Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Debris recharge rates in torrented gullies on the Queen Charlotte Islands Oden, Marian Elizabeth 1994

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-ubc_1994-0581.pdf [ 4.12MB ]
JSON: 831-1.0087695.json
JSON-LD: 831-1.0087695-ld.json
RDF/XML (Pretty): 831-1.0087695-rdf.xml
RDF/JSON: 831-1.0087695-rdf.json
Turtle: 831-1.0087695-turtle.txt
N-Triples: 831-1.0087695-rdf-ntriples.txt
Original Record: 831-1.0087695-source.json
Full Text

Full Text

DEBRIS RECHARGE RATES IN TORRENTED GULLIESON THE QUEEN CHARLOTTE ISLANDSbyMaiian Elizabeth OdenB.A., University of Colorado, 1991A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFMASTER OF SCIENCEinTHE FACULTY OF GRADUATE STUDIES(Department of Geography)We accept this thesis as conformingto the required standard.?,41.THE UNIVERSITY OF BRITISH COLUMBIAJuly 1994© Marian Elizabeth Oden, 1994In presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives, It is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission.(Signature)__________________________________Department of_______________The University of British ColumbiaVancouver, CanadaDateDE-6 (2188)11AbstractThis study is an examination of the rate at which organic debris and clastic sedimentaccumulate in a gully after it is scoured by a debris torrent. Of particular interest is theeffect that a change in land use from old-growth to clear-cut conditions may have on theserates. This change should result in a reduction in the delivery of large organic debris(LOD), which is a major factor in sediment storage in gullies. It is hypothesized that thischange in land use, and the subsequent reduction in the LOD supply, should result in asignificant difference in debris recharge rates between old-growth and clear-cut gullies.Twenty-nine gullies in both land-treatment groups were sampled on the west coast ofthe Queen Charlotte Islands. Sampling procedures involved the estimation of the volume ofLOD and sediment in storage (normalized by the gully surface area) and the determinationof the time elapsed since the last debris torrent. These data were then used to estimaterecharge rates (m3ha’year1)of LOD, sediment, and total debris.Recharge rates of each material were compared between land-treatment groups usingthe nonparametric Mann-Whitney test. This test revealed that LOD has been delivered toold-growth gullies at a significantly higher rate relative to clear-cut gullies. There was nosignificant difference in sediment and total debris recharge rates between gullies in the twogroups, but this outcome was partially a result of the small samples and the different debrisrecharge times in each data set. Graphical representations of the data permitted theidentification of possible temporal trends in sediment and debris accumulation, which maybe strengthened with larger data sets.Debris recharge rates have several applications. The estimate of sediment volumestored in a gully can be used in the construction of local sediment budgets, as onecomponent of a watershed sediment cascade is quantified. The calculation of debrisrecharge rates will provide insight into the transfer rate of sediment from hillslopes to loworder channels and to the storage capacity of the channels. Finally, debris recharge ratescan be used to improve knowledge of the frequency-magnitude characteristics of debristorrents in an area.111Table of ContentsAbstract 11List of Tables viList of Figures viiAcknowledgements ixChapter 1 Introduction 11.1 Overview 11.2 Frequency-magnitude characteristics 21.3 Debris recharge rates 31.4 Importance of debris recharge rates 41.5 Preliminary hypothesis 61.6 Study objectives 8Chapter 2 Study Area 92.1 Overview 92.2 Rennell Sound 92.3 Physiography 132.4 Bedrock geology and surficial materials 232.5 Climate 252.6 Vegetation and logging history 282.7 Summary 29Chapter 3 Methods 313.1 Research design 313.2 Field Techniques 313.2.1 Volume of debris in a gully 313.2.2 Debris torrent dates 32iv3.3 Estimation of volume of large organic debris 343.3.1 Previous work 343.3.2 Techniques used in this study 363.4 Estimation of sediment volume 363.4.1 Previous work 363.4.2 Techniques used in this study 38Chapter 4 Study Results, Analysis and Discussion 534.1 Overview 534.2 Volume of debris in storage 534.2.1 Large organic debris volumes 534.2.2 Sediment volumes 554.2.3 Total debris volumes 564.3 Debris torrent dates 574.4 Recharge rates 584.5 Analysis of study results 604.5.1 Statistical techniques used to compare sample means 604.5.2 Large organic debris recharge rates 624.5.3 Sediment recharge rates 644.5.4 Debris recharge rates 684.5.5 Other factors considered 734.6 Consideration of erosion rates at a global scale 784.7 Summary 80Chapter 5 Conclusions and Recommendations 815.1 Overview 815.2 Applications in geomorphology 815.3 Implications for forest managers 82V5.4 Recommendations for future projects 835.5 Final remarks 85References 86Appendix A Air photograph sets used for debris torrent dating. 92Appendix B Assessment of accuracy of techniques used in estimation ofsediment cross-sectional area. 93Appendix C Estimated volume of material in storage. 94Appendix D Methods used and estimated dates of most recent debris torrents. 97Appendix E Prediction of magnitude of future debris torrents. 98viTablesBasin characteristics of the Rennell Sound area.Physiography and dominant lithology of the sampledgullies: old growth.Physiography and dominant lithology of the sampledgullies: clear-cut.Recharge rates of large organic debris, sediment, and total debrisin gullies.Comparison of large organic debris recharge rates between land-treatment groups.Comparison of sediment recharge rates between land-treatmentgroups.Comparison of debris recharge rates between land-treatmentgroups.Comparison of LOD recharge rates in clear-cut gullies in relationto cut-boundary proximity.Comparison of sediment recharge rates between the dominantgeologic formations.Comparison of sediment recharge rates in relation to channel andsidewall material in clear-cut gullies.Comparison of erosion rates derived from rates of individualprocesses.Estimated volume of large organic debris.Estimated volume of sediment.Estimated total volume of debris.Sidewall erodibility rating.Magnitude of future debris torrents in gullies.Table 2.1Table 2.2Table2.2(cont)Table4.lTable 4.2Table 4.3Table4.4Table 4.5Table 4.6Table4.7Table 4.8Table C.1TableC.2Table C.3TableE.1Table E.2172122596364697477787994959699100viiFigure 1.1Figure 1.2Figure 1.3Figure 2.1Figure 2.2Figure 2.3Figure 2.4Figure 2.5Figure 2.6Figure 2.7Figure 2.8Figure 2.9Figure 2.10Figure 2.11Figure 2.12Figure 2.13Figure 2.142571012141415161718192024272728FiguresOld-growth gully scoured to bedrock by debris torrent in thewinter (199 1-1992) prior to photograph.Typical old-growth gully with abundance of large organic debris.Old-growth gully with recently delivered large organic debris.Study area location and major physiographic regions of the QueenCharlotte Islands.Rennell Sound study area with sampled gullies.Bonanza Creek sample sites.Shelly Creek sample sites.Gregory Creek sample sites.Riley Creek and Shields Bay sample sites.Typical hilislope well-dissected by deep gullies.Typical clear-cut gully with scoured bedrock channel andsidewalls, and prisms of fine sediment at base of sidewall.Typical old-growth gully with stable, vegetated sidewalls.Sediment wedge in old-growth gully associated with debris jam.Generalized bedrock geology of Rennell Sound area.Annual precipitation in Rennell Sound area, 1977-1982.Monthly precipitation in Rennell Sound area, 1977-1982.Large landslide resulting from extensive windthrow of timberon ridgetop.Channel cross-section viewed from left sidewall, beforeexcavation.Channel cross-section viewed from left sidewall, afterexcavation.Example of cross-section of excavated sediment and trapezoidsused to calculate area.Example of cross-section of sediment wedge used to calculatearea of sediment fill.Example of cross-section in which trapezoidal bedrock channelwas assumed.Example of cross-section in which bedrock outcropsFigure 3.1Figure 3.2Figure 3.3Figure 3.4Figure 3.5Figure 3.64040414345were extrapolated. 46viiiFigure 3.7 Comparison of cross-sectional area determined by measurementand by estimation. 48Figure 3.8 Frequency distribution of measured and estimatedcross-sections. 48Figure 3.9 Example of sediment loading in gully and approximation ofsediment deposit dimensions. 50Figure 3.9a Actual width and depth of deposit. 50Figure 3.9b Width and depth of deposit integrated between knowncross-sections. 51Figure 3.10 Example of plot used to calculate volume of sedimentin a channel. 52Figure 4.1 Typical old-growth gully in which large portions of the LODare buried by debris and covered with moss. 54Figure 4.2 Typical old-growth gully with abundance of LOD. 55Figure 4.3a Frequency distribution of LOD recharge rates. 61Figure 4.3b Frequency distribution of log-transformed LOD recharge rates. 61Figure 4.4 Comparison of average sediment recharge rates and time sincethe last debris torrent. 66Figure 4.5 Total debris accumulation in gullies less than 15 years old. 70Figure 4.6 Total debris accumulation in gullies less than 15 years old,with fitted exponential functions. 71Figure 4.7 Total debris accumulation in gullies less than 80 years old. 72Figure 4.8 Abundance of LOD in clear-cut gully located in closeproximity to a cut boundary. 74Figure 4.9 Clear-cut gully of same age as gully in Figure 4.8, but locatedin open clear-cut area and lacking LOD. 75ixAcknowledgementsI would like to thank my supervisory committee: Dr. Michael Bovis for his support,patience and wit, and Dr. Olav Slaymaker for his interest and constructive reviews. Specialthanks go to Jim Schwab for providing useful data and suggestions.This work was supported by the Federal and Provincial Fish/Forestry InteractionProgram, and I owe thanks to Steve Chatwin and Dan Hogan for their interest and support.Thanks also to the numerous individuals on the Queen Charlotte Islands who went out oftheir way to help me during the field season: Mike Hennigan, Keith Moore, Brian Eccles,Greg Wiggins, Norm Nalleweg, Al Cowan, Ray Krag, Cohn Richardson, Mike England,Yoshio Nishimura, and the representatives of Husby Forest Products Ltd. Special thanks goto Lars Uunila for his assistance and fortitude in adverse conditions, and Scott Davidson forhis help and energy with a shovel.All of this would not have been possible without encouragement from my parents andtheir gift of independence. My deepest gratitude goes to Steve Rice for going far beyond thecall of duty.1Chapter 1 Introduction1.1 OverviewTraditionally, the term debris flow has encompassed all mass movement eventsthat involve a rapidly-moving, poorly-sorted, saturated mixture of elastic and organicmaterial. The debris typically travels as a slurry, with a steep front made up of largeboulders, followed by a more fluid mass of finer material. In the last twenty years inwestern North America, a distinct type of debris flow has been identified by manygeomorphologists and geotechnical engineers. The term debris torrent has been appliedto debris flows that involve significant amounts of organic debris, including logs andstumps, a low proportion of fine matrix material and abundant water (Swanston andSwanson 1976, Hungr et al. 1984, VanDine 1985). This type of mass movement iscommon in the coastal mountain belt of British Columbia.Debris torrents can occur in first- to fifth-order streams, but are most commonlyinitiated in first-order streams, also termed gullies. A gully is operationally defined hereas a long, linear depression in a hilislope, with a confmed channel and sidewalls higherthan three meters (Maynard, unpublished report). The longitudinal profile ranges fromgentle to steep and from uniform to stepped, while the channel in cross-section rangesfrom v-shaped to u-shaped with moderate to steep sidewalls. The channel may beperennial, ephemeral, or inactive.Debris torrents can be very destructive to both the natural and built environments.The moving mass entrains sediment and organic material (together referred to as ‘debris’)in its path, commonly leaving a channel scoured to bedrock in its wake (Swanston andSwanson 1976, Pearce and Watson 1983) (Figure 1.1). Obstructions in its path that aremore solid (e.g. roads, bridges) can be damaged or destroyed by a torrent, and with asudden loss of momentum, a plug of debris can be deposited downstream in a channel.This mass of material can disturb sensitive fish-spawning habitat and temporarilycontaminate municipal water supplies.21.2 Frequency-magnitude characteristicsAs a result of the destructive nature of debris torrents, a better understanding ofthe frequency and magnitude of future events is desired by many professionals in thecoastal belt of the province. In particular, forestry and fisheries managers seek thisinformation for assessment of the potential impact of debris torrents on downstreamchannels. This is of particular importance in British Columbia as many coastal streamsserve as habitat for salmon spawning.Debris torrent frequency is a function of the occurrence of triggering events,which are commonly of two types: (1) a level of stream discharge is reached whichmobilizes a critical proportion of the debris stored in the gully; or (2) a mass of debris isintroduced into a gully which has sufficient impetus to flow down the gully and entrainFigure 1.1 Old-growth gully scoured to bedrock by debris torrent in the winter(199 1-1992) prior to photograph. Southwest side of upper Gregory Creek.3stored material. Intense precipitation, or the failure of a natural debris dam, maygenerate the former scenario, while failure of a gully sidewall or an adjacent hillslope isgenerally responsible for the latter (VanDine 1985, Jordan 1987). In both cases, theavailability of stored debris in a gully is crucial to the development of a debris torrent.The magnitude of a torrent is a function of the volume of the mass triggering theevent, as well as the volume and stability of entrainable material stored in a gully. Thevolume of stored material can sometimes be the main determinant of torrent magnitude,as small slides can readily produce large torrents if debris is present in the channel. Thevolume of debris in storage will be controlled by the rate of debris accumulation in thegully, the length of the gully, and the time elapsed since the last debris torrent. Forexample, longer gullies and those with high volumes of material in channel storageshould produce larger torrents than shorter gullies lacking stored material. In addition, along time interval between events will allow more debris accumulation than a shortinterval, which should result in a larger torrent. Hence, a short gully with few storagesites and a short time interval since the last event should be characterized by a chronicflushing of sediment and small, frequent failures. In contrast, a long gully with manystorage sites and a long time interval since the last event, should produce events of alarger magnitude. The accumulation of debris between events is the main focus of thisstudy.1.3 Debris recharge ratesIn coastal gullies, the volume of debris in a channel is dependent on the rate ofaccumulation, or ‘recharge’, of sediment and organic debris. With consideration of thetime since the last debris torrent, the volume of debris can be used to estimate a ‘debrisrecharge rate’ for a gully. This assumes the channel was completely scoured of debris bythe last torrent. Therefore, debris recharge rates are controlled by the delivery of debrisfrom adjacent hillslopes and the storage characteristics of the channel.4Sediment is transferred from hillslopes to a gully by several processes, includingrockfall, debris slide, soil creep, and ravelling of loose material (Selby 1982). Once inthe gully’s channel, finer sediment is often transported downstream by ephemeral fluvialactivity, while coarser sediment may be trapped by stream-profile irregularities (e.g.bedrock knobs or steps) and channel obstructions. Of particular significance for sedimentstorage is large organic debris (LOD), operationally defined here as logs, stumps andbranches greater than ten centimeters in diameter (Hogan 1987). LOD is an efficientsediment trap and usually is abundant in gullies (Froehlich 1973, Megahan and Nowlin1976).LOD is delivered to a gully by a variety of mechanisms, including massmovement, logging, mortality, and windthrow (Swanson et al. 1976, Bisson et al. 1987).Windthrow and debris slides tend to be the predominant input mechanisms in coastalgullies, which are commonly littered with LOD (Figure 1.2). Once