. It isimportant to consider how much damage to sidewallsand vegetation will occur before deciding to clean.

Debris Jams in Gullies Log jams may result fromeither debris flows coming to rest or from watertransport of woody debris. Jams may be breached inthree main ways. First, the LWD of the jam itselfmay fail and develop into a debris flow as theimpounded saturated sediment is released. Ingeneral, jams tend to occur in areas of netdeposition, so, because of inadequate channel slope,such catastrophic failure is not likely to occur. Asecond, more likely failure mechanism involves adebris flow from higher up the gully system, whichbreaches the jam and incorporates the storedmaterial into the flow. The first post-logging debrisflow is generally the largest and is also likely to travelthe farthest. Consequently, if a debris jam forms inthe lower part of the gully channel as a result offrom the first post-logging event, subsequent smallerevents are less likely to re-mobilize the jam. Waterflows may erode part or all of the jam in a shorttime, or over a number of flood events. If the jam isvery recent, the probability of subsequent jam failureor release is greater than if the jam has survived a

number of high flows. This is particularly true ofjams that are wedged into erosion-resistant locations,such as bedrock sidewalls, or jams that incorporatepieces of woody debris several metres in length. Theseare unlikely to fail or erode until the wood in the jambecomes rotten. Even in such cases, breakdown of thejam and gradual release of stored sediment is likely tooccur over a period of many years.

Debris Jams Subject to Moderate to High Water FlowsJams should have all wood not buried in sedimentremoved, so the sediment wedge behind the jamdoes not increase in volume over time. If a jam islocated adjacent to erodible sidewalls, part of themiddle of the jam should be notched, to maintainwater flows down the middle of the gully. This willalleviate erosion of the sidewalls.

Debris Jams in Gullies with High Debris FlowPotential If a jam is structurally weak (i.e., notwedged against bedrock or composed of small debrispieces), sits high in the gully system where channelgradients are close to 50%, and has a high potentialfor future debris flows, the jam should be removed ifpracticable so that catastrophic breaching is avoided.

Conclusions

Despite the acknowledged importance of woodydebris in the sediment dynamics of gullies, there arevery few quantitative data documenting the real-time interactions between sediment and coarsewoody debris in gullies. The study results reportedhere provide data on the effects of woody debris inboth logged and unlogged channels, includingobservations on the effects of woody debris removalfollowing timber harvest. A three-year monitoring ofgullies in the Queen Charlotte Islands and thesouthern Coast Mountains reveals similar rank-orderings of gully types with regard to sedimentoutput when results are normalized by gully size.Logged torrented gullies show the highest outputs inthe Queen Charlotte gullies, followed in rank orderby logged slash-clear gullies and unlogged old-growth gullies. In Coquitlam basin, slash-cleargullies have the highest rank. In both study areas,logged slash-full gullies registered the lowestsediment outputs over the three-year study period.

74

A longer-term synoptic study of sedimentrecharge rates over periods of decades in torrentedgullies produced normalized rates of the same orderof magnitude as those noted in the three-yearprocess study. In both the process and synopticstudies, variations in sediment output and storagerate are explained by the variable trap-efficiency oflogging debris for sediment moving down gullychannels in the various types of gullies. This is highin the case of logged slash-full gullies, very low inrecently torrented or slash-cleared gullies, andmoderate in most unlogged gullies.

The recommendations presented here for post-harvest removal of woody debris from logged gulliesemphasize the importance of local assessments ofdebris flow and water transport hazards, in additionto the time elapsed since logging and the sensitivityof downstream areas to debris loadings. Althoughsome of the treatments recommended stem directlyfrom the results presented in this study, many haveyet to be corroborated by measurements of sedimentmovement in gullies. An important objective forfuture work is to fill these gaps in knowledge toallow more precise evaluations of the benefits andcosts of woody debris removal.

Acknowledgements

We would like to acknowledge the assistance andsupport of the individuals and organizations thathave contributed to this study. The research wassupported by several contracts from the B.C. Ministryof Forests, and in part by an NSERC operatinggrant. Other organizations and individuals providedfacilities or data to assist us: Fletcher-Challenge(Sandspit), the Greater Vancouver Water District,B.C. Hydro, Husby Forest Products, and Jim Schwab,Ministry of Forests, Smithers. Thanks are due toSteve Chatwin and Mike Church who produced thefirst draft of a proposal from which these studieswere developed. We are also grateful to Derek Bonin,Dan Hogan, Ray Krag, and Jim Schwab for theiradvice and support.

