Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Modelling channel morphodynamics associated with large wood in an intermediate-sized stream Davidson, Sarah 2011

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

Item Metadata

Download

Media
24-ubc_2011_fall_davidson_sarah.pdf [ 15.29MB ]
Metadata
JSON: 24-1.0072311.json
JSON-LD: 24-1.0072311-ld.json
RDF/XML (Pretty): 24-1.0072311-rdf.xml
RDF/JSON: 24-1.0072311-rdf.json
Turtle: 24-1.0072311-turtle.txt
N-Triples: 24-1.0072311-rdf-ntriples.txt
Original Record: 24-1.0072311-source.json
Full Text
24-1.0072311-fulltext.txt
Citation
24-1.0072311.ris

Full Text

Modeling Channel Morphodynamics Associated with Large Wood in an Intermediate-Sized Stream by Sarah Davidson B.Sc, McGill University, 2007 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF Master of Science in THE FACULTY OF GRADUATE STUDIES (Geography) The University Of British Columbia (Vancouver) October 2011 c￿ Sarah Davidson, 2011 Abstract The primary objective of this research is to investigate the relationship between wood load and reach-scale morphodynamics using a model system. Previous re- search has shown that large wood significantly impacts channel dynamics, espe- cially in small and intermediate sized forested streams where wood pieces are similar in length to channel width. Five experiments, each comprised of several five hour runs, were conducted using a stream table with wood loads scaled to 0 m3/m2, 0.011 m3/m2, 0.016 m3/m2, 0.022 m3/m2, and 0.028 m3/m2. The addi- tion of large wood significantly increased the flow resistance and decreased the reach-averaged velocity in all experiments. These hydraulic changes were associ- ated with decreased sediment transport and increased sediment storage within the reach. Over time, increases in bed and water surface slope compensated for the loss of potential energy to flow resistance and enabled the system to reach a new steady state. The heterogeneity in the spatial distribution of the hydraulic changes – with flow velocity and shear stress decreasing upstream of obstructions and increasing downstream of log steps – increased the facies complexity and pool frequency in the reach at the new steady state, and thereby increased habitat complexity. These results show that wood load is a primary control on channel morphodynamics and the availability of aquatic habitat in intermediate sized streams. ii Table of Contents Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii Table of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iv List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vi 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.1 Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.2 Large Wood in Stream Networks . . . . . . . . . . . . . . . . . . 3 1.3 Effects of In-Channel Large Wood . . . . . . . . . . . . . . . . . 5 1.4 Disturbance and Large Wood . . . . . . . . . . . . . . . . . . . . 7 1.5 Stream Restoration . . . . . . . . . . . . . . . . . . . . . . . . . 8 1.6 Research Objectives . . . . . . . . . . . . . . . . . . . . . . . . . 10 2 Experimental Design . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 2.1 Prototype System . . . . . . . . . . . . . . . . . . . . . . . . . . 12 2.2 Physical Model . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 2.3 Study Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 3 Large Wood Dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . 20 3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 3.2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 3.2.1 Large Wood and Jam Dynamics . . . . . . . . . . . . . . 21 iii 3.2.2 Hydraulic Effectiveness . . . . . . . . . . . . . . . . . . 22 3.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 3.3.1 Distance and Timing of Wood Transport . . . . . . . . . . 23 3.3.2 Jam Frequency and Size . . . . . . . . . . . . . . . . . . 27 3.3.3 Hydraulic Effectiveness . . . . . . . . . . . . . . . . . . 36 3.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 4 Stream Hydraulics and Sediment Transport . . . . . . . . . . . . . . 43 4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 4.2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 4.2.1 Flow Velocity . . . . . . . . . . . . . . . . . . . . . . . . 44 4.2.2 Water Stage . . . . . . . . . . . . . . . . . . . . . . . . . 48 4.2.3 Shear Stress Partitioning . . . . . . . . . . . . . . . . . . 49 4.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 4.3.1 Flow Velocity . . . . . . . . . . . . . . . . . . . . . . . . 50 4.3.2 Water Stage . . . . . . . . . . . . . . . . . . . . . . . . . 57 4.3.3 Shear Stress Partitioning . . . . . . . . . . . . . . . . . . 65 4.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 5 Channel Morphology . . . . . . . . . . . . . . . . . . . . . . . . . . 73 5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73 5.2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 5.2.1 Sediment Transport and Storage . . . . . . . . . . . . . . 75 5.2.2 Channel Morphology . . . . . . . . . . . . . . . . . . . . 77 5.2.3 Sedimentological Characteristics . . . . . . . . . . . . . . 77 5.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78 5.3.1 Sediment Transport and Storage . . . . . . . . . . . . . . 78 5.3.2 Channel Morphology . . . . . . . . . . . . . . . . . . . . 90 5.3.3 Sedimentological Characteristics . . . . . . . . . . . . . . 101 5.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111 6.1 Large Wood Dynamics . . . . . . . . . . . . . . . . . . . . . . . 111 6.2 Channel Hydraulics and Morphology . . . . . . . . . . . . . . . . 112 iv 6.3 Implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115 6.4 Future Research . . . . . . . . . . . . . . . . . . . . . . . . . . . 