in a gully, LOD mayremain there for long periods of time, depending on the frequency of debris torrents inthe gully, which are the dominant output process. The establishment of LOD jams iscommon, and these readily cause deposition of sediment, producing sediment wedges(Keller and Swanson 1979). A series of log jams and wedges along a gully will result ina local widening of the channel floor and the development of a stepped longitudinalprofile. Jams govern the storage and release of sediment and are important storagecomponents in the sediment budget of forest streams (Swanson and Lienkaemper 1978,Marston 1982). Sediment budgets are used in an attempt to quantify the rates ofproduction, transport, and discharge of sediment from hilislopes to the highest-orderstream within a drainage basin.1.4 Importance of debris recharge ratesThere has been little research directed at quantifying the rate of debrisaccumulation in coastal gullies. This is unfortunate because an understanding of the5debris recharge cycle can be used to define frequency-magnitude characteristics of debristorrents, and to construct watershed sediment budgets.Many geomorphic processes display a strong relationship between event frequencyand event magnitude, with larger events generally less frequent than smaller ones. Fordebris torrents, this relationship is often difficult to determine, as field evidence of earlierevents can easily be destroyed by larger, subsequent events. As a result, event frequencycannot ordinarily be used to predict magnitude, and vice versa. However, if evidencedoes exist that allows the reconstruction of either magnitude (with units of cubic metersper gully surface area, m3ha’) or frequency of past events, the debris recharge rate(m3ha-lyrl) of a gully can be used to predict the frequency or magnitude, respectively,for that gully. For example, if the time interval (in years) between past events can beFigure 1.2 Typical old-growth gully with abundance of large organic debris. GullyG5, Gregory Creek. Torrent date: 1917.6determined for a given gully, then the debris recharge rate can be used to estimate aminimum magnitude of an event that could occur in a given year. If, on the other hand,only the magnitude of a past event can be estimated, then the debris recharge rate can beused to predict the number of years needed to recharge the gully with the same volume ofmaterial.Debris recharge rates are significant also because, with their estimation, a portionof one component of a sediment budget for a drainage basin can be calculated. Thecalculation of the rates involves the estimation of volume of sediment currently in storagein a gully, resulting in the quantification of the storage component of a low-orderchannel. In addition, the measured recharge rate will be an indicator of the rate at whichsediment is produced on a hillslope and transferred to a channel system.1.5 Preliminary hypothesisIt is well established that the amount of clastic sediment stored in coastal gullies isstrongly influenced by LOD conditions in a gully’s channel (Megahan and Nowlin 1976,Keller and Swanson 1979). Important characteristics are the total volume of LOD in thechannel, the orientation of LOD pieces relative to the channel axis, and whether theobstruction is a single piece or an accumulation of several pieces (Hogan, 1987).Because of the importance of LOD in sediment storage, a disruption of LOD supply to achannel should affect the accumulation rate of sediment in a gully.The process of windthrow, which delivers much of the LOD to guffies undernatural conditions, is especially active in old-growth and second-growth stands. As aresult, LOD supply to a gully in these stands is uninterrupted over time. Figure 1.3shows an old-growth gully which last torrented in 1986 (photograph was taken in 1992)and which contains a moderate amount of LOD. Only a portion of the LOD cancurrently trap sediment, but with time, the pieces bridging the gully will also play an7active role in sediment storage as they rot and collapse. In contrast, LOD supply to agully in a clear-cut area tends to increase during and immediately after harvest. Toewsand Moore (1982) report that logging activities produced an initial five-fold increase inthe amount of LOD, relative to that of undisturbed channels, over a six-year period.Subsequently, if no buffer strips exist along a gully, the future LOD supply would besignificantly reduced as the main debris source has been removed. This situation should8persist until the surrounding regenerated stand is old enough to contribute LOD to thechannel once more. Grette (1985) estimated that this could be fifty to sixty years in aconiferous stand.A change in land use from old-growth to clear-cut conditions should result in asignificant reduction in the recharge rate of LOD to a gully. Therefore, it is hypothesizedin this study that the change in land use should produce a difference in debris rechargerates between these two land-treatment groups.1.6 Study objectivesLOD supply and sediment recharge rates in gullies are controlled by many factorsincluding geology, climate, and land use on adjacent slopes. With the assumption thatother controlling factors are constant in an area, the overall objective of this project is toquantify the impact of a change from old-growth to clear-cut conditions on debrisrecharge rates. In order to accomplish this, several steps are involved, enumerated asobjectives below.The primary objectives of this study are:(1) To estimate the volume of debris currently stored in a selection of oldgrowth and clear-cut gullies.(2) To determine the elapsed time since the most recent debris torrent in thosegullies.(3) To determine the debris recharge rates, from (1) and (2).(4) To assess any differences in debris recharge rates between old-growth andclear-cut gullies.9Chapter 2 Study Area2.1 OverviewThe Queen Charlotte Islands lie 130 kilometers off the coast of British Columbiaand are part of the Insular Belt of the Canadian Cordillera. The Insular Belt includes twomajor subdivisions: the Insular Mountains of Vancouver Island and the Queen CharlotteIslands, and the Hecate-Georgia Depression. On the Queen Charlotte Islands, the InsularMountains are further subdivided into the Queen Charlotte Ranges and the SkidegatePlateau, and the Hecate-Georgia Depression is expressed as the Queen CharlotteLowlands (Sutherland Brown 1968) (Figure 2.1). The mountains of the Queen CharlotteRanges form the spine of the Islands, with most of the peaks reaching between 800 and1100 meters above sea level.The landscape of the Queen Charlotte Islands is highly susceptible to hilislopefailures (Alley and Thomson 1978, Rood 1984, Gimbarzevsky 1988). Rood (1984)estimated that an average of 43% of the total volume of sediment mobilized by hillslopefailures in forested and logged areas on the Islands directly enters stream channels. Atleast 200 of the streams provide habitat for pink, chum and coho salmon (Poulin 1984,Chatwin and Smith 1992). This sudden delivery of sediment by debris torrents has had adetrimental impact on anadromous fish habitat (Tripp and Poulin 1986, Hogan andSchwab 1991), by affecting stream morphology and sediment texture (Hogan 1989, Rice1990).2.2 Rennell SoundThe study area selected is located in the Rennell Sound area, located on the westcoast of Graham Island, on the western edge of the Skidegate Plateau (Figure 2.1). ThePlateau is a long, northwest-trending surface, that slopes gently to the east. Along theshores of Rennell Sound, the Plateau rises abruptly to 600 meters, and it is believed thatthis is a fault-line scarp associated with the Rennell-Louscoone Inlet fault zone (Alley and10Figure 2.1 Study area location and majorphysiographic regions of the Queen Charlotte IslandsGraham Island‘9/ Queen Charlotte7Study ç Islands—- Area’Rennell SoundMoresby Island133°W Tasu SoundBritishColumbia NPacific —Ocean 0 kilometers 5011Thomson 1978). The close proximity of Rennell Sound to this fault zone places it in oneof the most seismically active areas in Canada (Sutherland Brown 1968, Alley andThomson 1978).The Rennell Sound area is characterized by frequent hillslope failures, whichinclude debris slides, debris avalanches, debris flows and debris torrents. The frequencyof occurrence of these failures is much higher in clear-cut areas (22.10 km-2yr’) relativeto old-growth forested areas (0.39 knr2yr’), and the average volume of the failures in theclear-cut areas (1.81 m3ha-’yr’) is considerably higher than that in old-growth forestedareas (0.02m3ha-’yrl) (Rood 1984).The Rennell Sound area was chosen for three main reasons: (1) there arenumerous, accessible debris torrent tracks; (2) these are located in both clear-cut andold-growth tracts; and (3) the approximate dates of many debris torrents had previouslybeen determined (Hogan and Schwab 1991, J.W. Schwab pers. comm. 1992). Inaddition, any information gained concerning debris torrent activities and debris rechargerates would be useful for the assessment of impacts on the many valuable spawningstreams in the area.Gully selection was based on the pre-determined dates from Schwab (pers. comm.1992), and from air photograph and field reconnaissance, which were used to assessaccessibility and gully complexity. Air photographs revealed information on theerosional and depositional features associated with debris torrents, as well as on thestream order of the system. Gullies that had more than one channel entering a mainstem, and which were more than two-hours’ walking distance from a road, were avoided.Data were collected in twenty-nine gullies, thirteen in old-growth stands andsixteen in clear-cut areas, in July and August 1992. These gullies are located in Bonanza,Gregory, Riley, and Shelly creek basins, and along the eastern shores of Rennell Sound atShields Bay (Figure 2.2). Bonanza Creek trends northeast to southwest, and the sampledgullies are located on the southeast-facing slope (Figure 2.3). Shelly, Gregory, and RileyFigure 2.2 Rennell Sound study area withsampled gullies0 3kilometers*1-.// : Gregory Creek/-. .(Figure 2.5)*******12Bonanza Creek(Figure 2.3)“___._%___j__•-..Riley Crek±.. .*.. (Figure 2.6)Shelly Creek(Figure 2.4)Drainage———— Logging roadWatershed boundary* AES stationSampled gullies:* Old-growthr Clear-cutI/13creeks flow west into Rennell Sound, and the sampled gullies are located on the north-facing slope of Shelly Creek (Figure 2.4) and on both sides of Gregory and Riley creeks(Figures 2.5 and 2.6).2.3 PliysiographyThe watersheds of the Rennell Sound area are characterized by steep, confinedheadwater areas and broader, distal valleys. The surrounding hills and ridges are roundedand reach elevations of 840 meters. The slopes above the valley floor are steep, rangingfrom 20° to 800, and are well-dissected by deep gullies entrenched in bedrock,colluvium, and till (Figure 2.7). The valley of Bonanza Creek is generally deeper thanthe others and is flanked by longer slopes, resulting in longer debris torrent tracks (anaverage of 550 meters long, compared with 260 meters in the other basins).Furthermore, the wide valley floors adjacent to Bonanza Creek and Lower Riley Creekbuffer the stream channels from debris torrents, and the most recent events have notreached these creeks. In contrast, most of the gullies in the upper sections of thesebasins, and in the other basins studied, have delivered torrents directly into the mainchannel. The details of each basin are summarized in Table 2.1.The sampled gullies are predominantly trapezoidal in cross-section, and range indepth from 3.5 to 9 meters, with an average depth of 5 meters. Longitudinal profiles aregenerally stepped, as a result of both bedrock irregularities and debris jams. Channellongitudinal slopes average 270, sidewalls average 430, and gully headwalls approach60°. At the time of data collection (June-August), some of the channels had a smalltrickle of water, while others were dry but contained evidence of recent fluvial activity.The clear-cut gullies typically have bare bedrock sidewalls and debris slide scars,both of which are prone to dry ravelling (Figure 2.8). Sediment that enters the channel isprimarily trapped by bedrock irregularities, a few stumps, and small pieces of organic14Figure 2.3 Bonanza Creek sample sites\\ •‘N__\ ‘( ‘>. 3 // 1 / 500 1007\ I & \ 131 \ ?\ \“1/.... .... N. \\J !c N V 1 \,;/fl3\/\. —--/.‘ .\ ‘I/ \ 4- \>\ L NNL—K\ \\ \\ .K \ \ \ —.. .... ... ......,... ...N-. \ .2\ \ I‘N \\\N\ N \ \ N\ \.‘. ...\0 kilometer 1Contour interval lOOm‘. Sampled gullyDrainage—— Logging road.4 CutblockFigure 2.4 Shelly Creek sample sitesN jJJJI L-1--ii-c x—-\-‘100 — — —I..300 500 Si00 kilometer 1 \Contour interval lOOmFigure25GregoryCreeksamplesites\Sampledgully*Drainage-—.LoggingroadCutbiock000kilometer1ContourintervallOOm(110kilometer1Contourinterval1OOmFigure2.6RileyCreekandShieldsBaysamplesitesShieldsBaySampledgullyDrainageLoggingroad17Figure 2.7 Typical hilislope well-dissected by deep gullies. South side of lowerRiley Creek.Table 2.1 Basin characteristics of the Rennell Sound area.basin namedrainage loggedarea area(km2) (2)relief(m)steeplandareaZ(2)loggingextent(% ofbasin area)Bonanza CreekGregory CreekRiley CreekShelly CreekShields Bay45.235.528.34.9number ofgulliessampled5. Basin area steeper than 200, determined from 1:50,000 NTS maps (after Rood 1984).18debris. LOD is sparse, and because of the lack of storage sites, it seems that the coarsersediment is transported downstream during the winter rainstorms. These conditions resultin a bedrock channel flanked by small prisms of sediment that have developed primarilyfrom sidewall ravelling.Old-growth gullies typically have stable, vegetated sidewalls, composed ofbedrock or colluvial material with debris slide and slump scars (Figure 2.9). Commonly,sediment that enters the channel is readily trapped by LOD, as well as by bedrock steps.LOD is abundant, and in those gullies that have not been scoured by torrents for severaldecades, a jumble of branches and debris is common. Debris jams are frequent and oftenproduce sediment wedges upstream (Figure 2.10). Where active streams are present, thedepth of accumulated debris is often revealed by fluvial incision to bedrock. Dry gulliesFigure 2.8 Typical clear-cut gully with scoured bedrock channel and sidewalls, andprisms of fme sediment at base of sidewall. Gully R13, Riley Creek.Torrent date: approximately 1984.19Figure 2.9 Typical old-growth gully with stable, vegetated sidewalls. Gully G9,Gregory Creek. Torrent date: 1978.generally contain mossy debris in the channel, with occasional bedrock outcrops. Thephysiographic details of the sampled old-growth and clear-cut gullies are listed in Table2.2.Figure 2.10 Sediment wedge in old-growth gully associated with debris jam. View isdownstream; bedrock channel in foreground. Gully G4, Gregory Creek.Torrent date: 1978.20Table2.2Physiographyanddominantlithologyofthesampledgullies:oldgrowth.gullymeanmeanmeaninitiationdominantchannel!surfacechannelchannelsidewallsidewallzonerocksidewallgullyzarea(ha)blength(m)cslope(deg)clength(m)Cslope(deg)celevation(m)typedmaterialeGi0.20218299.24119511G20.29341238.44621311G30.19232178.44512212G40.20235178.34514612G50.24247239.93826811G60.615121712.05024411G90.31410217.64325911GlO0.181802510.14212531Gil0.322802811.34114031G120.0550359.04118931R40.28352248.04130511R50.16296295.44743011Slog!0.141183811.85342111a0=GregoryCreek;R=RileyCreek;S=ShellyCreekbproductofaveragesidewallheight andgullylengthcinitiationandtransportzonesoflastdebristorrent(afterRood1984)dafter Hesthamineretal.1991,mapscale1:50,0001=YakounFormation-shale,siltstone,sandstone, andesite;2=Plutons-quartzdiorite, quartzmonzodiorite;3=KungaGroup-shale,argillite,limestone,sandstonee1=bedrock;2basaltillfSiinitiatedinoldgrowthandtorrentedthroughclear-cutarea;twosegmentstreatedseparatelyTable2.2(cont)Physiographyanddominantlithologyof thesampledgullies:clear-cut.datesgullymeanmeanmeaninitiationdominantchannel!adjacentsurfacechannelchannelsidewallsidewallzonerocksidewallslopesgully’area(ha)blength(m)’slope(deg)clength(m)cslope(deg)Celevation(m)typedmaterialeloggedBi0.61895246.846430111979B20.17232197.551213111977B30.43780245.542488111978-79B40.17276206.245201111978-79R20.06158304.147168211975-77R30.19255327.646162211975-77R70.11341273.339341111974R80.23318297.343348121974R90.15331274.643390111974RiO0.10310275.242250121978R120.20386215.135207121974-78R130.04245271.550165121974-75R140.06152293.737122111974-75Slccf0.10180325.339299111975SB10.05134273.440128211973SB20.07190253.731183221973aB=BonanzaCreek;R=RileyCreek;S=ShellyCreek;SB=ShieldsBaybproduct of averagesidewall height andgullylengthcinitiationandtransport zonesoflast debristorrent(afterRood1984)dafter Hesthammeretal.1991, mapscale1:50,0001=YakounFormation-shale, siltstone, sandstone, andesite;2=Plutons-quartzdiorite, quartzmonzodioritee1=bedrock;2=basal tillfSiinitiatedinoldgrowthandtorrentedthroughclear-cutarea;twosegmentstreatedseparatelyt3232.4 Bedrock geology and surficial materialsThe bedrock geology of the Rennell Sound area was originally described andmapped by Sutherland Brown (1968), and more recently updated by Hesthammer et al.