Many undergraduate and graduate studentsassisted us over the years in the challenging andhazardous task of monitoring and measuring gulliesunder often adverse weather conditions. We wouldlike to acknowledge Lars Uunila, Scott Babakaiff,Sue Young, John Matechuk, Craig Nistor, and

Gordon Clark. We are especially indebted to ScottDavidson, who played a major role in field datagathering for the Moresby Island projects.

References

Alley, N.F. and B. Thomson. 1978. Aspects ofenvironmental geology, parts of GrahamIsland, Queen Charlotte Islands. B.C. Min.Environ., Res. Anal. Branch, Bull. 2.

Buchanan, P. and K.W. Savigny. 1990. Factorscontrolling debris avalanche initiation.Can. Geotech. J. 27:659–675.

Chatwin, S.C., D.E. Howes, J.W. Schwab, and D. N.Swanston. 1994. A guide for management oflandslide-prone terrain in the PacificNorthwest. B.C. Min. For., Land Manage.Handb. No. 34, 2nd ed.

Dietrich, W.E. and T. Dunne. 1978. Sediment budgetfor a small catchment in mountainous terrain.Zeitschrift fur Geomorphologie, Suppl. Band29:191–206.

Fannin, R.J. and T.P. Rollerson. 1993. Debris flows:some physical characteristics and behaviour.Can. Geotech. J. 30:71–81.

Froelich, H.A. 1973. Natural and man-caused slashin headwater streams. In Loggers Handb. 33.Pacific Logging Congress, Portland, Oreg.

Hesthammer, J., J. Indrelid, P.D. Lewis, and J.W.Haggart. 1991. Geology of southwesternGraham Island, Queen Charlotte Islands,British Columbia. Geolog. Survey of Canada.Open File 2319.

Howes, D.E. 1987. A method for predicting terrainsusceptible to landslides following forestharvesting: a case study from the southernCoast Mountains of British Columbia. InForest Hydrology and Watershed Management.R.H. Swanson, P.Y. Bernier, and P.D.Woodward (editors). IAHS Publ. No. 167,pp. 143–154.

75

Howes, D.E. and E. Kenk. 1988. Terrain classificationsystem for British Columbia (revised ed.). B.C.Min. Environ., Victoria, B.C. Manual No. 10.

Krag, R.K., E.A. Sauder, and G.V. Wellburn. 1986. Aforest engineering analysis of landslides inlogged areas on the Queen Charlotte Islands,British Columbia. B.C. Min. For., Victoria, B.C.Land Manage. Rep. No. 43.

Millard, T.H. 1993. Sediment in forested and loggedgullies, coastal British Columbia. MSc thesis,Univ. B.C., Geography Department, Vancouver,B.C.

Oden, M.E. 1994. Debris recharge rates in torrentedgullies on the Queen Charlotte Islands. MScthesis, Univ. B.C., Geography Department,Vancouver, B.C.

Roberts, R.G. and M. Church. 1986. The sedimentbudget in severely disturbed watersheds,Queen Charlotte Ranges, British Columbia.Can. J. For. 16:1092–1106.

Roddick, J.A. 1965. Vancouver North, Coquitlam,and Pitt Lake map areas, British Columbia.Geolog. Survey of Canada, Memoir 335.

Rood, K.M. 1984. An aerial photograph inventory ofthe frequency and yield of mass wasting eventson the Queen Charlotte Islands, BritishColumbia. B.C. Min. For., Victoria, B.C. LandManage. Rep. No. 34.

Rood, K.M. 1990. Site characteristics and landslidingin forested and clearcut terrain, QueenCharlotte Island, B.C. B.C. Min. For., Victoria,B.C. Land Manage. Rep. 64.

Sidle, R.C., A.J. Pearce, and C.L. O’Loughlin. 1985.Hillslope stability and land use. Amer.Geophys. Union, Water Res. Monogr. 11.

Sutherland-Brown, A. 1968. Geology of the QueenCharlotte Islands. B.C. Dep. Mines andPetroleum Resources, Bull. No. 54.

Swanson, F.J., R.L. Fredrickson, and F.M.McCorison. 1982. Material transfer in awestern Oregon forested watershed. In Analysisof coniferous forest ecosystems in the westernUnited States. R.L. Edmonds (editor).Hutchinson and Ross, Stroudsburg, Pa.,pp. 233–266.

Swanston, D.N. and F.J. Swanson. 1976. Timberharvesting, mass erosion and steepland forestgeomorphology in the Pacific Northwest. InGeomorphology and Engineering. D.R. Coates(editor). pp. 199–221.

Wilford, D.J. and J.W. Schwab. 1983. Soil massmovements in the Rennell Sound area, QueenCharlotte Islands, British Columbia. Can.Hydrol. Symp. 82, pp. 521–541.