116 Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118 v List of Tables Table 2.1 Scaling relations for wood. . . . . . . . . . . . . . . . . . . . 14 Table 2.2 Scaling relations . . . . . . . . . . . . . . . . . . . . . . . . . 14 Table 2.3 Experimental treatments . . . . . . . . . . . . . . . . . . . . . 18 Table 2.4 Naturally occuring wood loads . . . . . . . . . . . . . . . . . 19 Table 3.1 Jam classification . . . . . . . . . . . . . . . . . . . . . . . . 29 Table 3.2 Significant relationships between wood metrics . . . . . . . . . 32 Table 3.3 Reach-scale wood characteristics . . . . . . . . . . . . . . . . 34 Table 3.4 Jam formation . . . . . . . . . . . . . . . . . . . . . . . . . . 35 Table 4.1 Velocity comparison . . . . . . . . . . . . . . . . . . . . . . . 51 Table 4.2 Significant relationships between wood metrics and flow velocity 56 Table 4.3 Water stage and surface gradient . . . . . . . . . . . . . . . . 57 Table 4.4 Significant relationships between wood metrics and water stage change . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 Table 4.5 Significant relationships between wood metrics and water sur- face gradient . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 Table 4.6 Shear stress comparison . . . . . . . . . . . . . . . . . . . . . 65 Table 4.7 Significant relationships between wood metrics and shear stress partitioning . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 Table 4.8 Significant relationships between wood metrics and Manning’s n 69 Table 5.1 Facies classification . . . . . . . . . . . . . . . . . . . . . . . 78 Table 5.2 Sediment storage . . . . . . . . . . . . . . . . . . . . . . . . . 83 Table 5.3 Significant relationships between wood metrics and sediment storage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86 vi Table 5.4 Pool characteristics . . . . . . . . . . . . . . . . . . . . . . . 91 Table 5.5 Significant relationships between wood metrics and pool char- acteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94 Table 5.6 Morphologic characteristics . . . . . . . . . . . . . . . . . . . 98 Table 5.7 Significant relationships between woodmetrics and channel mor- phology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99 Table 5.8 Facies characteristics . . . . . . . . . . . . . . . . . . . . . . 103 Table 5.9 Significant relationships between wood metrics and sedimento- logical characteristics . . . . . . . . . . . . . . . . . . . . . . 106 vii List of Figures Figure 2.1 The model apparatus and large wood pieces . . . . . . . . . . 15 Figure 2.2 Rotating sediment feeder . . . . . . . . . . . . . . . . . . . . 16 Figure 2.3 Cumulative plots of sediment storage . . . . . . . . . . . . . 17 Figure 3.1 Boxplots of mean velocity for pieces with varying piece length, rootwad presence and recruitment angle . . . . . . . . . . . . 24 Figure 3.2 Changes in wood orientation over time . . . . . . . . . . . . . 26 Figure 3.3 Wood transport over time . . . . . . . . . . . . . . . . . . . . 27 Figure 3.4 Orientation of individual pieces and jam members with rootwads 28 Figure 3.5 Orientation of all pieces with rootwads . . . . . . . . . . . . 28 Figure 3.6 Flow deflection jam . . . . . . . . . . . . . . . . . . . . . . . 30 Figure 3.7 Valley jam . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 Figure 3.8 Unstable jam . . . . . . . . . . . . . . . . . . . . . . . . . . 31 Figure 3.9 Bar apex jam . . . . . . . . . . . . . . . . . . . . . . . . . . 31 Figure 3.10 Volume versus number of pieces in jams . . . . . . . . . . . . 33 Figure 3.11 Jam frequency versus number of suspended pieces per unit area of channel bed . . . . . . . . . . . . . . . . . . . . . . . 34 Figure 3.12 Boxplots of mean blockage ratio for individual pieces and jam members . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 Figure 3.13 Orientation of jam members and individual pieces . . . . . . . 37 Figure 3.14 Mean blockage ratio versus the percentage of pieces in jams . 38 Figure 3.15 Dimensionless projected area versus wood load and jam fre- quency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 Figure 4.1 Filtering of electrical conductivity . . . . . . . . . . . . . . . 47 viii Figure 4.2 Boxplots of velocities derived from peak, harmonic mean and centroid methods . . . . . . . . . . . . . . . . . . . . . . . . 52 Figure 4.3 Boxplots of the effects of wood on velocity . . . . . . . . . . 53 Figure 4.4 Mean velocity versus wood load and piece frequency . . . . . 54 Figure 4.5 Mean velocity versus blockage ratio and dimensionless pro- jected area . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 Figure 4.6 Longitudinal water surface profiles . . . . . . . . . . . . . . . 58 Figure 4.7 Water stage increase versus wood load and piece frequency . . 60 Figure 4.8 Water stage increase versus mean blockage ratio and dimen- sionless projected area . . . . . . . . . . . . . . . . . . . . . 61 Figure 4.9 Mean water surface slope versus wood load and piece frequency 62 Figure 4.10 Mean water surface slope versus jam frequency and dimen- sionless projected area . . . . . . . . . . . . . . . . . . . . . 63 Figure 4.11 Relative shear stress partitioning versus wood load and piece frequency. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66 Figure 4.12 Relative Manning’s roughness versus wood load and piece fre- quency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 Figure 5.1 Very fine facies . . . . . . . . . . . . . . . . . . . . . . . . . 79 Figure 5.2 Fine facies . . . . . . . . . . . . . . . . . . . . . . . . . . . 79 Figure 5.3 Medium facies . . . . . . . . . . . . . . . . . . . . . . . . . 80 Figure 5.4 Coarse facies . . . . . . . . . . . . . . . . . . . . . . . . . . 80 Figure 5.5 Very coarse facies . . . . . . . . . . . . . . . . . . . . . . . . 