(199 1) (Figure 2.11). The Rennell Sound area is dominated by the Jurassic YakounFormation (JY), made up of weak volcanics and sediments. This formation has amoderate to high degree of fracturing, and the dominant rock types include shale,siltstone, sandstone, and andesite. Also present is the Kunga Group of the SandilandsFormation (TrJs), a sedimentary unit of Triassic and Jurassic age. This formation alsohas a moderate to high degree of fracturing and is well jointed. Dominant rock types areshale, argillite, limestone and sandstone. Minor outcrops of post-tectonic plutons (Jp)occur, consisting primarily of quartz diorite and quartz monzodiorite of Jurassic age(Hesthammer et al. 1991). The Masset Formation (Tv) and a Cretaceous Sandstone (Kss)are present within the study area basins but do not crop out in any of the sampled gullies.All of the formations weather rapidly and have a low resistance to surface erosion (Alleyand Thomson 1978; Lewis 1985). The dominant rock types in each sampled gully aresummarized in Table 2.2.The Queen Charlotte Islands were intensively glaciated during the FraserGlaciation by a locally generated ice sheet, which coalesced with the Cordilleran IceSheet to the east (Sutherland Brown 1968). All of the valleys in the mountain and plateauareas are U-shaped and many of the tributary valleys are hanging relative to the mainvalleys. There are cirques at various elevations, from sea level to 1100 meters, and theyare oriented in all directions. Smaller glacial features include rock drumlins, fluting,roche moutonnées, and striations. The glacial deposits consist of till, marine drift, andstony clays and outwash sands. These are exposed in bluffs along the eastern shores ofthe Islands and in road cuts and some stream banks around the Islands (Sutherland Brown1968). Most of the steeper slopes now lack glacial deposits, and only a few of the gulliessampled still retain a till veneer. In these gullies, till occurs at lower elevations, isCD CD C CDcr CD CD N CD I(_•)< 0Tj<<C)°0—I_I 11’C025discontinuous, and is approximately 1-3 meters thick. The texture is clay-loam to sandy-loam, depending on the parent bedrock type and the degree of soil development (Alleyand Thomson 1978).Since deglaciation, the combination of steep slopes and highly weatherablebedrock has produced a thick blanket of colluvium, with sandy-loam to loamy-sandtexture, over much of the landscape (Alley and Thomson 1978). Colluvium is thedominant surficial material of the study area and is commonly present in gully sidewalls.Debris torrents from gullies have reworked these colluvial blankets to produce colluvialaprons and fans in most of the study basins (Townshend 1979, Wilford 1984). Thesedepositional areas preserve the most coherent record available of recent debris torrentactivity.2.5 ClimateThe Queen Charlotte Islands are characterized by a humid, cool-temperateclimate, with mild winters and cool summers. The monthly mean temperatures rangefrom +1°C to +4°C in January and + 13°C to +15°C in July. The annual precipitationis quite variable around the Islands because of orographic and rain-shadow effects. Atsea level it ranges from 1,200 mm on the east side of the Queen Charlotte Lowlands tomore than 4,200 mm on the west side of the Queen Charlotte Ranges. The west coast isconsidered to be one of the wettest places in Canada, and at the higher elevations, annualprecipitation has been estimated to be greater than 7,000 mm (Williams 1968). Theseestimates are based on data collected since 1963 by the Atmospheric Environment Service(AES), which operates several climatological stations on the Islands. Of these, the onlypermanent AES station located on the west coast is at Tasu Sound (Figure 2.1),approximately 60 kilometers south of Rennell Sound. The extrapolation of Tasu Sounddata to Rennell Sound is not desirable because of the extreme spatial variability ofprecipitation; as a result, long-term climatological records are lacking for the Rennell26Sound area.Short-term records exist for the period 1977-82 from three temporary stationsmaintained by AES (Figure 2.2). Over this period, the monthly mean temperaturesranged from -0.5°C to +7°C in January and from + 10.5°C to + 14°C in July. Themean annual precipitation ranged from 2,225 mm near sea level to nearly 3,000 mm at250 meters (Figure 2.12). The majority of this precipitation occurred as rain betweenOctober and April, with October and December generally being the wettest months(Figure 2.13). It is believed that these precipitation values are underestimates because ofthe locations of the stations, and that the actual values would have exceeded 4,000 mm onslopes that serve as significant topographic barriers to storms, such as the shores ofRennell Sound (Wilford and Schwab 1982, Schwab 1988).With regard to slope failure initiation, the duration and intensity of the stormscharacteristic of the Rennell Sound area are considered by some to be more importantthan precipitation amounts (Wilford and Schwab 1982, Rood 1990). Schwab (1988)analyzed precipitation data for an eleven-year period and discovered that rainstormswhich triggered slope failures in the Rennell Sound area generally delivered more than130 mm in 24 hours, and occurred almost every year in the period 1974-1984. Withinthis ten-year period, the most significant storm was a five-day event in October andNovember 1978, which delivered more than 400 mm of rain and triggered more than 260slope failures in the Rennell Sound area. The estimated recurrence interval of this eventon the west coast of the Queen Charlotte Islands is ten to twenty years (Schwab 1983).The west coast is also considered to be one of the windiest places in Canada, withmaximum hourly wind speeds averaging 33 kmhr1 over a three-year period (Wilford andSchwab 1982) and recorded wind gusts exceeding 190 kmhr’ (Alley and Thomson 1978),These high winds are responsible for very extensive and destructive incidents of timberblow-down in the Rennell Sound area (Figure 2.14).27Figure 2.12 Annual precipitation inRennell Sound area, 1977-198240003500‘‘3000 .1977 1978 1979 1980 1981 1982 MeanYear ArmualPrecipitation____Camilla Station:. Gospel Point Station N 1 Penthouse Stationelevation = 251m a.s.1. elevation 34m a.s.l. I I elevation 378m a.s.1.Figure 2.13 Monthly precipitation inRennell Sound area, 1977-1982500‘—450400350300.200. .....i1I II1I1I1I IJan Feb Mar April May June July Aug Sept Oct Nov DecMonth282.6 Vegetation and logging historyRennell Sound lies within the Coastal Cedar-Pine-Hemlock Biogeoclimatic Zone(Wilford 1984), the most productive forest zone in British Columbia (Valentine et al.1978). Stable slopes support stands of shore pine (Pinus contorta), western hemlock(Tsuga heterophylla), western red cedar (Thuja plicata), and yellow cedar(Cha,naecyparis nootkatensis). Less stable slopes support Sitka spruce (Picea sitchensis)and western hemlock (Schwab 1988). Disturbed slopes with bare mineral soil, such asdebris torrent tracks, are most commonly colonized by the successional species red alder(Alnus rubra), which helps stabilize slopes and reduce erosion. Logged slopes areinitially revegetated primarily by shrubs, herbs, bryophytes and young conifers, whileFigure 2.14 Large landslide resulting from extensive windthrow of timber onridgetop. North side of Shelly Creek.29stream banks and valley flats support a dense cover of these plants together with red alder(Roberts and Church 1986).There has been logging on the slopes of the Rennell Sound area since the early1970’s. Of the study basins, Bonanza Creek has the greatest area of clear-cuts (5.0 kIn2),while Shelly Creek is the most extensively logged (18% of the basin area) (Table 2.1).The road systems are extensive, particularly in the Bonanza Creek basin (Figure 2.3), andmany failures in the area are associated with these roads. Rood (1984) estimated thatroad-related debris flows occurred seventy times more frequently than events in forestedareas. However, of the sixteen clear-cut gullies sampled, only three had debris torrentstriggered by road failures.The yarding methods used in the area include high-lead and grapple yarding, andmore recently, large-capacity helicopters. There is concern that forest harvestingpractices have increased the incidence of hillslope failures in the Rennell Sound area(Schwab 1983, Rood 1984, Schwab 1988), and less damaging harvesting practices arecurrently being investigated. For several years, the Fish/Forestry Interaction Program, incooperation with the Forest Engineering Research Institute of Canada, has been assessingthe potential for large-capacity helicopters to log steep slopes without creating thehillslope disturbances more conventional methods create. Cutblocks are divided intotreatment groups with various levels of partial-cut logging (10-50% harvest) as well asclear-cutting. Sediment movement on slopes and within gullies, and the presence of slopefailures after yarding are being monitored (Chatwin and Smith 1992, R. Krag pers.comm. 1992).2.7 SummaryRennell Sound is characterized by glacially over-steepened slopes, fractured andjointed bedrock, frequent seismic activity, intense precipitation and high winds. Theseenvironmental conditions work together to promote high rates of sediment production andwidespread hilislope instability which has been accelerated locally by logging practices.This rapid pace of geomorphic activity provides an opportunity to quantify debrisrecharge rates in numerous gullies, and to assess the impact of adjacent land use on theserates.3031Chapter 3 Methods3.1 Research designIn order to assess the effect of a change in land use on debris recharge rates ingullies, two primary assumptions were made: (1) that all other controls on rechargerates, such as relief, geology, climate and glacial history, are fairly homogeneous withinthe study area; and (2) that the last debris torrent scoured the channel to bedrock. It isimportant to note that the clear-cut gullies examined had torrented at least once since thecompletion of harvesting operations. This criterion is essential for the assurance that thedegree of LOD loading in a channel would accurately reflect a long-term change fromunlogged to logged conditions, rather than the initial increase of LOD supply to a channelresulting from forest-harvesting operations alone.3.2 Field techniquesDebris recharge rates are based on the measurement of three variables: thevolume of LOD in a channel, the volume of sediment stored in a channel, and the timeelapsed since the last debris torrent. The estimation of sediment volume requires visiblebedrock in a channel and an assumption that all of the material present in the channel,down to the bedrock profile, has entered the channel since the last debris torrent. Theestimation of the time since the last event requires the dating of field evidence, includingscars visible on air photographs and vegetation growing in the channel and on the torrentdeposit. A detailed discussion follows on the methods used to measure debris volume ina gully and those used to determine the date of a debris torrent.3.2.1 Volume of debris in a gullyThe transport zone of the most recent debris torrent was identified in the field asthe section between the start of channelization (within or near the initiation zone) and thestart of the deposition zone, defined by evidence of deposited material and channel slope32(Rood 1984). It was assumed that a torrent will lose momentum and begin to deposit onslopes of less than 150.The measurement of the total volume of debris in a channel involved estimation ofthe volume of LOD and the volume of sediment in storage. The calculation of LODvolume is based on the dimensions of length and average diameter of single pieces andjams. For each LOD obstruction, details of the following qualitative observations werenoted: configuration (single piece or jam); orientation (parallel, perpendicular or anobtuse angle to the channel axis); age (as an indicator of strength and stability); and anestimation of the volume of sediment trapped.The measurement of sediment stored involved detailed surveys of channelconditions, including longitudinal profile, cross proffles at 25 meter intervals, sidewallangles and heights, and the dimensions of sediment wedges. In some of the gullies, partsof the sediment fill were excavated to bedrock. The longitudinal profile was surveyedusing a hip-chain, abney level and ranging pole; channel cross-sections were measuredwith a 30 meter tape and stadia rod. At each cross-section, sidewall angles weremeasured with a Brunton compass, and sidewall heights were measured either with a tapemeasure or were estimated from a stadia rod lying parallel to the wall surface. Inespecially deep gullies, a photograph of two ranging poles lying end-to-end on the wallwas used to estimate sidewall height. Sediment sources, sidewall failures, and theidentification of the sidewall material were noted. Finally, the dimensions of theinitiation zone were measured, and the volume of recent slumps and slides into thechannel were visually estimated. These data were supplemented with an extensive set ofground photographs illustrating channel conditions.3.2.2 Debris torrent datesThe dating of debris torrent events is central to the problem of estimating debrisrecharge rates in gullies, but knowledge of the exact timing of a past debris torrent is33difficult to acquire. Some of the techniques only reveal a range of possible dates ratherthan an exact year. As a result, four sources of information were combined in this study.First, previous work was consulted. Schwab (pers. comm. 1992) used air photographand dendrochronologic records to date and map more than 300 hillslope failures,including debris torrents, in Bonanza, Gregory, and Riley creek basins and along theshore at Shields Bay. Using information about Rennell Sound storm activity obtainedfrom newspapers, with precipitation data as corroborating evidence, Schwab was able toidentify four years in which large storms had produced significant numbers of hilislopefailures (1891, 1917, 1935 and 1978). Hogan and Schwab (1991) estimated that morethan 85% of all landslide-derived debris entering Gregory and Riley creeks occurredduring these four storms. In this study, air photograph and field reconnaissance wereused to assess the failures mapped by Schwab, and this resulted in the identification ofless than forty debris torrent tracks suitable for data collection.Schwab’s torrent dates and map proved to be invaluable starting points for torrentdating, but several of the clear-cut gullies in his study had re-torrented since his work wascompleted in 1989. As a result, air photographs, dendrochronology, and interviews withknowledgeable people in the area, were used to date gullies that had re-torrented, as wellas those selected for the project which had not been dated by Schwab.A chronological sequence of air photographs, dating from 1933 to 1989(Appendix A), was used to identify the time span during which an individual event hadoccurred. By studying a sequence of photographs taken over one location andinterpreting any visible changes, it was possible to identify the most probable date for arecent debris torrent (e.g. when a slide scar first appeared on a photograph, or when theheadwall region of a gully moved upslope). The date of the photograph immediatelyprior that year, on which there was no visible change, was then used as the earliest yearin which the torrent could have occurred. For example, if a debris torrent scar is visibleon a 1981 photograph, but not on a 1977 photograph, it can be deduced that the torrent34occurred in the period 1977 to 1981.In addition to Provincial air photographs, oblique air photographs were used forselected gullies in Bonanza, Gregory, and Riley creek basins and along the shore atShields Bay. These photographs were obtained from a helicopter flight in 1992.Tree-ring dating was used to confirm the individual dates of Schwab and tonarrow the time intervals established from the air photographs. Torrent dates wereestimated by counting growth rings and interpreting disruptions in growth patterns.These disruptions were