81 Figure 5.6 Relative sediment transport over time . . . . . . . . . . . . . 82 Figure 5.7 D50 and D90 versus sediment output rate . . . . . . . . . . . . 83 Figure 5.8 Comparison of storage estimates . . . . . . . . . . . . . . . . 85 Figure 5.9 Sediment storage versus piece frequency . . . . . . . . . . . 87 Figure 5.10 Storage efficiency versus wood load . . . . . . . . . . . . . . 88 Figure 5.11 Storage-transport relations . . . . . . . . . . . . . . . . . . . 89 Figure 5.12 Actual and predicted storage volume over time . . . . . . . . 92 Figure 5.13 Longitudinal thalweg profile . . . . . . . . . . . . . . . . . . 93 Figure 5.14 Pool spacing and total pool length versus the percentage of pieces in jams . . . . . . . . . . . . . . . . . . . . . . . . . . 95 ix Figure 5.15 Percentage of log-affected pools versus piece frequency . . . 96 Figure 5.16 Boxplots of mean pool length and depth . . . . . . . . . . . . 97 Figure 5.17 Median bed gradient versus mean blockage ratio . . . . . . . 100 Figure 5.18 Mean bed gradient versus wood load and piece frequency . . . 101 Figure 5.19 Relative mean bed gradient versus dimensionless projected area and jam frequency . . . . . . . . . . . . . . . . . . . . . . . 102 Figure 5.20 Relative facies number versus mean blockage ratio . . . . . . 104 Figure 5.21 Relative D50 versus the precentage of pieces in jams and the dimensionless projected area . . . . . . . . . . . . . . . . . . 105 x Acknowledgments I would first like to thank Brett Eaton for his continued support through every stage of this project. Completion of this thesis would not have been possible without his technical expertise, enthusiasm, and thoughtful editing. I am also thankful to Dave Luzi and Jacqueline Armstrong for sharing their technical knowledge during the early stages of the research. This thesis also benefitted from the advice of Rob Millar, Mike Church, Dan Moore, and Marwan Hassan. I am deeply grateful for the support of my family. I would like to thank my parents for providing encouragement – as well as many weekends of childcare, often on short notice – when it was most needed. And most importantly, to my son Austin: thank you for lending me perspective during the hard times, and for giving me the motivation to complete this work. You are a constant source of inspiration to me. xi Chapter 1 Introduction Due to the branching nature of stream networks, small and intermediate sized streams represent a large proportion of drainage networks (Hassan et al., 2005). Since the 1970s research has repeatedly shown that where wood is similar in size to the bankfull channel width, it exerts a strong influence on stream function. Throughout most of the 20th century, however, wood was removed from streams to facilitate transport, enhance fish passage, and to prevent flooding (Bisson et al., 1987; Gippel, 1995; Brooks et al., 2004; Lester and Boulton, 2008; Nagayama and Nakamura, 2010). The addition of large wood has now become a common stream restoration practice (Bernhardt et al., 2005; Sweka and Hartman, 2006; Kail et al., 2007; Lester and Boulton, 2008; Nagayama and Nakamura, 2010). While the effects of large wood on channel processes have been described extensively, a quantitative understanding of these impacts remains lacking. In this chapter the existing literature regarding wood distribution, the impacts of in-stream wood, and the effects of natural disturbance and stream restoration projects will be reviewed. 1.1 Definition The terminology used to describe large wood pieces varies throughout the liter- ature. The earliest studies referred to large wood pieces as large organic debris (LOD; Beschta, 1979; Keller and Tally, 1979), large organic matter (LOM; Keller and Swanson, 1979), organic debris (OD; Mosley, 1981), coarse woody debris 1 (CWD; Nakamura and Swanson, 1993; Lisle, 1995), or large woody debris (LWD; Bilby and Ward, 1991; Beechie and Sibley, 1997; Gurnell and Sweet, 1998). The term large wood (LW) has most recently been adopted (Brooks et al. 2004, 2006). Regardless of the terminology used to describe it, large wood is commonly de- fined as any piece of wood (including branches, stems and rootwads) greater than a threshold diameter and length. The threshold diameter is nearly universally defined as 0.1 m, while the threshold length varies from 1-2 m. Various terminology and definitions have also been adopted to refer to accu- mulations of large wood. The term ‘dam’ has been used to refer both to individual pieces of wood which span a channel, as well as accumulations of large wood. Def- initions of debris ‘jams’ also vary; jams have variously been defined as accumula- tions of five or more key pieces (Nakamura and Swanson, 1993) or accumulations of five or more clustered wood pieces (Beechie and Sibley, 1997). More generally, jams have been defined as stacked log accumulations (Brooks et al., 2006), loca- tions where wood pieces interact to produce aggregate patterns (Bocchiola et al., 2008), or complex, porous accumulations of heterogeneous material (Manners et al., 2007). While the term ‘accumulation’ generally broadly refers to a grouping of two or more pieces of wood, the term has been qualified by some as a structure composed of at least three pieces of wood, with at least 2 points of contact (Czarnomski et al., 2008). This qualification of ‘accumulation’ is analogous to the more re- cent process-based definitions of ‘jams’ which specify the presence of interactions among the individual pieces (Bocchiola et al., 2008). The formation of an aggre- gate structure, and the accompanying implications for channel hydraulics, require that individual pieces of wood interact; multiple pieces of wood do not constitute an accumulation or a jam if they do not interact to produce a cumulative effect on stream processes. Finally, while the terms ‘log step’ and ‘log jam’ are most often used inter-changeably, some authors have distinguished between these structures on the basis of size (Faustini and Jones, 2003). For the purposes of this research, ‘jams’ are defined as complex, porous accumulations of at least three pieces of wood with at least two points of contact. 