evidenced on a tree’s surface as a scar in the cambial tissueresulting from the impact of a piece of organic or clastic debris. Tree cores were taken tocount the growth rings and to determine the time elapsed since the tree surface wasscarred. These were obtained using a Swedish increment borer. Red alders growing in adebris torrent initiation zone, in the transport zone, and on the deposit were dated, as thisspecies is one of the first to germinate soon after a failure has exposed bare mineral soil(Alley and Thomson 1978, Wilford and Schwab 1982). It was assumed that there wouldbe at least a one-year lag time between torrent occurrence and the germination of redalder. Within a gully, all red alders available were dated and compared to minimizeerrors resulting from the presence of double or false growth rings in any single tree.Finally, several individuals who regularly travel on the remote Rennell Soundroads were interviewed, and they were able to provide detailed information about themore recent debris torrent occurrences. These people were able to recall, or had recordof, the precise year of an event’s occurrence, and include Ministry of Forests employees(e.g. recreation managers and road engineers) and loggers. Their accounts helped to datethe most recent events for which all other evidence was lacking.3.3 Estimation of volume of large organic debris3.3.1 Previous workThe importance of LOD in steep-channel dynamics in the coastal belt has received35much attention in the last twenty years (Froehlich 1973, Swanson et al. 1976, Bisson etal. 1987, Sedell et a!. 1988). LOD delivery to a channel is of particular importance todebris recharge in gullies since the partial blockage of a gully at LOD-load pointsprovides many sites for sediment build-up.The techniques used by others to estimate LOD input and volume are reviewedhere. Swanson and Lienkaemper (1978) evaluated the delivery of LOD to Cedar Creek,in the Coast Ranges of Oregon, in two ways: (1) by considering the residence time ofindividual pieces in a channel (determined by dendrochronology); and (2) by examiningthe LOD loading of streams flowing through forest stands in different stages of firerecovery. For the latter approach, debris size and residence time in a stream were used,among other things, to determine whether the LOD had entered the stream before or afterthe fire. In turn, it was estimated that in first-order streams, nearly 120 years wererequired for the LOD supplied from a post-fire stand to match the LOD supplied from apre-fire stand (Swanson and Lienkaemper 1978).In a study of the effects of streamside logging on LOD input to Carnation Creek,on the west coast of Vancouver Island, Toews and Moore (1982) monitored debrisloading annually for a period of six years. Debris loading was assessed in two ways: (1)by counting the number of new pieces (3 meters or longer) entering a channel reach 50-75 meters long; and (2) by monitoring the changes in LOD volume per unit area of thecreek over a given year. The volume of LOD was estimated by summing the volume ofindividual log cylinders. Pearce and Watson (1983) quantified the effect of twolandslides on the development of log jams in first- and second-order streams. The totalvolume of LOD in a jam was calculated as a rectangular prism estimated from the heightand length of the jam. The measurement of jam width was not described in their paper.Lienkaemper and Swanson (1987) recorded LOD input to old-growth streams overa ten-year period by annually mapping the distribution of individual logs and LODaccumulations. Individual pieces were recognized from year to year based on their36shape, location, and distinguishing characteristics. Volume, V, of a single piece of LODwas calculated using the formula for the frustum of a paraboloid:V = ‘/8 [ir (D12 + D2)*L] [3.1]where D1 and D2 are the diameters (m) at each end and L is the length (m) of the piece.A modified version of this equation was used in this project, as described in the followingsection.3.3.2 Techniques used in this studyData collection included a detailed account of all LOD encountered during gullysurveying, and was used to calculate volume of LOD in each of the sampled gullies. Thevolume of a single piece, V, was calculated using a modified version of equation 3.1,which utilizes a single estimate of log diameter, D:jl = 1/4 [ir D2 *L] [3.2]where D is average diameter and L is length.The volume of LOD included in a jam, Vj, was initially calculated as a rectangularprism, after Pearce and Watson (1983):= WHL - (1/2L*tanj3*W) [3.3]where W is the width of the jam (average of surface width and base width); H is theheight of the jam; L is length of the jam; and fi is the channel slope angle, extrapolatedfrom slope angle measurements upstream and downstream of the jam. This volume wasthen corrected for sidewall angles by subtracting triangular prisms from each side, basedon an extrapolation of the sidewalls to the base of the jam. In each gully, total LODvolume was calculated as the sum of all individual and jam volumes.3.4 Estimation of sediment volume3.4.1 Previous workThe volume of sediment in storage in a gully is difficult to measure accurately37because of the fundamental problem of determining the depth of the sediment. While it isrelatively straightforward to monitor changes in storage volume by repeated cross-sectional surveying of a channel surface (Megahan and Nowlin 1976, Millard 1993),estimating the total volume of stored sediment is dependent on locating the underlyingchannel base. Several studies have involved the estimation of volume of sediment wedgesassociated with in-channel obstructions (Hogan 1987, Nakamura and Swanson 1993) andlong-term disturbances (Roberts and Church 1986).A sediment wedge upstream of an obstruction has been most commonlycharacterized as a rectangular prism (Megahan 1982, Hogan 1987, Nakamura andSwanson 1993), with dimensions obtained by detailed surveying. The depth of eachwedge is often difficult to determine if minimal bedrock is visible, but in all cases it wasassumed to be equal to the difference between the top of an LOD jam and the base of thejam at the downstream end of the wedge. Megahan (1982) and Nakamura and Swanson(1993) identified wedge length as the distance from the upstream end of the obstruction tothe upstream end of the wedge. Hogan (1987) defmed length as the distance from the logjam to the point where a line of average channel slope would intersect the sedimentsurface, as the upstream end of the wedge was sometimes difficult to identify. Hogandetermined wedge width from detailed topographic maps (1:200) made from his surveydata, while Megahan (1982) and Nakamura and Swanson (1993) used an average value ofwidth measurements made along the wedge. The rectangular prism model assumesvertical sidewall angles and therefore needs to be modified if the sidewalls are inclinedotherwise.Pearce and Watson (1983) estimated the volume of a sediment wedge upstream ofa log jam as a triangular prism, tapering upstream. The surface area of the wedge wassurveyed, but the method used to determine each of the critical dimensions is notdescribed in their paper. Channel slopes above and below the wedge and jam weremeasured, then extrapolated beneath the sediment wedge and incorporated into the final38calculation of volume. Their study area is characterized by near-vertical gully sidewalls,and again it was assumed that the walls extended vertically to the bedrock channel basebelow the sediment wedge.Roberts and Church (1986) characterized a sediment wedge as a rectangular prismand as a semicircular prism, to accommodate the fact that most channels are moresemicircular than rectangular in cross-section. Both models were modified to incorporatechannel slope, stream incision, and antecedent storage factors. First, the profile of thebedrock surface below the wedge was assumed to be smoothly concave between bedrockcontrol points, and a best-fit mathematical curve describing survey points and bedrockoutcrops was defined. The survey data of the wedge were overlaid on the modelledcurve, and these plots were used to estimate the length and depth of a wedge. Width ofthe sediment wedge was measured from air photographs, and sediment volume wascalculated for each wedge model using the estimated dimensions. Next, the volume ofeach wedge model was modified in three ways: (1) from the rectangular model, twotriangular wedges near each streambank were subtracted to incorporate repose angles ofthe sediment; (2) the volume occupied by a surveyed incised stream was subtracted fromboth models; and (3) the volume of material stored behind log steps existing before thesediment wedge developed was subtracted from both models. Finally, total volume of thesediment wedge was calculated for both the rectangular and semicircular models (Robertsand Church 1986).3.4.2 Techniques used in this studyWithin an individual gully, the width and depth of sediment accumulation zonesare highly variable, as the slope of the sidewalls varies non-systematically and the gullybase is often irregular. As a result, the simple shape approximations which characterizethe above techniques used to quantify sediment volume were considered to beinappropriate, as large errors would have resulted. Instead, the volume of sediment in a39given gully was calculated using multiple measurements and estimates of the cross-sectional area of the sediment fill, along with an approximation of the bedrocklongitudinal profile. The calculated sediment volume includes buried branches, twigs androots.Measurement ofsediment cross-sectional areaThe cross-sectional area of the sediment at a site was measured by excavating atrench in the fill using picks and shovels. Fourteen cross-sections in clear-cut and sixteencross-sections in old-growth gullies were excavated down to bedrock. Within a gully,sites were selected based on the feasibility of two people completing the excavation withhand tools.At each excavation site, the surface profile of the channel fill was surveyedperpendicular to the channel axis at 25 centimeter intervals, using a 30 meter tape and astadia rod. The site was then excavated to bedrock, and the exposed bedrock surface wassurveyed in the same manner as the sediment surface (Figures 3.1 and 3.2).The cross-sectional area of each of the excavation sites was determined from anoverlay of the sediment and bedrock surface profiles. The resultant sediment polygonwas divided into trapezoids, and the area, A1, of each trapezoid was calculated from:A1 = (a/2 + b/2)*c [3.4]where a and b are the lengths of the parallel sides and c is the horizontal width of a giventrapezoid (Figure 3.3). Trapezoids were then summed to yield the total cross-sectionalarea of the sediment fill.The method of excavation was very time-consuming, and so the data obtained bythis direct measurement are limited. As a result, detailed information is available foronly a few points in a gully that is typically several hundred meters long. It wastherefore necessary to supplement the measured data with estimated cross-sectional data.40Figure3.1Channelcross-sectionviewedfromleftsidewall,beforeexcavation.GullySlcc,ShellyCreek.Torrentdate:approximately1984.Figure3.2Channelcross-sectionviewedfromleftsidewall,afterexcavation.GullySlcc,ShellyCreek.Torrentdate:approximately1984.Figure3.3Exampleofcross-sectionofexcavatedsedimentandtrapezoidsusedtocalculatearea0.0-0.5-1,0-1.5-2.023Distanceacrosschannel(m)014542Estimation ofsediment cross-sectional areaWithin each gully estimates of cross-sectional area of sediment fill were based onthe survey measurements of channel cross-profiles and the longitudinal profile. Cross-sectional areas were estimated using four methods: (1) the cross-profile dimensions ofsediment wedges were plotted with gully sidewall angles; (2) polygons of sediment wereidentified on plotted cross-sections by extrapolating bedrock profile trends, and the areasof these polygons were graphically integrated; (3) bedrock cross-sections in the channelwere identified as sites of zero sediment storage; and (4) linear regression was used topredict the area of sediment at the start of channelization of the gully and the start of thedeposition zone of the most recent debris torrent. Methods (1)- (3) were appliedopportunistically at as many sites within each gully as possible, while (4) was applied tothe entire data set, consisting of the estimated and measured areas within a gully, in orderto identify cross-sectional area at the end-points of the longitudinal profile.(1) Sediment wedgesWidth and depth measurements at the downstream end of a sediment wedge werecombined with measured gully sidewall angles to reconstruct the cross-sectional area ofthe sediment fill. Sidewall angles were extended linearly beneath the sediment surface,and cross-sectional area, A, was calculated either as a triangle or as a trapezoid, using theformula:A = w*d- {[‘/2d*(1/tanflj)] + [1/2d*(1/tanfl) } [3.5]where w is width of the sediment wedge, d is depth of the wedge, and Jj and fi2 are thegully sidewall angles (Figure 3.4).43Figure 3.4 Example of cross-section of sedimentwedge used to calculate area of sediment fill- :. / f32/‘. . . .: .: /‘.. . . . . ....- . . . •.:. ... . .. .. I• . .. .. . . .... ///‘ )/• :. . .- .. //— — — :_ — .._‘_ — _•___ —.— — I/‘ /I/(2) Bedrock profile trends and sediment polygonsCross-sectional information was gathered every 25 meters along each channel.The data include the cross-profile of the sediment surface and a note identifying thesurface type (e.g. bedrock, sediment, woody debris) at each measurement station, whichwere 25 centimeters apart across the channel. All cross-sections were plotted and anybedrock outcrops in the channel bottom or sidewalls were identified. On the cross-sections, outcrops were then connected to approximate the bedrock profile beneath thesediment fill. For cross-sections which had only small bedrock outcrops, a trapezoidalw Width of sediment wedged Depth of sediment wedge131, [32 Sidewall angles—- Sidewalls extended belowsediment surface] Sediment fillcross-section was assumed (Figure 3.5). Typically, channel-bed outcrops were extended44laterally to intersect each other or linear projections of observed bedrock sidewalls(Figure 3.6), the angles of which were measured in the field with a clinometer. Polygonsof sediment were thus isolated, and the area of each polygon was estimated using the dotplanimeter method. The areas were summed for total cross-sectional area of sediment.While linear connections between exposed bedrock points are approximations, it isassumed that any resultant over-estimation of sediment area is compensated by equallyprobable under-estimation.(3) Bedrock cross-sectionsLack of sediment accumulation is just as significant in defining the volume ofsediment in a channel and its spatial distribution as is the presence of sediment.Therefore, those sites and sections in the channel that were primarily bedrock wereassigned a cross-sectional area value equal to zero.