2 1.2 Large Wood in Stream Networks The quantity and distribution of wood in a channel depends on wood recruitment rates, as well as the location of a reach within the stream network. The volume of in-stream large wood available to influence channel processes is fundamentally a function of the relative rates of recruitment and export of riparian trees. While the creation of sediment budgets has become commonplace in geomorphological liter- ature, the development of analogous wood budgets is a relatively new phenomenon (Benda and Sias, 2003; Hassan et al., 2005; Czarnomski et al., 2008). Wood recruitment results from both chronic and episodic inputs from ripar- ian forests (Bisson et al., 1987; Fetherston et al., 1995; Czarnomski et al., 2008). Chronic inputs are dominated by mortality and subsequent toppling of stream- side trees, as well as input of live or dead trees through lateral bank erosion, while episodic inputs result from disturbances which increase tree mortality or top- pling, including fire, wind throw, insect infestation, and hillslope mass movements (Benda and Sias, 2003). These recruitment processes are associated with a range of temporal scales; increases in recruitment associated with rapid mass movements, such as avalanches and debris flows, last only minutes, while increased recruit- ment following fire or insect infestation spans decades (Bragg, 2000; Benda and Sias, 2003). The relative importance of these input processes varies according to riparian forest characteristics. Chronic inputs, especially those related to tree mortality, dominate in mature forest stands. Because it takes approximately 100 years af- ter a significant disturbance in coniferous coastal forests before chronic mortality- related inputs resume (Benda and Sias, 2003), episodic disturbances such as fire may cause immediate increases in wood input, both due to increased debris flow activity and slow input of burned wood, followed by a long period of reduced wood recruitment (Bragg, 2000; Naiman et al., 2002; Benda and Sias, 2003). Similarly, while timber harvesting often inputs large quantities of woody debris into stream systems (as a result of both harvesting practices and subsequent mass movement processes), the effect is often short lived as the recruited wood is usually easily degraded, and second-growth forests provide vastly reduced quantities of the large wood needed to initiate jam formation (Bilby and Ward, 1991). 3 Wood export is a function of transport processes as well as abrasion and in situ decomposition (Naiman et al., 2002; Hassan et al., 2005). Generally there is an inverse relationship between wood load, defined as the volume of wood per unit of channel area, and stream size or drainage area (Keller and Tally, 1979; Lienkaem- per and Swanson, 1987; Bilby and Ward, 1991; Gippel, 1995; Bragg, 2000; Wohl and Jaeger, 2009). Assuming uniform rates of wood recruitment throughout a chan- nel network, the relative volume of in-channel wood must vary with stream size: as channel width increases the relative bank length per unit of channel area decreases. Further compounding this effect, transport is fundamentally dependent on wood size relative to channel dimensions (Gurnell et al., 2002). Channel width is such an important determinant of wood dynamics that channel size is often defined rel- ative to tree height (Church, 1992; Gurnell et al., 2002; Piegay, 2003). In small streams where piece size is large relative to channel dimensions, transport requires either high flow conditions or channelized mass movements (Benda and Sias, 2003) and many wood pieces remain suspended across the channel banks (Bisson et al., 1987; Fetherston et al., 1995; Wohl and Jaeger, 2009). Thus, transport is rare and wood storage is significant. Conversely, in large streams wood is transported downstream until an obstacle is encountered (Lienkaemper and Swanson, 1987; Wohl and Jaeger, 2009); wood accumulations therefore become less frequent, but larger in size, in the downstream direction (Gurnell et al., 2002). In intermediate channels wood pieces are similar in size to channel dimensions and form frequent jams (Wohl and Jaeger, 2009). Current conceptual models suggest that wood abundance decreases predictably in the downstream direction as stream size becomes sufficiently large that wood is easily transported, and channel-spanning jams become rare (Fetherston et al., 1995; Gurnell et al., 2002; Naiman et al., 2002; Wohl and Jaeger, 2009). Inves- tigations of the relationship between wood load - the volume of wood per unit area of channel bed (m3/m2) - and channel width or drainage area, which both increase downstream, have largely corroborated this conceptual model of longitu- dinal wood distribution (Lienkaemper and Swanson, 1987; Bilby and Ward, 1991; Gippel, 1995; Bragg, 2000; Wohl and Jaeger, 2009). Relationships between wood load and channel elevation or slope, both of which decrease in the downstream direction, are less clear. 4 1.3 Effects of In-Channel Large Wood In-stream wood pieces act as large roughness elements which dissipate a signifi- cant proportion of the potential energy in a channel through flow resistance. Field studies suggest that large wood contributes up to 98% of the elevation loss in intermediate-sized, forested streams (Heede, 1972; Keller and Swanson, 1979; Keller and Tally, 1979; Bilby and Likens, 1980; Thompson, 1995; Faustini and Jones, 2003; Andreoli et al., 2007). Log jams create a stepped profile in which low gradient reaches are punctuated by sudden elevation drops at log steps. As a result, a much larger proportion of the channel is at a low gradient in channels with significant wood accumulations (Faustini and Jones, 2003), and there may be less energy available to transport sediment and to erode the bed and banks (Heede, 1972; Gippel, 1995; Lisle, 1995). Thus, channels with significant large wood and frequent log steps may have more stable banks, despite local increases in bank erosion where wood deflects flow toward the banks (Hassan et al., 2005). Reductions in the magnitude of the shear stress acting on the bed promote de- position of sediment, primarily upstream of channel obstructions (Fetherston et al., 1995; Gippel, 1995; Abbe and Montgomery, 2003; Brooks et al., 2004, 2006). A large proportion of the storage in forested channels, representing multiple times the annual sediment yield, is directly attributed to wood (Bilby, 1981; Thomp- son, 1995); comparisons of the volume of sediment stored by wood and annual sediment yields show that the volume stored is 1.5-30 times the annual sediment input to stream channels (Megahan, 1982; Hassan et al., 2005; Andreoli et al., 2007). The deposition of scoured material upstream of obstructions increases bar amplitude and may force a transition from plane-bed to riffle-pool morphologies (Montgomery and Buffington, 1997; Brooks et al., 2004). While the net effect of the increased channel roughness provided by log steps is to decrease grain shear stress, roughness elements also concentrate flow ob- structions, inducing local scour. The most pronounced morphological effect of large wood is often an increase in pool frequency. A significant percentage of pools in high gradient streams are associated with wood accumulations (Keller and Tally, 1979; Beechie and Sibley, 1997; Jackson and Sturm, 2002; Faustini and Jones, 2003). In channels with erodible banks, scour around large wood may also 5 cause local widening (Robison and Beschta, 1990; Nakamura and Swanson, 1993; Thomspon, 1995; Abbe and Montgomery, 2003). The alternating regions of high and low transport capacity associated with wood obstructions also produce local variations in the surface grain size distri- bution. Large wood steps promote the deposition of fine sediment upstream of obstructions, while coarsening the bed downstream where flow is concentrated, thereby enhancing facies complexity (Buffington and Montgomery, 1999b; An- drews, 2010). The net effect of wood addition is generally a decrease in surface grain size (Manga and Kirchner, 2000), which creates a negative feedback by en- hancing sediment export from storage reservoirs (Lisle and Church, 2002). The effects of large wood on channel processes and reach morphology vary with stream size, gradient, and flow level. The strength of the relationship between wood abundance and pool frequency decreases with decreasing gradient, and in- creasing stream size (Keller and Tally, 1979; Beechie and Sibley, 1997). Given that significant step-pool formation (as opposed to the development of small localized scour pools) requires the blockage of streams, it is reasonable that pool formation should decrease with increasing channel size; in larger channels there are less fre- quent complete blockages of streams due to the smaller size of the wood relative to the channel width. In lower gradient streams, pool frequency is instead dictated by the presence of riffles and bars (Fetherston et al., 1995; Montgomery and Buff- ington, 1997; Gurnell and Sweet, 1998). Similarly, the strength of the relationship between wood abundance and sediment storage decreases with decreasing stream gradient, and increasing stream size (Thompson, 1995). The hydraulic and morphologic conditions associated with large wood are often favourable for aquatic fauna. Fish abundance is generally strongly correlated to pool volume, which is in turn determined by the amount of boulders and wood in a reach (Fausch and Northcote, 1992; Sweka and Hartman, 2006). Pools dampen the effects of flow fluctuations as they provide refuge from high velocities at high flows, as well as refuge from drought at low flows (Hakala and Hartman, 2004; Lester and Boulton, 2008). More directly, large wood provides cover for fish, as well as a substrate for biofilm growth (Lester and Boulton, 2008). Wood presence is also associated with a decreased nutrient sprialling length and increased secondary production (Gurnell et al., 2002), while coarse substrates in scour pools provide 6 spawning habitat (Bisson et al., 1987; Floyd et al., 2009; Nagayama and Nakamura, 2010). 1.4 Disturbance and Large Wood Wood recruitment, which largely determines the quantity of wood available to alter channel morphology, is affected by both natural and anthropogenic disturbances. Anthropogenic disturbance has historically involved wood removal from streams (Bisson et al., 1987; Gippel, 1995; Erskine and Webb, 2003; Nagayama and Naka- mura, 2010), as well as disruption of wood recruitment through riparian land use practices (Brooks et al., 2004; Sweka and Hartman, 2006; Kail et al., 2007; Lester and Boulton, 2008). Field studies involving wood removal, which was common in North America prior to the 1970s, suggest that it promotes increased transport from sediment reservoirs, resulting in sedimentation of downstream scour pools (Beschta, 1979; Bilby, 1981; Mosley, 1981; Smith et al., 1993; Gurnell and Sweet, 1998). By increasing sediment transport from upstream storage reservoirs and enhancing sedimentation in downstream scour