(4) Linear regressionThe fmal method used in the absence of measured or estimated values of thecross-sectional area at the start and end of the sampled section (transport zone) was linearregression analysis. The downstream distance from the initiation zone of the last debristorrent was the independent variable, while cross-sectional area of sediment (bothmeasured and estimated by the previous four techniques) was the dependent variable. Aregression line was fitted and the equation was used to predict cross-sectional area valuesat the start of the transport zone, and at the start of the deposition zone of the most recentdebris torrent.Figure3.5Exampleof cross-sectioninwhichtrapezoidalbedrockchannel wasassumed0.0-0,5-1.0-1.5-2.0012345Distance acrosschannel (m)0.0-0.5-1.0-1.5-2.00Figure3.6Exampleofcross-sectioninwhichbedrockoutcropswereextrapolated123456Distanceacrosschannel(m)47Comparison ofmeasurement and estimation techniquesIn order to assess the validity of the cross-sectional fill estimates, estimated areasclose to each measured cross-section were selected for comparison. Proximity is anecessary condition, as sediment accumulation is quite variable over short distances.Some of the excavated cross-sections were located near bedrock sections (cross-sectionalarea estimated as zero) and therefore were not included in the comparison. Figure 3.7illustrates that the two methods do not share a 1:1 relationship and that the estimationtechniques tend to underestimate cross-sectional area.The two data sets were statistically compared to assess the similarities of the twotechniques. The variance and mean of each were compared using an F test and t test ofsample means, respectively, assuming the data are somewhat normally distributed (Figure3.8). The mean area of the measured cross-sections is 0.87 m2 with a sample variance of0.14 m2 (n=20), and the mean area of the estimated cross-sections is 0.71 m2 with asample variance of 0.26 m2 (n =20). The F and t tests revealed that the samples couldhave been drawn from the same population, with a significance level of a equal to 0.01(Appendix B).As a result of this comparison, the estimation techniques used to determine cross-sectional area are considered to be sufficiently accurate. These estimations were used tosupplement measured cross-sections in the nine gullies with excavated sites, and wereused independently in the other twenty gullies that lack excavation data.Estimation of total volume ofsedimentThe estimation of total sediment volume in a gully was based on two assumptions:(1) the bedrock longitudinal profile is continuous and uniform between bedrock referencepoints in the channel; and (2) the cross-sectional area of sediment varies ilnearly betweenlocations where the area was measured or estimated. These conditions permitted thecalculation of sediment volume by integration of cross-sectional area between two48Figure 3.7 Comparison of cross-sectional areadetermined by measurement and by estimation2.0xxCCt1.0c. 0.5 x xCtx0.0 I0.0 0.5 1.0 1.5 2.0Area from measurement (m2)Figure 3.8 Frequency distribution of measuredand estimated cross-sections25%____________Measured20% Estimated15%10% .. ....5% . .0% T r T I I I I I0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0Cross-sectional area (m2)49adjacent cross-sections.Figure 3.9 illustrates how cross-sections were positioned at significant locationsand how the geometry of sediment deposits was approximated. Figure 3.9a is anexample of how the deposits may look in plan view and longitudinal profile view, whileFigure 3.9b is an example of how the deposits’ dimensions were integrated. Volume wasdetermined in each gully by plotting approximated cross-sectional areas against channeldistance below the initiation zone of the last torrent (Figure 3.10). The area under this‘curve’ is the interpolated volume of sediment in the gully, representing an infinitenumber of cross-sectional areas multiplied by distance. The volume between adjacentcross-sections is equal to the area of the triangles and trapezoids which are delineated byadjacent data points in Figure 3.10. The appropriate formula was used to calculatevolume between adjacent cross-sections, and the volumes in each channel segment weresummed to obtain total sediment storage.50Figure 3.9 Example of sediment loading in gully andapproximation of sediment deposit dimensionsOverview of sedimentfill in typical gullyFigure 3.9a Actual width and depth of depositX-S1 L_____________X—S2 LMii jX-S3X-S4Plan view--JLongitudinal profile viewL:1Lx-s5X-S6cts.cr52Figure 3.10 Example of plot used to calculatevolume of sediment in a channel2.0 X-S6r1.51.0 x-s50.5X-S2d /\xs30.0 X-S40 100 200 300 400Distance below torrent initiation zone (m)-Area of each polygon (delineated by dashed lines)was calculated to determine the volume of sedimentbetween adjacent cross-sections-53Chapter 4 Study Results, Analysis and Discussion4.1 OverviewAs discussed in Section 1.6, the objectives of this project involve the estimation ofdebris volume in storage in a gully (objective 1), the estimation of the date of the lastdebris torrent (objective 2), and the determination of debris recharge rates (objective 3).Furthermore, it was hypothesized that the removal of a timber stand from a hilislopewould result in a significant change in the delivery rate of LOD and in the recharge rateof debris. This will be assessed in terms of differences in debris recharge rates betweenold-growth and clear-cut gullies (objective 4).This chapter provides a brief discussion of the results in terms of each of theseobjectives, as well as an assessment and statistical analysis of the differences in LOD,sediment, and total debris recharge rates between the two land-treatment groups.Possible temporal trends in debris accumulation are discussed, and various controls ofrecharge rates are explored as possible explanations for any differences in geomorphicactivity between the two land-treatment groups. Finally, the data are used as an indicatorof local erosion rates and are compared with erosion rates from various regions aroundthe world.4.2 Volume of debris in storage4.2.1 Large organic debris volumesThe volume of single pieces of LOD and log jams were estimated using themethods described in Section 3.3.2. When considering the accuracy of these volumeestimates (Appendix C, Table C. 1), two field conditions are pertinent: (1) partial orcomplete burial of LOD by sediment and the growth of mosses were common (Figure4.1) and made measurement or estimation of the dimensions more difficult; and (2) inold-growth gullies, LOD was sometimes so abundant as to make volume estimation ofindividual pieces nearly impossible (Figure 4.2). Nevertheless, the volume of LOD54present in a channel was important to estimate, and it is included in the calculation of thetotal volume of debris for two main reasons: (1) it is assumed that the LOD present in achannel has entered a gully since the last debris torrent, and (2) it is probable that theLOD present will be entrained by the next debris torrent large enough to mobilize thedebris.The gullies sampled are all of similar physiography and size, but in order to makean accurate comparison of the LOD, sediment, and debris volume calculations betweengullies, the volume of each was normalized by the area that contributes material to eachgully. Contributing area is defined as the product of average sidewall length and channellength of the gully (column 2 in Table 2.2), plus the surface area of the initiation scar ofthe last debris torrent. The latter area is a particularly important source of debris in theFigure 4.1 Typical old-growth gully in which large portions of the LOD are buriedby debris and covered with moss. Gully G5, Gregory Creek.Torrent date: 1917.55more recently torrented gullies, as scars are typically unstable and barren. Errors in thecalculation of gully area will be a result of the fact that sidewall length was oftenestimated rather than measured.4.2.2 Sediment volumesThe volume of sediment stored in a gully was determined using the techniquesdescribed in Section 3.4.2. Measured and estimated cross-sectional areas were used todetermine the volume in the nine guffies with excavation sites, while only estimatedvalues were used in all other guffies. The total volume in each gully was normalized bycontributing surface area, as described in the previous section. The results are presentedin Appendix C, Table C.2.Figure 4.2 Typical old-growth gully with abundance of LOD. Gully G6, GregoryCreek. Torrent date: 1935.56The accuracy of the estimation techniques, compared with measurementtechniques, has been reviewed (Section 3.4.2) and should be considered in the assessmentof possible errors in the volume calculations. Another possible source of error is relatedto the integration of the cross-sectional area of fill between known or estimated points.This technique assumes a linear change in the depth and width of the fill and a continuousbedrock profile between the points. Bedrock irregularities beneath the sediment fill arenot accounted for, and the volume calculation for that section of fill would include someerror. However, it was not feasible to excavate all of the zones of debris accumulation,so it was necessary to make these assumptions.4.2.3 Total debris volumesThe calculation of the total volume of debris present in a gully involved thesummation of the volume of LOD and the volume of sediment in storage. Again, totalvolume was normalized by the contributing area of a gully. The results for each of thesampled guffies are listed in Appendix C, Table C.3.To first approximation, the comparison between normalized debris volume in thetwo land-use groups reveals that, in general, there is nearly twice as much debris in thetorrented old-growth gullies than in the torrented clear-cut gullies, with only a handful ofexceptions. Although this comparison does not consider the time elapsed since the lastdebris torrent, initial explanations for some of the cases are hypothesized as follows:(1) The average channel slope of Slog is high at 38° (compared to a mean slopeof 26° for all of the other gullies sampled), making the deposition of material lesslikely to occur than on gentler slopes. This may account for the very low amountof debris in this gully.(2) It is believed that clear-cut gullies which are located at or very near a clear-cut boundary will contain more LOD than those in open clear-cut tracts. This is57because a supply of LOD will still be available for delivery to the upper reaches ofthe gully. This may explain the relatively high amount of debris in which isadjacent to a cut boundary.(3) B2 is generally free of debris along most of the gully, except for one sectionwhere a large stump was wedged at a channel constriction. Upstream of thestump, a very large sediment wedge (approximately 250 m3 of material) hasdeveloped. Upstream of the wedge and downstream of the stump the channel ismainly bedrock, so this sediment accumulation is not characteristic of the gullysstorage capacity. This demonstrates the control a single piece of LOD can haveon sediment storage.4.3 Debris torrent datesThe timing of the occurrence of the last debris torrent in each gully was estimatedfrom the dates provided by Schwab (pers. comm. 1992), aerial photography,dendrochronology, and interviews with knowledgeable people in the area, as described inSection 3.2.2. In some gullies, there was not enough evidence to pinpoint the year thetorrent occurred, but an interval of time could be identified. The single dates and timeintervals estimated and the methods used in each gully are listed in Appendix D.Clear-cut gullies that torrented prior to the mid-1970’s are lacking. These weredifficult to find for two reasons: (1) logging of steep slopes on the east side of RennellSound only dates back to the early 1970’s, and (2) most of the clear-cut guffies had retorrented since the 1970’s. As a result, the conditions of debris accumulation in olderclear-cut gullies are not known, and comparison of debris recharge rates between the twoland-treatment groups will not be as complete as it would be if gullies of similar agegroups had been compared.584.4 Recharge ratesAlthough one of the main objectives of this study is to estimate debris rechargerates (combined LOD and sediment), the recharge of LOD and sediment were alsodetermined separately (Table 4.1). This allowed a more detailed analysis of thedifferences in storage characteristics and the recharge of material between old-growth andclear-cut gullies. In a given gully, the normalized volumes of LOD, sediment, and totaldebris were divided by the number of years since the last torrent to yield recharge rates incubic meters of material per gully area per year (m3ha’year’). For those gullies inwhich only a range of possible dates of occurrence could be determined, the averagenumber of years was used.To first approximation, it appears that LOD is delivered to the old-growth gulliesat a rate more than twice that to the clear-cut gullies, and that sediment and debris aredelivered to the clear-cut gullies at rates twice as high as those to the old-growth gullies.The standard deviations of the sediment and debris recharge rates in the clear-cut gulliesare nearly three times as high as that of the old-growth gullies, implying much morevariability in the clear-cut data. The statistical similarities and differences are exploredand discussed in the following section.59Table 4.1 Recharge rates of large organic debris, sediment, and total debris in gullies.LOD sediment debrisrecharge recharge rechargerates rates ratesgully (m3ha-’yr’) (m3ha-’yr’) (m3ha-’yr1)old growthGi 10 133 143G2 10 141 151G3 28 120 148G4 14 131 145G5 2 20 22G6 1 20 21G9 18 42 60GlO 1 40 41Gil 31 96 127G12 2 49 50R4 1 88 88R5 2 72 74SlogZ <1 31 32mean 9 76 85standarddeviation 11 45 52clear-cutBi 24 530 554B2 23 190 213B3 0 74 74B4 <1 75 75R2 0 193 193R3 1 58 59R7 2 208 210R8 3 26 30R9 3 60 64RiO 1 167 168R12 8 299 307R13 1 89 91R14 <1 42 42Slcc’ 2 138 139SB1 0 64 64SB2 <1 13 14mean 4 139 143standarddeviation 10 131 137a The most recent debris torrent in Si initiated in old growth and torrented through clear-cut; the twosegments are treated as separate gullies (Slog and Si).604.5 Analysis of study results4.5.1 Statistical techniques used to compare sample meansIn the geosciences, parametric statistical techniques are often used in data analysis,but require two main assumptions: (1) that the data are normally distributed, and (2) thatthe samples are fairly large (n> 30). In this study, neither of these assumptions are met.The data were mostly right skewed (Figure 4.3a) and transforming the datalogarithmically still resulted in a non-normal distribution (Figure 4.3b). Secondly, thesample sizes were small (old-growth gullies n= 13, clear-cut gullies m= 16). As a result,the nonparametric Mann-Whitney test was selected as a substitute for the t-test forexamining the equality of two sample means. In this test, the only requirements are thatthe two samples were drawn independently of one another and that the populationdistribution shapes are the same (Figure 4.3).The Mann-Whitney test involves three steps. First, the data in both samples (oldgrowth, OG and clear-cut, CC) are pooled and ranked in ascending order. Next, a teststatistic, T, is calculated for either of the samples (OG or CC) using the formula:T = E R(0G1)- n(n + U [4.1]2where 0G1 is the ith observation from sample OG, R(OG) is the rank of this observationin the pooled sequence, and n is the size of sample 0G. Finally, the T statistic iscompared to critical values of Tin tables. For a two-tailed test of equality, critical rangesare identified in which T must fall for the null hypothesis of equal population means to berejected at a given confidence level. In a one-tailed test, used to determine if one sampleis significantly greater than the other, a single critical T value must be exceeded for thenull hypothesis to be rejected (Siegel 1956).Both one- and two-tailed tests were used in this study. It was hypothesized withconfidence that LOD would be delivered to old-growth gullies at a higher rate than toclear-cut gullies, so the more powerful one-tailed test was used in the comparison of61Figure 4.3a Frequency distribution ofLOD recharge rates6O%50%-.Old growth, n=1340%-Clear-cut, n=16/.io:j. fl RH ,0 3 6 9 12 15 18 21 24 27 30 33 36LOD recharge rates (m3/ha/year)Figure 4.3b Frequency distribution of log-transformed LOD recharge rates40%[ Old growth,30% Clear-cut, n=14(_)2O%r/10%••.0%Log-transformed LOD recharge rates (m3/halyear)62LOD recharge rates. However, the differences in sediment and debris recharge rateswere less apparent, a priori, and thus two-tailed tests were used in these comparisons.4.5.2 Large organic debris recharge ratesLOD has accumulated at a faster rate in old-growth gullies (mean =9.2m3halyr’,standard deviation =11.0 m3ha-1yrl) compared with clear-cut gullies (mean = 4.3 mha1yr, standard deviation =7.7m3ha-1yr). When recharge rates are compared statisticallywith the Mann-Whitney test, over a common period of record (0-14 years since the lastdebris torrent), old-growth rates are indeed found to be significantly higher (a =0.05)(Table 4.2). The same result is apparent when the entire period of record is considered,but is less meaningful as there are no clear-cut gullies sampled with torrenting older than14 years. The lower input rates following clear-cutting are a consequence of the removalof the main LOD source, the standing forest. After the first torrent in a gully followinglogging, LOD supply is limited to infrequent delivery of stumps and yarding debris asroots deteriorate and gully sidewalls fail.It is difficult to make comments about the pattern of LOD accumulation in clearcut gullies beyond the 14-year period. The even-aged structure and youth of second-growth stands initially results in lower rates of LOD delivery compared with old-growthareas, but when second-growth reaches a certain age (fifty or sixty years old? - Grette1985), LOD delivery rates may approach those of old-growth gullies. The implication isthat within the 80-year rotation period for coastal forest harvesting, LOD supply ratesmay have temporarily recovered to the ‘background’ rates typical of old-growth forests.With a comparison of the