pools, wood removal reduces bed complexity (Bilby and Ward, 1991). By decreasing wood recruitment, riparian logging and land clearing have been shown to decrease in-stream wood load (Richmond and Fausch, 1995; Faustini and Jones, 2003; Czarnomski et al., 2008). This, in turn, has been related to decreased in-stream sediment storage and pool frequency (Megahan, 1982; Bilby and Ward, 1991; Richmond and Fausch, 1995). Despite changes in forestry practices since the 1970s to prohibit removal of in-stream wood and protect riparian vegetation, streams remain depleted in second-growth stands, as well as urban and agricul- tural areas (Kail et al., 2007; Lester and Boulton, 2008; Nagayama and Nakamura, 2010). Natural disturbances to riparian forests also affect channel processes, primarily through changes in wood recruitment. Recent research suggests that while sys- tems dominated by chronic inputs may develop stable wood loading conditions, systems driven by recurrent disturbances (e.g. fire or insect infestation) experi- ence prolonged oscillations in wood recruitment, wherein initial increases in wood recruitment are followed by extended periods of depletion as riparian forests re- 7 generate (Bragg, 2000; Benda and Sias, 2003). These systems may never reach a steady state of wood loading. Given the importance of in-stream wood in deter- mining channel hydraulics, frequently disturbed channels likely experience cyclic oscillations in channel morphology due to repeated endogenous changes in bank strength and wood recruitment (Eaton and Giles, 2009). 1.5 Stream Restoration As the ecological role of large wood has become increasingly accepted, the intro- duction of large wood has gained popularity as a tool for restoration in degraded, wood depleted streams. Reviews of documented restoration projects suggest that wood addition has become common practice in North American streams (Bern- hardt et al., 2005; Nagayama and Nakamura, 2010), but is less frequently used in Europe (Kail et al., 2007) and Australia (Erskine andWebb, 2003; Lester and Boul- ton, 2008), as well as Asian and developing countries (Nagayama and Nakamura, 2010). According to these reviews, wood installation is most often employed in moderate sized (<20 m width), high gradient (>1%), gravel bed streams (Lester and Boulton, 2008; Nagayama and Nakamura, 2010). Installation techniques vary according to stream type. ‘Hard engineering’ tech- niques, in which wood is fixed in place, are most commonly used despite sig- nificantly higher installation costs compared with the addition of un-fixed wood (Sweka and Hartman, 2006; Kail et al., 2007). Log dams or drop structures, ori- ented either perpendicular or obliquely to the flow, are the most common struc- ture in moderate sized streams (Nagayama and Nakamura, 2010). Engineered log jams, designed to simulate natural log accumulations, are generally used in larger streams where the channel width is greater than the length of the largest wood pieces (Brooks et al., 2004, 2006; Nagayama and Nakamura, 2010). Despite the increased use of large wood for stream restoration in North Amer- ica, the effects of wood addition on physical habitat and ecological diversity remain relatively unknown, as 42-90% of projects in North America and Europe receive no post-implementation monitoring (Bash and Ryan, 2002; Bernhardt et al., 2005, Kail et al., 2007). Furthermore, perhaps as a result of the lack of replicates in such projects, few scientific studies concerning the impacts of wood addition have been 8 published (Nagayama and Nakamura, 2010). Reported results suggest that wood addition has the potential to alter both chan- nel morphology and fish abundance. Morphological adjustments generally begin during the first high flows following wood addition (Brooks et al., 2004; Kail et al., 2007) as gravel accumulates upstream of wood structures (Brooks et al., 2006; Floyd et al., 2009). Pool area generally increased following wood installation in streams with gradients below 3% (Hilderbrand et al., 1997; Warren and Kraft, 2003; Brooks et al., 2004, 2006; Sweka and Hartman, 2006; Nagayama and Naka- mura, 2010). The addition of large wood was often associated with increases in fish abundance. By increasing winter habitat, large wood installation commonly increased smolt abundance in the following spring (Cedarholm et al., 1997; So- lazzi et al., 2000; Roni and Quinn, 2001; Johnson et al., 2005; Nagayama and Nakamura, 2010). Gravel deposition also provided addition spawning habitat and increased redd abundance (Crispin et al., 1993; Floyd et al., 2009). The relationship between fish habitat and abundance is complex; measurable increases in pool area and habitat quality had no effect on fish abundance in the Williams River, Australia (Brooks et al., 2004, 2006), while fish abundance in- creased in West Virginian streams following wood emplacement while pool area remained constant (Sweka and Hartman, 2006). The effect on fish abundance was also scale-dependent, with greater increases in abundance in studies involv- ing longer experimental reaches (Johnson et al., 2005; Sweka and Hartman, 2006). These results suggest that while wood addition generally increases fish habitat – especially in moderate sized gravel bed streams with gradients of 1-3% – increases in actual fish abundance depend on additional biological and environmental factors. According to these results, restoration is most successful when it mimics nat- ural processes of wood recruitment. ‘Soft engineering’ techniques, involving the addition of un-fixed wood, produce more natural channel features, as wood stabi- lizes in locations where it would naturally occur (Kail et al., 2007). The addition of whole trees, including root wads and