old-growth LOD recharge rates in the early period aftera torrent (0-14 years, mean = 10.6m3ha’yrl) to all of the old-growth data (0-75 years,mean = 9.2m3ha-lyrl), it is suggested that gullies have an LOD capacity after which theaccumulation rate stabilizes or declines. If the accumulation rate stabilizes, this impliesan equilibrium between LOD input and output. This may occur once stored LOD breaks63Table 4.2 Comparison of large organic debris recharge rates between land-treatmentgroups.samplesample standardmean deviationn (3ha-lyr’) (m3ha-lyr’)old growth 11 10.6 11.0clear-cut 16 4.3 7.7one-tailed test! H0: OG = CCupper region: H1: OG > CCT0b (OG) = 126 a = 0.01 = 134a = 0.05 = 121a = 0.10 = 114Reject H0, Accept H1, a = 0.05down sufficiently to allow its removal by gully runoff. Alternatively, a decline in LODinput rates may exist and result in a slow net loss of LOD. This is conceivable as thenumber of standing trees left in unstable positions by the undercutting action of theprevious debris torrent is reduced as gully sidewalls revegetate. This apparent reductionof LOD accumulation after several decades may also be a result of sampling difficulties.In the older gullies, accurate measurement of LOD storage was not possible because ofthe partial or total burial of LOD by sediment and by the subsequent growth of mosses.The data are sparse and it is difficult to define the long-term patterns of recharge.It is clear, however, that in the early years after a debris torrent, LOD accumulates morequickly in old-growth gullies than in clear-cut gullies. Until second-growth stands reacha critical age, it is reasonable to expect lower rates of LOD accumulation in clear-cutgullies.644.5.3 Sediment recharge ratesAlthough old-growth gullies have recharged with more LOD than the clear-cutgullies, this is not matched by higher sediment recharge rates, as originally hypothesized(Table 4.3). In fact, the mean sediment recharge rate in the clear-cut gullies is almosttwice that in the old-growth gullies for the period of common record, although the highvariability in each sample precludes a statistically significant distinction (a=0.Ol, n 11,m=16).Table 4.3 Comparison of sediment recharge rates between land-treatment groups.samplesample standardmean deviationn (3ha-’yr’) (m3ha-’yr-l)old growth 11 85.7 44.7clear-cut 16 139.1 130.4Mann-Whitney test results:two-tailed test: H0: OG = CCH1: OG+CCT0b (OG) 70 range: a = 0.02 = 41-135a 0.05 = 47-129a 0.10 = 54-122Accept H0, a = 0.02It is useful to consider the possible causes of the higher rates in the clear-cutgullies. Sediment recharge rates depend on the delivery of both LOD and sediment to achannel. Since clear-cut gullies receive lesser amounts of LOD, the higher meansediment recharge rate in the clear-cut areas implies more rapid sediment delivery. Thisis attributed to the greater instability of sidewalls and bank tops where vegetation hasbeen removed, and to generally increased sediment production from disturbed hilislopes65(Schwab 1983, Rood 1984, Sidle et a!. 1985).Alternatively, higher sediment recharge rates in clear-cut areas may reflect theages of the sampled gullies, in particular the lesser mean age of the clear-cut gullies. Themean age of the old-growth gullies is twenty-one years since the last torrent, with a modeof fourteen years, compared with a mean of seven years and mode of eight years in theclear-cut gullies (Appendix D). Recently disturbed gullies have a higher sedimentaccumulation rate than any of the other gullies, as shown in Figure 4.4. This graphshows that sediment is accumulating at a high rate in the younger, clear-cut gullies andthat the rate progressively declines as time since an event increases (for predominantlyold-growth gullies). Unfortunately, there are not any old-growth gullies younger than sixyears nor any clear-cut gullies older than fourteen years since the last debris torrent.Consequently, the comparison of sediment and debris recharge rates is between differingage groups.If the difference in age groups is ignored, it is possible to fit a common powerfunction (R=bt -a, where R is sediment recharge rate and t is time) through most of thedata points. This implies that there is no difference in sediment delivery between the twoland-treatment groups, and that any differences in debris recharge rates will be a result ofdiffering LOD delivery rates. However, the r2 of 0.47 for the combined data is very low(standard error of the R estimates, SE{R}, = 0.28, in log units), and it is difficult toassign much significance to the visual trend. In addition, evidence from gullies in bothland-treatment groups recently scoured in the study area, and other areas in the coastalbelt, suggest that sediment delivery from clear-cut gully sidewalls will be higher thanfrom old-growth gully sidewalls and will remain higher for many years after a debristorrent. This is a result of the lack of a shade-producing canopy for many decades inclear-cut gullies. This canopy permits the growth of mosses and other bryophytes, whichtend to stabilize sidewalls and allow the germination of other vegetation.Consideration of the pattern of variation in each land-use group separatelyFigure4.4Comparisonofaveragesedimentrechargeratesandtimesincelastdebristorrent1000I) 4 100 C-) 10100Timesincelastdebristorrent(years)110C’67supports the argument that the entire data set should not be described by a common powerfunction. There is virtually no relationship between sediment recharge rates and timeelapsed since a debris torrent in the old-growth gullies (r2=0.34, SE{R} =0.26, in logunits), whereas the relationship is somewhat stronger for the clear-cut data (r2=0.52,SE{R} =0.29, in log units). In addition, the slopes of the power-function models aregreatly different (old growth= -0.59, clear-cut= -0.80). Any evidence of a commontrend is probably a function of the lack of both types of gullies on each end of the timescale.The limitations of the data clearly pose problems for interpretation. Obviously amore ideal comparison would be between the old-growth gullies sampled and clear-cutgullies that are of a similar age. This was the original goal of the project, but it quicklybecame apparent that most of the clear-cut gullies are much younger than old-growthgullies, either because of re-torrenting or because older gullies were not accessible.The discussion thus far has considered the two components of total debris in agully: large organic debris and sediment. Based on the observed trends, the followingarguments have been made:(1) The short-term recharge of LOD is more rapid in old-growth gullies than inclear-cut gullies. The implication of this is that old-growth gullies will generallyhave more LOD and therefore greater sediment storage capacity.(2) Although the difficulty in measuring LOD in older gullies may have led to theunderestimation of the volume of LOD present, it seems that with time, LODaccumulation rates stabilize and possibly decline.(3) Although the data on sediment delivery rates indicate a possible commonfunctional line for both data sets, the gaps in the data are significant enough toquestion the trend. It is therefore suggested that the sediment recharge rates are68higher in clear-cut gullies than in old-growth ones in the early years after a debristorrent because of the differences in sidewall stability. Given the small amountsof LOD in the clear-cut gullies, this is the most probable explanation for the lackof a statistically significant difference in sediment accumulation between clear-cutand old-growth gullies.(4) Clear-cut gullies lacking LOD will quickly achieve their sediment storagecapacity while old-growth gullies will continue to recharge with sediment as longas LOD continues to add to their storage capacity. As a result, moderately agedand older old-growth gullies should contain more LOD and sediment than clearcut gullies of the same age.It is now important to consider the implications and the cumulative effects of these trendson the total debris recharge into gullies.4.5.4 Debris recharge ratesFor the period of common record (0-14 years) the statistics show that there is nosignificant difference between old-growth and clear-cut gully debris recharge rates(a =0.01, n =11, m =16) (Table 4.4). However, a graph of the mean values of thesedata (Figure 4.5) suggests that while debris accumulation in clear-cut gullies is increasingslowly over this time period, the accumulation rate in old-growth gullies is occurring at afaster rate. This is consistent with the trends suggested in sections 4.5.2 and 4.5.3,particularly the fact that LOD recharge to clear-cut gullies is relatively low, and as aresult, the sediment storage capacity related to LOD in the clear-cut gullies is beingreached sooner than in the old-growth gullies. It is possible to describe the old-growthdata with a weak exponential function (Figure 4.6; r2=0.47, SE{R} =0.24, in log units),but the model for the clear-cut data is less obvious (r2=0.07, SE{R} =0.29, in log units).69Table 4.4 Comparison of debris recharge rates between land-treatment groups.samplesample standardmean deviationn (3ha-’yr’) (m3halyr’)old growth 11 96.2 44.7clear-cut 16 143.4 137.8Mann-Whitney test results:two-tailed test: H0: OG = CCH1: OG+CCTobs (OG) = 74 range: a = 0.02 = 41-135a = 0.05 = 47-129a = 0.10 = 54-122Accept H0, a = 0.02The data are limited for longer time periods, but on the basis of the available old-growth data (Figure 4.7), it appears that these gullies reach a somewhat stable level ofdebris accumulation, as the balance between LOD input and output is achieved and thestorage sites are filled to their capacity. In contrast, the clear-cut pattern can only behypothesized, but it would include a delayed increase in debris accumulation related tothe regeneration of the adjacent forest. Neither data set can be described by a strongpower function (old growth: r2=0. 19, SE{R} =0.28, in log units; clear-cut: r2=0.06,SE{R} =0.29, in log units) or exponential function (old growth: r2=0.08, SE{R} =0.29,in log units; clear-cut: r2=0.07, SE{R} =0.29, in log units).The implications of these hypothesized long-term patterns with regard to futuretorrent activities in the two types of gullies are that until LOD is introduced in quantity,the clear-cut gullies should be characterized by a chronic evacuation of the sediment fromthe channel. This could result in debris torrents of a relatively small magnitude but arelatively high frequency. Once LOD is present, long-term accumulation should begin.Figure4.5Totaldebrisaccumulationingullieslessthan15yearsold1800><Oldgrowth•Clear-cut1500n=8Meanvaluesinclude95%confidencelimits1200C900-m=5•600-•><••C300 0-IIIIIIIII0123456789101112131415Timesincelastdebristorrent(years)Figure4.6Totaldebrisaccumulationingullieslessthan15yearsold,withfittedexponential functions10000-><Oldgrowth---Oldgrowthregression•Clear-cut——.Clear-cutregressionMeanvaluesinclude95%confidencelimitsC-.p-I’ E1000- -.R=938eO02L100II0123456789101112131415Timesincelastdebristorrent(years)C EFigure4.7Total debrisaccumulationingullieslessthan80yearsold1000100 10110100Timesincelastdebristorrent (years)73Old-growth gullies, on the other hand, are less likely to exhibit a chronic sedimentoutput, but the potential magnitude of future events is likely to be significantly higherthan in clear-cut gullies, but probably at a lower frequency. These conditions may haveimplications for managers and policy-makers, particularly those who must assess theimpact of logging slash on gullies.4.5.5 Other factors consideredAt the time of data collection, it was assumed that the sampled gullies differedonly in their land use. However, variability from other factors was inevitable and it ispossible that the results may be further explained by these conditions. To investigate thelikelihood of this, the influence of the position of clear-cut boundaries on LOD rechargerates is considered, as are the influences of bedrock and surficial geology on sedimentrecharge rates.Clear-cut boundariesThe most recent debris torrents in the sampled clear-cut gullies were initiated inthree locations: in an old-growth stand, at a cut boundary, or within a clear-cut area,some distance downslope of the cut boundary. It is hypothesized that those gullies inwhich a torrent was initiated at or near a cut boundary should have higher LOD rechargerates (and hence, higher debris recharge rates) because of the proximity to an old-growthstand in the upper reaches of the gully. These trees are more susceptible to windthrow(Chatwin et al. 1991), as they are exposed to the study area’s strong winds.The results in Table 4.5 show that the presence of a cut boundary has nostatistically significant effect on average recharge of LOD to clear-cut gullies. However,a visual comparison between two clear-cut gullies of similar age demonstrates that theeffect can be locally significant. Figure 4.8 is a photograph of the abundance of LOD ina gully adjacent to a cut boundary, compared to one located elsewhere in the same74Table 4.5 Comparison of LOD recharge rates in clear-cut gullies in relation to cut-boundary proximity.1.14.8samplestandarddeviation(m3ha’yr-l)7.026.1Mann-Whitney test results:two-tailed test: H0: cut boundary = open slopeH1: cut boundary + open sloperange: = 0.005 = 1-27= 0.025 = 4-24= 0.05 = 5-23Accept H0, cr = 0.005nsamplemean(3ha-’yr1)cut-boundaryopen clear-cut214Tobs (cut b.) = 17Figure 4.8 Abundance of LOD in clear-cut gully located in close proximity to acut boundary. Gully R6, southeast of Gully R7, Riley Creek (not includedin analysis). Torrent date: 1978.75cutblock (Figure 4.9). Based on the visible difference, it is reasonable to expect that alarger sample of guffies adjacent to cut boundaries may have produced different statisticalresults.Bedrock geologyLithology and structure can potentially affect sediment accumulation rates.Although the geological make-up of the gullies in the two land-treatment groups isdifferent (e.g. plutons are present only in old-growth gullies; Kunga rocks occur only inclear-cut gullies; both groups are dominated by Yakoun rocks), at the time of sampling itwas assumed that these differences would not affect debris recharge rates. Thisassumption was based on the fact that the formations are all highly fractured, well-jointedFigure 4.9 Clear-cut gully of same age as gully in Figure 4.8, but located in openclear-cut area and lacking LOD. Gully R9, Riley Creek. Torrent date:1978.76volcanic and sedimentary units which would weather at similar rates. In Table 4.6, astatistical evaluation of the effect of bedrock geology is presented. The results show thatwithin each land-treatment group, the two major geologic types do not exhibit significantdifferences (a =0.005) in sediment recharge rates over the period of common record (0-14 years) (Table 4.6 (a) and (b)). To maximize the rigor of the statistical tests, it wasdesirable to maximize the degrees of freedom of a sample. This was achieved bycombining the data of both rock types in each land-use group. This permits a morepowerful comparison of the effect of a change in land use, given the limited data. If thedata from the two geologic types in each treatment group are not combined, and only thecommon rock type (Yakoun) is considered, a significant difference is detected in thesediment recharge rates (a =0.05; Table 4.6 (c)). However, this test is not as rigorousas one using the data sets of the combined rock types, as the sample sizes (and degrees offreedom) are smaller. From these results it was concluded that bedrock geology does notsignificantly influence sediment recharge rates in the sampled gullies.Surficial geologyThe basal till that is present in the channels and sidewalls of gullies in the RennellSound area is not very consolidated, being derived from weak shales and othersedimentary units. It is possible that the till and highly fractured bedrock weather atsimilar rates, but if this is not the case, then it is necessary to exclude the handful ofgullies that contain significant amounts of till from all of the analyses. Sediment rechargerates were compared within a land-treatment group between gullies with till and gullieswith bedrock. Statistical results show that for the period of common record, there is nosignificant difference in recharge rates in clear-cut gullies between the two material types(a0.005, n=5, m=ll) (Table 4.7). Therefore, the presence of basal till could not beused to explain the high sediment recharge rates in the clear-cut gullies. Only two of thethirteen old-growth gullies sampled have till present.77Table 4.6 Comparison of sediment recharge rates between the dominant geologicformations.samplesample standardgeologic mean deviationformation n (3ha-’yr’) (m3halyrl)old growthYakoun 8 94.7 42.4Kunga 3 61.6 30.0clear-cutYakoun 12 158.2 141.4Plutons 4 82.1 77.5Mann-Whitney test results:(a) Yakoun versus Kunga (within OG):two-tailed test: H0: Yakoun = KungaH1: Yakoun + KungaTob (Kunga) = 13 Tt range: a = 0.005 = 1-23a = 0.025 = 4-20a = 0.05 6-18Accept H0, a 0.005(b) Yakoun versus Plutons (within CC):two-tailed test: H0: Yakoun = PlutonsH1: Yakoun + PlutonsT0b (Plutons) = 24 T0nt. range a = 0.005 = 6-42a = 0.025 10-38a 0.05 = 13-35Accept H0, a = 0.005(c) Yakoun (OG) versus Yakoun (CC):one-tailed test! H0: Yakoun(OG) = Yakoun(CC)upper region: H1: Yalcoun(OG) > Yakoun(CC)T0b (Y - OG) = 72 a = 0.01 = 78a = 0.05 = 69a = 0.10 = 65Reject H0, Accept H1, a = 0.0578Table 4.7 Comparison of sediment recharge rates in relation to channel and sidewallmaterial in clear-cut gullies.samplesample standardmean deviationn (3ha-’yr’) (m3halyrl)basal till (CC) 5 119.0 118.3bedrock (CC) 11 148.3 141.4Mann-Whitney test results:two-tailed test: H0: till = bedrockH1: till + bedrockTobs (till) = 38 range: a = 0.005 = 8-47a = 0.025 13-42a = 0.05 = 16-39Accept H0, a = 0. 