branches, has also been shown to improve wood stability, enhance pool area, and provide better cover for fish (Kail et al., 2007; Nagayama and Nakamura, 2010). Further, improvements in fish abundance and richness are likely related to the extent of channel rehabilitated, suggesting that the scale of restoration projects should be maximized (Nagayama and Nakamura, 9 2010). Longer-term monitoring of projects is also needed to account for hysteresis in reversing degradation, as fish response to channel improvements may occur over a longer time scale than the initial response to degradation (Brooks et al., 2006). Greater monitoring must be accompanied by increased reporting of results (Na- gayama and Nakamura, 2010). Finally, an increased focus on process-based, or passive, restoration is recommended (Sweka and Hartman, 2006; Kail et al., 2007; Beechie et al., 2010; Nagayama and Nakamura, 2010). Active restoration, in- cluding the installation of fixed or un-fixed wood, must be considered an interim measure, and used only to improve aquatic habitat in the short term while natural processes of wood recruitment are restored. 1.6 Research Objectives While the importance of large wood has been established, a quantitative under- standing of the effects of large wood remains elusive. The ability to test hypotheses through manipulation of wood loads in the field is restricted by the large tempo- ral and spatial scales at which the relevant processes operate; it is both difficult and costly to remove large volumes of wood, and nearly impossible to measure the processes at the high flows (Manners et al., 2007). The majority of studies thus far have therefore been phenomenological, involving comparisons between reaches with differences in wood loading (Heede, 1972; Keller and Swanson, 1979; Naka- mura and Swanson, 1993; Beechie and Sibley, 1997; Abbe andMontgomery, 2003; Andreoli et al., 2007). While the limited investigations involving experimental re- moval of wood lend insight into the system dynamics in the short term following decreases in wood load, they reveal little about the effectiveness of wood addition for stream restoration, or the effects of natural fluctuations in wood loading. Studies involving physical models present a unique opportunity to investigate the wood-channel dynamics at larger simulated temporal and spatial scales. Most flume studies to date have focused primarily on wood transport processes, rather than the actual dynamics associated with stable, in-stream large wood (e.g. Brau- drick et al., 1997; Braudrick and Grant, 2001; Bocchiola et al., 2006). Of the few studies that examine the impacts of in situ wood on stream morphology, most 10 have investigated the effects of individual wood pieces, rather than accumulations (e.g. Cherry and Beschta, 2001; Bocchiola et al., 2008). The primary objective of this research is to quantitatively evaluate the effects of large wood on channel dy- namics using a physical model. More specifically, the research aims to investigate the dynamics of in-stream large wood, as well as the effects of wood addition and removal on channel hydraulics and bed morphology. 11 Chapter 2 Experimental Design Physical models present a unique opportunity to investigate the effects of wood on channel dynamics and bed morphology at larger simulated temporal and spatial scales than is feasible in field studies. Most flume studies to date have focused primarily on wood transport processes, rather than the actual dynamics associated with in situ large wood (e.g. Braudrick et al., 1997; Braudrick and Grant, 2001; Bocchiola et al., 2006). A series of five experiments were conducted on a stream ta- ble – modeled after a prototype stream – to investigate the effects of wood addition on flow hydraulics and channel morphology in an attempt to bridge the existing gap in the geomorphological literature. 2.1 Prototype System Fishtrap Creek, the natural prototype for the model system, is located in the interior of British Columbia, approximately 50 km north of Kamloops. The stream is a tributary to the larger North Thompson river and drains a watershed area of 158 km2, nearly two-thirds of which (68% or 98 km2) was burned by a high intensity forest fire in August 2003 (Eaton et al., 2010). The Fishtrap Creek study reach is located directly upstream of the Water Sur- vey of Canada stream gauge in a channel segment bounded by floodplain on either side. The original 130 m study reach established in 2004 was expanded in both directions in 2006 to include a total length of 440 m (Eaton et al., 2010). The study 12 reach is relatively steep (1.5-2.0%) with a coarse bed; the median grain size (D50) of the surface is approximately 50-55 mm, while the D95 is 128 mm. The mean annual peak flow is approximately 7.5 m3/s, and occurs almost exclusively during the freshet in April through June (Eaton et al., 2010). The average bankfull width of the channel is approximately 10-12 m, suggest- ing a high potential for wood retention; Fishtrap Creek is classified as an intermedi- ate or medium sized stream, according to the definitions of channel size presented in section 1.2, with a ratio of maximum piece length to bankfull width of approx- imately 2:1. The reach has a jam frequency of 1.25 jams per 100 m, which is equivalent to an average jam spacing of 6.7 channel widths. The average wood load in the reach is 0.0206 m3/m2 (Andrews, 2010). The size distribution of the 192 wood pieces surveyed within the study reach is presented in Table 2.1. The morphology of the study reach has changed since the 2003 fire in response to endogenous changes in both bank strength and wood loading. Dramatic channel widening occurred 2007 despite average peak flows in all post-fire years. This widening is likely due to a dramatic decrease in bank strength associated with decay of dead riparian root networks (Beechie et al., 2006; Eaton and Giles, 2009). As a result, the study reach has shifted from a plane-bed to a riffle-pool morphology as bar amplitudes have increased since the fire (Montgomery and Buffington, 1997; Eaton et al., 2010). 