0354.6 Consideration of erosion rates at a global scaleThe level of geomorphic activity in an area is dependent on many factors,including relief, slope, climate, and bedrock geology. The Rennell Sound area ischaracterized by a high degree of geomorphic activity compared to many other areasaround the world. This can be demonstrated by a comparison of the study results withpublished data on erosion rates derived from the rates of individual geomorphic processes(Table 4.8). All efforts were made to compare basins of a similar size, but some of thedata presented do represent larger basins.An average sediment recharge rate for an area represents the rate at whichsediment is transferred from hillslopes to first-order channels, and this can be used as aminimum erosion rate of the hillslopes. A natural erosion rate for the study area wasestimated as the quotient of the average sediment volume in the old-growth channelsdivided by the average area of the contributing basins, per year.Table4.8Comparisonoferosionratesderivedfromratesofindividualprocesses.basingeomorphicerosionlocationarea(km2)climatebedrockreliefzprocessrate(B)bsourcedebrisslump,temperatevolcanic,debrisslide,RennellSound0.2maritimesedimentarysteepravelling3000-BritishtemperateColumbia2.4-21.6alpinegraniticnormalvarious7-21Slaymaker1977subtropicaldebrisTanakaandJapann/ahumidn/asteepavalanche1000Mori1976temperatesandstone,NewZealand100.0maritimesiltstonesteeplandslide3800Selby1976NortherntemperateIrelandn/amaritimebasaltssteeprockfall277-1617Douglas1980YairandSinain/aaridsandstonenormalrockfall100-400Gerson1974periglacialamphibolite,debrisslide,Spitsbergen5montanesandstonenormaldebrisflow4-54Rapp1985continentalgranodiorite,rockfall,UnitedStates2.1alpinegneisssteepdebrisflow6Caine1986asteep=mountainous,steeply-dissectedregions,individualslopes>250;normal=allothersbB=Bubnoffunits,where1B1mm/1000years(fromSaundersandYoung1983)80A convenient measure of erosion rates is the Bubnoff unit, B, where 1 B= 1mm/1000 years, or the removal of 1 m3km-2 of material per year. The use of this unitallows the comparison of data from various sources that are presented in different units.It is apparent that the west coast of the Queen Charlotte Islands is one of the morenaturally active regions of the world. The estimated erosion rate of 3000 B is comparableto the rate for the steep slopes of the west coast of New Zealand, also of a temperatemaritime climate with weak sedimentary rocks (Selby 1976). However, if the rate for theRennell Sound area is further extrapolated to 30 meters of erosion per 10,000 years, it isimplied that the slopes should be free of till and the gullies should be considerablydeeper. Therefore, the intensive geomorphic activity in the twentieth century may not becharacteristic of the area on a scale of thousands of years. Nonetheless, the area is highlyactive in the present, and care should be taken in the consideration of any future land-usechanges, as these will most probably accelerate the already high erosion rates.4.7 SummaryIt is clear that the nature of the data available, and the age distribution of thegullies in the Rennell Sound area, have resulted in conflicting time periods for each land-treatment group. These conditions may have confounded the sediment and debrisrecharge rates and made it impractical to complete a conclusive analysis. Although, theMann-Whitney tests and regression analyses do not, in general, support the researchhypotheses, the graphical representations of the data reveal trends that may bestrengthened with larger data sets.81Chapter 5 Conclusions and Recommendations5.1 OverviewThis project was undertaken to investigate the rate at which large organic debris,sediment and total debris accumulate in old-growth and clear-cut gullies scoured by debristorrents. The data collected on LOD recharge rates support the belief that these rates willbe higher in old-growth gullies than in clear-cut gullies because of the presence of theold-growth stands. However, it is difficult to form conclusions about the influence ofland use on sediment and total debris recharge rates because of the small samples and theconfounding effect of the temporal differences in the samples. Nevertheless, this projecthas important applications and implications for geomorphologists and forest managers thatare interested in the processes occurring in coastal gullies. In addition, possible trendshave been identified which can be further investigated in future projects.5.2 Applications in geomorphologyThe results of this study have several important applications for geomorphologists.First, the project involved the estimation of total volume of stored debris in nearly thirtygullies in the Rennell Sound area. These volume estimates can be used in theconstruction of sediment budgets for the study area drainage basins. In addition, theestimates can be extrapolated to gullies in a similar environmental setting that are of acomparable size and nature to those sampled. With these constraints in mind, theextrapolated estimates may be used in the construction of sediment budgets outside of thestudy area.Second, the debris volume estimates can be used to predict a minimum magnitudeof future debris torrents based on current channel conditions, as event magnitude will be afunction of the total available and entrainable debris currently in the channel and anyerodible sidewall material (VanDine 1985). Magnitude is commonly predicted fromestimates in depositional zones (Innes 1983, Rood 1984). However, in coastal gullies,82debris torrents frequently deposit in higher-order streams and part of the torrent isimmediately washed downstream, while part is subsequently eroded by fluvial processes.As a result, magnitude estimates based on the remaining deposit will be significantunderestimates. Therefore, estimates based on current channel conditions may be moreaccurate (Appendix E).Finally, debris recharge rates can be used to improve the understanding offrequency-magnitude characteristics in the sampled gullies with the existing knowledge ofeither the frequency or magnitude of past events. For example, if the number of yearssince a debris torrent in a gully is known, then the debris recharge rate for that gully canbe used to estimate a minimum magnitude of an event that could occur in a given year.If, on the other hand, only the magnitude of a past event can be approximated, then thedebris recharge rate can be used to predict the number of years needed to recharge thegully with the same volume of debris.5.3 Implications for forest managersThe objectives of the managers of British Columbia’s forest industry include: (1)to minimize the loss of the land-base available for silvicultural activities, resulting fromhillslope failures and other geomorphic activities; and (2) to minimize the impact of theiractivities on other natural resources, such as fish-bearing streams. The Rennell Soundarea is known to have a naturally high level of geomorphic activity (Section 4.6), and thismust be considered in the management of the area’s natural resources. As a result, anyknowledge gained on the occurrence of debris torrents, which will affect both of theforest manager’s objectives, will be advantageous to these policy-makers.It is apparent that clear-cutting reduces the supply of LOD, the primary site forsediment storage in coastal gullies. It is possible that this reduction will last for as longas second-growth stands on the coast are harvested at an eighty-year rotation period.Therefore, it is important to consider the effect that removal of this LOD supply will83have on the long-term storage of sediment and on the magnitude and frequency of debristorrents. It is probable that in the absence of LOD, gullies in clear-cut areas will becharacterized by a chronic evacuation of sediment, related to stream discharge. Thisimplies that a clear-cut gully will frequently deliver sediment to higher-order streams, butthe events will be of a relatively low magnitude. This is in contrast to the low frequency-high magnitude events that are characteristic of undisturbed systems, in which largesediment wedges can develop upstream of log jams. Obviously a high magnitude,catastrophic failure will damage a pristine environment downstream, but these events arenaturally occurring, and the damage may be only temporary. In addition, the largerdebris torrents will include an abundance of LOD, which will then create log jams in thehigher-order channels. The log jams, in turn, create large, deep pools that serve asimportant salmonid rearing pools (Bisson et al. 1987). A chronic delivery of sediment,on the other hand, is less destructive initially, but the higher frequency may be moredifficult for the stream to manage, and the lesser LOD volumes may eventually forcesignificant form changes in the higher-order channels.It is the task of the forest managers to base their decision-making strategies on anassessment of the risks involved in altering the natural gully system, particularly in termsof the LOD supply. Formal steps are now being taken by all members of the forestindustry in British Columbia with the application of the new Forest Practices Code, toensure that potential hazards on the hillslopes and in gullies are assessed prior to thecommencement of logging activities (Ministry of Forests and Ministry of Environment,Lands, and Parks 1993).5.4 Recommendations for future projectsFurther research is required to assess the difference in debris recharge ratesbetween the two land-treatment groups, as most of the results presented here areinconclusive. Many difficulties were faced in both the collection of the data and its84analysis, and therefore, future projects should include three principal considerations.First, a field site should be selected which is characteristic of a large region, as this willensure that the results can be extrapolated to a regional scale. Because of the intensegeomorphic activity characteristic of the west coast of the Queen Charlotte Islands, theextrapolation of the results of this study is limited primarily to places on the west coast ofVancouver Island in similar environmental settings.Second, the selection of the field site should be based on the existence of oldergullies (>50 years since a debris torrent) from both old-growth and clear-cut areas. Thiswill ensure that both data sets are representative of the same time frame, which isessential to a strong comparison between the two groups.Finally, the estimation of the volume of debris in storage is a difficult part of thedata collection and is primarily based on some knowledge of the debris fill depth.Estimates from the integration of cross-sectional areas seem reasonable, but shouldinclude as many excavated cross-sections as possible. However, in conditions wheredebris excavation is not a feasible option, other techniques used to determine the distancebelow the surface to bedrock (e.g. seismic refraction, ground penetrating radar) should bethoroughly investigated prior to the field season.As would be expected, this project revealed more questions related to sedimentand organic debris recharge rates than could feasibly be answered. Therefore, futureprojects of a similar nature should investigate possible trends identified in this study andshould try to answer the following questions:1) Does a second-growth stand, several decades old, deliver LOD to a gully at arate comparable to that of an old-growth stand?2) Are sediment recharge rates higher in young (<10 years) clear-cut gullies thanin young old-growth gullies? If so, how many years are required for the clear-cutsidewalls to stabilize and deliver sediment at a rate comparable to that of oldgrowth sidewalls?853) Does the accumulation rate of debris stabilize after several decades, once agully in either land-treatment group reaches its storage capacity (as suggested inFigure 4.7)?5.5 Final remarksDebris torrents occur naturally, so it is only possible to minimize rather thaneliminate the impact of these events on sensitive downstream environments. This studyidentified possible trends in debris accumulation in old-growth and clear-cut environmentsand suggested directions for related future projects.86ReferencesAlley, N.F. and B. Thomson. 1978. Aspects of environmental geology, parts of GrahamIsland, Queen Charlotte Islands. British Columbia Ministry of Environment,Resource Analysis Branch, Bulletin Number 2, 65p.Bisson, P.A., Bilby, R.E., Bryant, M.D., Dolloff, C.A., Grette, G.B., House, R.A.,Murphy, M.L., Koski, K.V., and J.R. Sedell. 1987. Large woody debris inforested streams in the Pacific Northwest: past, present, and future. In Salo,E.O. and T.W, Cundy (eds.), Streamside Management: Forestry and FisheryInteractions. Proceedings of a Symposium. University of Washington Institute ofForest Resources, Conthbution, 57, 143-190.Bovis, M., Church, M., Hungr, 0., and R. Kellerhals. 1988. Engineering problems ofsteep mountain streams. Unpublished proceedings from workshop, University ofBritish Columbia.Caine, T.N. 1986. Sediment movement and storage on alpine slopes in the ColoradoRocky Mountains. In Abrahams, A.D. (ed), Hilislope Processes. BinghamtonSymposia in Geomorphology: International Series, no. 16, 115-137.Chatwin, S.C., Howes, D.E., Schwab, J.W., and D.N. Swanston. 1991. A Guide forManagement of Landslide-Prone Terrain in the Pacfic Northwest. BritishColumbia Ministry of Forests, Land Management Handbook No. 18, 212p.Chatwin, S.C. and R.B. Smith. 1992. Reducing soil erosion associated with forestryoperations through integrated research: an example from coastal BritishColumbia, Canada. In Erosion, Debris Flows and Environment in MountainRegions. Proceedings of the Chengdu Symposium, July 1992, IAHS Publicationno. 209, 1992, 377-385.Douglas, G.R. 1980. Magnitude frequency study of rockfall in Co. Antrim, NorthernIreland. Earth Surface Processes, 5, 123-129.Froehlich, H.A. 1973. Natural and man-caused slash in headwater streams. In LoggersHandbook, 33. Portland, OR: Pacific Logging Congress, 8p.Gimbarzevsky, P. 1988. Mass wasting on the Queen Charlotte Islands: a regionalinventory. British Columbia Ministry of Forests, Land Management Report No.29, 96p.Grette, G.B. 1985. The role of large organic debris in juvenile salmonid rearing habitatin small streams. M.Sc. thesis, University of Washington, Seattle, lO5p.87Hesthammer, J., Indrelid, J., Lewis, P.D. and J.W. Haggart. 1991. Geology ofSouthern Graham Island, Queen Charlotte Islands, British Columbia. GeologicalSurvey of Canada, Open File 2319.Hogan, D.L. 1987. The influence of large organic debris on channel recovery in theQueen Charlotte Islands, British Columbia, Canada. In Erosion and Sedimentationin the PacWc Rim. Proceedings of the Corvallis Symposium, August, 1987,IAHS Publication no. 165, 343-353.Hogan, D.L. 1989. Channel response to mass wasting in the Queen Charlotte Islands,British Columbia, Canada: A preliminary evaluation of temporal and spatialchanges in stream morphology. Presented at Watershed 89 Conference, Juneau,Alaska, 1989, USDA Forest Service, Alaska Region.Hogan, D.L. and J.W. Schwab. 1991. Stream channel response to landslides in theQueen Charlotte Islands, British Columbia: changes affecting pink and chumsalmon habitat. In Proceedings of 15th Northeast Pacfic Pink and Chum SalmonWorkshop. Parksville, British Columbia, 222-236.Hungr, 0., Morgan, G.C., and R. Kellerhals. 1984. Quantitative analysis of debristorrent hazards for design of remedial measures. Canadian Geotechnical Journal,21, 663-677.Innes, J.L. 1983. Lichenometric dating of debris-flow deposits in the ScottishHighlands. Earth Surface Processes and Landforms, 8, 579-588.Jordan, P. 1987. impacts of mass movement events on rivers in the southern CoastMountains, British Columbia: summary report. Prepared for Water ResourcesBranch, Environment Canada, 6lp.Keller, E.A. and F.J. Swanson. 1979. Effects of large organic material on channel formand fluvial processes. Earth Surface Processes and Landforms, 4, 361-380.Lewis, T. 1985. Assessment of environmentally sensitive areas of part of the QueenCharlotte Timber Supply Area. Prepared for British Columbia Ministry ofForests, Vancouver, British Columbia.Lienkaemper, G.W. and F.J. Swanson. 1987. Dynamics of large woody debris instreams in old-growth Douglas-fir forests. Canadian Journal of Forest Research,17, 150-156.Marston, R.A. 1982. The geomorphic significance of log steps in forest streams.Annals of the Association ofAmerican Geographers, 72, 99-108.88Maynard, D.E. Unpublished report. Gully stability classtfication system for coastalBritish Columbia.Megahan, W.F. 1982. Channel sediment storage behind obstructions in foresteddrainage basins draining the granitic bedrock of the Idaho Batholith. In SedimentBudgets and Routing in Forested Drainage Basins. USDA Forest Service GeneralTechnical Report, PNW-GTR- 141, Pacific Northwest Forest and RangeExperiment Station, 114-121.Megahan, W.F. and R.A. Nowlin. 1976. Sediment storage in channels draining smallforested watersheds in the mountains of central Idaho. In Proceedings of the 3rdFederal Inter-Agency Sedimentation Conference. Denver, CO, 4.115-4.126.Millard, T. H. 1993. Sediment in forested and logged gullies, coastal British Columbia.M.Sc. thesis, University of British Columbia, Vancouver, 2O2p.Ministry of Forests and Ministry of Environment, Lands, and Parks. 1993. ProposedForest Practices Rulesfor British Columbia. Victoria: Queen’s Printer for BritishColumbia.Nakamura, F. and F.J. Swanson. 1993. Effects of coarse woody debris on morphologyand sediment storage of a mountain stream system in western Oregon. EarthSuiface Processes and Landforms, 18, 43-61.Pearce, A.J. and A. Watson. 1983. Medium-term effects of two landsliding episodes onchannel storage of sediment. Earth Surface Processes and Landforms, 8, 29-39.Poulin, V.A. 1984. A research approach to solving fish/forestry interactions in relationto mass wasting on the Queen Charlotte Islands. British Columbia Ministry ofForests, Land Management Report No. 27, l6p.Rapp, A. 1985. Extreme rainfall and rapid snowmelt as causes of mass movements inhigh latitude mountains. In Church, M. and 0. Slaymaker (eds.), Field andTheory: Lectures in Geocryology. Vancouver: University of British ColumbiaPress, 36-56.Rice, S.P. 1990. The spatial variation of bed material texture in coupled basins on theQueen Charlotte Islands. M.Sc. thesis, University of British Columbia,Vancouver, l2’7p.Roberts, R.G. and M. Church. 1986. The sediment budget in severely disturbedwatersheds, Queen Charlotte Ranges, British Columbia. Canadian Journal ofForest Research, 16, 1092-1106.89Rood, K.M. 1984. An aerial photograph inventory of the frequency and yield of masswasting on the Queen Charlotte Islands, British Columbia. British ColumbiaMinistry of Forests, Land Management Report No. 34, 55p.Rood, KM. 1990. Site characteristics and landsliding in forested and clearcut terrain,Queen Charlotte Islands, British Columbia. British Columbia Ministry of Forests,Land Management Report No. 64, 46p.Saunders, I. and A. Young. 1983. Rates of surface processes on slopes, slope retreatand denudation. Earth Surface Processes and Landforms, 8, 473-501.Schwab, J.W. 1983. Mass Wasting: October-November 1978 storm, Rennell Sound,Queen Charlotte Islands, British Columbia. British Columbia Ministry of Forests,Research Note No. 91, 23p.Schwab, J.W. 1988. Mass wasting impacts to forest land: forest managementimplications, Queen Charlotte Timber Supply Area. In Lousier, J.D. and G.W.Still (eds.), Degradation offorested land: forest soils at risk. Proceedings of10th British Columbia Soil Science Workshop, 1986; British Columbia Ministryof Forests, Land Management Report No. 56, 104-115.Sedell, J.R, Bisson, P.A., Swanson, F.J. and S.V. Gregory. 1988. What we knowabout large trees that fall into streams and rivers. In From the Forest to the Sea:a Story of Fallen Trees. USDA Forest Service General Technical Report, PNWGTR-229, Pacific Northwest Forest and Range Experiment Station, 47-80.Selby, M.J. 1976. Slope erosion due to extreme rainfall: a case study from NewZealand. Geografiska Annaler, Series A, 58A, 13 1-138.Selby, M.J. 1982. Hilislope Materials and Processes. Oxford: Oxford UniversityPress, 264p.Sidle, R.C., Pearce, A.J. and C.L. O’Loughlin. 1985. Hillslope Stability and LandUse. Water Resources Monograph Series Number 11, American GeophysicalUnion, 140p.Siegel, S. 1956. Nonparametric Statistics for the Behavioral Sciences. New York:McGraw-Hill Book Co., 116-127.Slaymaker, 0. 1977. Estimation of sediment yield in temperate alpine environments.Erosion and solid matter transport in inland waters- Symposium. Proceedings ofthe Paris Symposium: IAHS-AISH Publication no. 122, 109-117.90Sutherland Brown, A. 1968. Geology of the Queen Charlotte islands. British ColumbiaDepartment of Mines and Petroleum Resources, Bulletin No. 54, 226p.Swanson, F.J. and G.W. Lienkaemper. 1978. Physical consequences of large organicdebris in PacWc Northwest streams. USDA Forest Service General TechnicalReport, PNW-GTR-69, Pacific Northwest Forest and Range Experiment Station,12p.Swanson, F.J., Lienkaemper, G.W. and J.R. Sedell. 1976. History, physical effects,and management implications of large organic debris in western Oregon streams.USDA Forest Service General Technical Report, PNW-GTR-56, PacificNorthwest Forest and Range Experiment Station, l5p.Swanston, D.N. and F.J. Swanson. 1976. Timber harvesting, mass erosion, andsteepland forest geomorphology in the Pacific Northwest. In D.R. Coates (ed)Geomorphology and Engineering. Stroudsburgh, PA: Dowden, Hutchinson andRoss, Inc., 199-221.Tanaka, M. and M. Mon. 1976. Photogrammetric estimate of erosion rate in theTanzawa mountains, central Japan. (in Japanese) Geographical Review of Japan,49(4), 236-243.Toews, D.A.A. and M.K. Moore. 1982. The effects of streamside logging on largeorganic debris in Carnation Creek. British Columbia Ministry of Forests, LandManagement Report No. 11, 29p.Townshend, R.B. 1979. Geotechnic Report - Rennell Sound, Queen Charlotte Islands,Unpublished report prepared for Queen Charlotte Timber Ltd., Vancouver, byPeril Bay Geotechnic Services, 1 lOp.Tripp, D.B. and V.A. Poulin. 1986. The effects of logging and mass wasting onsalmonid spawning habitat in streams on the Queen Charlotte islands. BritishColumbia Ministry of Forests, Land Management Report No. 50, 38p.Valentine, K.W.G., Sprout, P.N., Baker, T.E., and L.M. Lavkulich. 1978. The soillandscapes of British Columbia. British Columbia Ministry of Environment,Resource Analysis Branch, Victoria, l97p.VanDine, D.F. 1985. Debris flows and debris torrents in the Southern CanadianCordillera. Canadian Geotechnical Journal, 22, 44-68.Wilford, D.J. 1984. The sediment-storage function of large organic debris at the base of91unstable slopes. In Meehan, W.R., Merrell, T.R., Jr. and T.A. Hanley (eds.),Fish and Wildlife Relationships in Old-Growth Forests. Proceedings ofsymposium in Juneau, Alaska, 1982. American Institute of Fishery ResearchBiologists, 115-119.Wilford, D.J. and J.W. Schwab. 1982. Soil mass movements in the Rennell Soundarea, Queen Charlotte Islands, British Columbia. Proceedings of CanadianHydrology Symposium. Fredericton, N. B., 521-541.Williams, G.D.V. 1968. Climate of the Queen Charlotte Islands. In Calder, J.A. andR.L. Taylor (eds.), Flora of the Queen Charlotte Islands. Canada Department ofAgriculture, Plant Research Institute, Ottawa, 16-49.Yair, A. and R. Gerson. 1974. Mode and rate of escarpment retreat in an extremelyarid environment. Zeitschrzflfür Geomorphologie, Supplementband, 21, 202-215.92Appendix A Air photograph sets used for debris torrent dating in the Rennell Soundarea.approximateyear photograph numbers scale1933/1937 BC 26: 30-35, 83-94; 1:15000BC 27: 51-54, 107-110 1:150001964/1966 BC 4358: 47-50, 109-114,118, 121-132, 193-199 1:160001976/1977 BC 7842: 74-45 1:21000BC 77063: 119-128, 226-232 1:20000BC 77064: 54-58 1:200001981 BC 81059: 137-145, 194-197,201-203, 209-225, 274-281,288-293 1:12000BC 81083: 1-15, 49-54 1:120001989 BC 89005: 105-107, 149-150,220-227, 234-241 1:1700093Appendix B Assessment of accuracy of techniques used in estimation of sediment cross-sectional area.distance measured estimatedbetween cross-sectional cross-sectionalcross-sections area area(m) (m2) (2)o 0.2 0.10 0.8 0.60 0.8 0.50 0.4 0.31 1.1 0.52 0.8 0.85 0.7 0.35 0.7 0.55 1.2 0.15 1.2 1.68 1.4 1.38 1.1 1.79 0.2 0.710 0.9 0.711 0.7 0.814 1.0 1.817 0.8 0.618 0.9 0.423 1.8 0.533mean 0.9 0.7variance 0.14 0.26F-ratio = 1.80 F1j: a = 0.01 = 3.08a = 0.05 = 2.19a 0.10 = 1.84= 1.12 a 0.01 = 2.70a = 0.05 = 2.02a = 0.10 = 1.6894Appendix C Estimated volume of material in storage.Table C. 1 Estimated volume of large organic debris.V, volume volume persingle LOD Vi, volume total LOD gully gully areagully piece (m3) log jam (m3) volume (m3) area (ha) (m3hal)old growthGi 21 9 30 0.21 143G2 30 10 40 0.29 138G3 59 20 79 0.20 39504 15 26 41 0.21 195G5 34 2 36 0.28 129G6 18 20 38 0.65 58G9 85 13 98 0.39 251GlO 6 1 7 0.56 13Gil 139 11 150 0.35 429G12 2 0 2 0.08 25R4 2 0 2 0.43 5P5 3 1 4 0.17 24Slog 1 0 1 0.20 5clear-cutBl <1 16 17 0.69 25B2 1 36 37 0.20 185B3 <1 0 1 0.49 2B4 <1 <1 2 0.21 10P2 0 0 0 0.12 0R3 1 1 2 0.24 8R7 1 0 1 0.17 6R8 2 6 8 0.37 22R9 10 0 10 0.21 48RiO <1 2 3 0.33 9R12 1 2 3 0.36 8R13 2 0 2 0.13 15R14 <1 0 1 0.09 11Slec 29 0 2 0.16 181SB1 0 0 0 0.07 0SB2 <1 0 1 0.13 895Table C.2 Estimated volume of of cross-sections volume perestimated & sediment gully area gully areagully measured volume (m3) (ha) (m3ha’)old growthGi 12 388 0.21 1848G2 18 576 0.29 1986G3 ii (3a) 339 0.20 1695G4 19 (3) 385 0.21 1833G5 17 432 0.28 1543G6 27 744 0.65 1145G9 17 231 0.39 592GlO 11 309 0.56 552Gil 15 468 0.35 1337G12 4 51 0.08 638R4 10(7) 228 0.43 530R5 16 121 0.17 712Slog 8 (3) 50 0.20 250clear-cutBl 21 365 0.69 529B2 15 310 0.20 1550B3 25 343 0.49 700B4 22 94 0.21 448R2 5 70 0.12 583R3 11 156 0.24 650R7 11(3) 71 0.17 418R8 19 (1) 68 0.37 184R9 10 (2) 174 0.21 829RiO 9 381 0.33 1155R12 7 107 0.36 297R13 9 (5) 95 0.13 731R14 9 53 0.09 589S1 6(3) 171 0.16 1069SB1 6 36 0.07 514SB2 8 14 0.13 108anumber of cross-sections measured96Table C.3 Estimated total volume of debris.volume perdebris gully area gully areagully volume (m3) (ha) (m3ha’)old growthGi 418 0.21 1990G2 616 0.29 2124G3 418 0.20 2090G4 426 0.21 2029G5 467 0.28 1668G6 782 0.65 1203G9 330 0.39 846GlO 316 0.56 564Gil 617 0.35 1763G12 53 0.08 663R4 229 0.43 533R5 124 0.17 729Slog 50 0.20 250mean 373 1266standarddeviation 223 697clear-cutBi 381 0.69 552B2 347 0.20 1735B3 343 0.49 700B4 94 0.21 448R2 70 0.12 583R3 158 0.24 658R7 72 0.17 424R8 76 0.37 205R9 184 0.21 876RiO 382 0.33 1158R12 110 0.36 306R13 96 0.13 738Ri4 54 0.09 600Si 200 0.16 1250SB1 36 0.07 514SB2 14 0.13 108mean 164 678standarddeviation 129 41597Appendix D Methods used and estimated dates of most recent debris torrents in gullies.gully ageapproximate at time ofdating number of torrent data collectiongully method’ trees cored date (avg. years)old growthGi 1,2,3 6 1978 14G2 1,2- 1978 14G3 1,2,3 5 1978 14G4 1,2,3 8 1978 14G5 1,2,3 1 1917 75G6 1,2,3 3 1935 57G9 1,2,3 6 1978 14GlO 1, 2, 3 8 1978 14Gil 1, 2, 3 4 1978 14G12 1,2- 1978 14R4 2,3 4 1986 6R5 2- 1977-87 10Slog 2, 3 3 1981-87 8clear-cutBi 4- 1991 1B2 2, 3 3 1981-87 8B3 2, 3 5 1981-84 9.5B4 2,3,4 7 1986 6R2 2,3 3 1989 3R3 2,3 2 1981 11R7 2- 1989-91 2R8 1,2,3 2 1985 7R9 2,3 4 1978 14RiO 1,3 1 1985 7R12 4- 1991 1R13 2, 3 4 1981-87 8R14 2, 3 3 1978 14Slec 2, 3 3 1981-87 8SB1 2, 3 1 1981-87 85B2 2- 1981-87 8a 1 = J. Schwab, unpublished data; 2 = air photographs; 3 = dendrochronology; 4 = interviewswith knowledgeable people in area.98Appendix E Prediction of magnitude of future events in sampled gullies.The magnitude of a future debris torrent in each of the sampled guffies wasestimated in two ways. First, the estimated volume of debris in a channel (Appendix C)was summed with the volume of the triggering slide (calculated from the measureddimensions). This approach is based on the assumption that all of the material in thechannel would be entrained by a torrent and the channel would be scoured to bedrock.This assumption is supported by the literature (Swanston and Swanson 1976, Townshend1979, Pearce and Watson 1983) and by the fact that gullies in the study area whichtorrented recently were completely scoured. This approach results in a minimumestimate, as it does not account for two possible sources of debris that may contributeadditional material to a torrent: (1) the volume of any sidewall material (e.g. weatheredbasal till, colluvium) that would likely be eroded during an event; or (2) the volume ofany living trees that may be entrained by a torrent.The second method of estimating magnitude uses the variable ‘sidewall erodibility’to modify a volume estimate (after Hungr et al. 1984, Bovis et al. 1988). Magnitude, M,is calculated as:M = (v + iZv)*e [E.1]where v is the estimated total debris volume (m3), iz,,, is the volume of the initiation zoneof the most recent debris torrent (measured in the field), and e is a dimensionless sidewallerodibiity coefficient, which incorporates sidewall steepness, height and stability (TableE. 1). Although this approach permits the use of the detailed calculation of volume ofdebris in a channel, it also may result in an underestimation of the magnitude of a futureevent, as it does not incorporate the volume of any living trees that may be entrained.The results of both estimation methods are listed in Table E.2.99Table E. 1 Sidewall erodibility rating.sidewallerodibilityheight’ stability1’ coefficient, estrong bedrock, or low bankH less than 1 meter 0H=1-1.99m stable 1H = 1 - 1.99 m moderately stable 2H = 1 - 1.99 m unstable 4H =2-4.99m stable 2H = 2 - 4.99 m moderately stable 4H = 2 - 4.99 m unstable 10H=5-lOm stable 4H = 5 - 10 m moderately stable 8H = 5 - 10 m unstable 20a height of sidewall likely to be affected by sliding and erosion.b stability categories:stable - intact, vegetated slope, angle less than 350•moderately stable - vegetated slope with less than 30% of surface affected by shallow sliding,potential sliding, creep or erosion, or a bare steep slope composed of dense material.unstable - slope with more than 30% of surface affected by shallow sliding, potential sliding, creep orerosion.(After Bovis et al. 1988)The magnitude estimates from method (1) are very similar between the old-growthand clear-cut gullies. However, if the estimates from method (2) are compared, it isapparent that future debris torrents from the clear-cut gullies will be larger than thosefrom the old-growth gullies. This is strongly related to the instability of the clear-cutsidewalls and demonstrates the significant volume of material that unstable sidewalls cancontribute to a torrent. However, these estimates are based on the assumption that atorrent will scour to the full height of the sidewall, which will likely occur only inextreme conditions. In addition, the estimates are not normalized by gully size, whichmakes it difficult to accurately compare the estimates from each land-treatment group.100Table E.2 Magnitude of future debris torrents in gullies.initiation method (1) sidewall method (2)zone volume magnitudez erodibility magnitudecgully (m3) (m3) coefficient, e b (m3)old growthGi 8 450 6 2700G2 10 650 8 5200G3 38 450 8 3600G4 75 500 8 4000G5 375 850 4 3400G6 375 1150 8 9200G9 400 750 8 6000GlO 2813 3150 8 25200Gil 150 750 8 6000G12 150 200 8 1600R4 750 1000 20 20000R5 89 200 8 1600Sid 300 500 8 4000clear-cutBi 480 850 20 17000B2 150 500 20 10000B3 281 600 20 12000B4 188 300 20 6000R2 392 450 10 4500R3 459 600 20 12000R7 294 350 4 1400R8 1346 1400 20 28000R9 405 600 7 4200RiO 823 1200 20 24000R12 2429 2550 8 20400R13 1247 1350 4 5400R14 170 200 4 800SB1 120 150 10 1500SB2 300 300 10 3000a sum of total debris volume and initiation zone volume.b function of sidewall height, steepness, relative stability, from Table E. 1.C product of magnitudea and sidewall erodibility coefficient, e.d Si debris torrent triggered in old growth and flowed through clear-cut - these two components treatedas single gully for magnitude calculations.


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items