2.2 Physical Model A series of experiments was conducted on a 5 m by 0.85 m stream table, configured to model the prototype stream described in section 2.1. The model stream repre- sented an approximate Froude-scale model of the study reach at Fishtrap Creek, created from Styrofoam within the larger stream table (figure 2.1). The ratio of the model dimensions to the prototype dimensions was constant and the relative width, depth, and bed material grain size (Wr, dr and Dr) were all represented by a characteristic relative length (Lr) of about 1/30. Discharge and time were scaled according to the following equations: Tr = L0.5r (2.1) 13 Qr = L2.5r (2.2) where Tr is the relative time and Qr the relative discharge. The resulting channel dimensions and governing conditions are shown in Table 2.2. Table 2.1: The results of field surveys, as well as the scaled wood sizes, are shown. Prototype Model Piece Length (m) Frequency (%) Piece Length (m) L/Wb Frequency (%) 2–4 27 0.1 0.29 30 4–8 36 0.2 0.58 36 8–16 35 0.4 1.2 34 16–32 2 - - - Large wood was also scaled geometrically, with piece lengths and frequency (table 2.1) selected based on results from wood surveys performed in the prototype study reach in 2007-2009 (Andrews, 2010). The 0.1 m and 0.2 m pieces were designed to represent segments of full (0.4 m) trees (figure 2.1). Table 2.2: Parameter lengths for the prototype and model systems are com- pared. Parameter Prototype Model Width 10-12 m 0.34 m D50 35 mm 1.4 mm Peak Discharge 7.5 m3/s 1.65 L/s Time 27 hours 5 hours The largest size class (16-32 m), which contributed only 2% of the large wood surveyed in Fishtrap Creek, was not included in the experiments. It was expected that such large wood, having a ratio of length to bankfull width of > 1.5, would remain suspended accross the channel for the duration of the experiments. Also, 0.4 m pieces representing the second largest size class (8-16 m) were considered sufficiently large to act as key members. 14 Figure 2.1: a) The stream table apparatus, and b) Wood pieces measuring 0.4 m, 0.2 m, and 0.1 m in length. 2.3 Study Design Five experiments were conducted to determine the effects of wood addition on the morphodynamics of the reach. Each experiment was comprised of a sequence of 5 hour runs, each designed to represent a single day of peak annual flow (1.65 L/s in the model or 7.5 m3/s in the prototype system). Assuming that only peak flows are morphologically effective, each run was equivalent to a year of effective flow. Throughout the runs sediment was added at the upstream end at a rate of 55-60 g/min using a rotating feeder (figure 2.2). Runs were first conducted without wood to establish steady state sediment transport rate and channel morphology. Experiments were allowed to proceed un- til a steady state – defined as a period of several hours in which sediment storage remained constant, determined from figure 2.3 – was achieved. In experiment 1 a pulse of sediment was then released from the upstream end of the reach to simulate a jam failure, while in experiments 2 to 5 wood was added to the reach at low flow (0.4 L/s). The experimental treatments are summarized in Table 2.3. 15 Figure 2.2: Sediment was input at the upstream end at a rate of 55-60 g/min using a rotating feed system. 16 Figure 2.3: Cumulative plots of sediment storage were used to determine steady state. Red lines indicate prolonged periods with constant sediment storage, defined as steady state, which occured prior to wood addition or removal. 17 Table 2.3: A summary of the experimental treatments employed in experi- ments 1 through 5. Experiment Treatment Description Wood Load Prototype Equivalent (m3/m2) x 102 (m3/m2) x 102 1 Sediment pulse release - - 2 Wood addition and removal 0.037 1.1 3 Wood addition and removal 0.053 1.6 4 Wood addition and removal 0.073 2.2 5 Wood addition 0.093 2.8 The wood loads added in experiments 2 to 5 were selected to encompass a broad range of reported values from field studies (table 2.4), while also including an experiment with a load similar to that of the prototype system. Wood addition was designed to mimic natural recruitment processes or ‘soft engineering’ techniques; wood was added to the reach at randomly selected cross sections and angles (either diagonal upstream, downstream, or perpendicular to the bank) and left un-fixed. To promote interaction between pieces and stabilize key members, as well as to emulate natural processes, root wads and branch snags were attached to many of the pieces. Wood was removed from the reach following the attainment of a new steady state in experiments 2 to 4 (figure 2.3).1 Sediment input remained constant throughout the experiments prior to, and following, the treatments. Thus, the experimental reach most closely replicated a treated reach bounded by an unaffected reach upstream. This represents a plausible scenario, as natural disturbances such as fire affect distinct geographic areas with sharp boundaries, and restoration activities are generally implemented at the reach scale. 1Wood was not removed in experiment 5 because the lab was closed for renovations prior to the development of a new steady state. 18 Table 2.4: Naturally occuring wood loads determined from field surveys. Reference Location Sample Size Wb (m) Wood Load (m2/m3) x 102 Cadol et al. (2009) La Selva, Costa Rica 30 3.1-15 0.41-6.1 Andreoli et al. (2008) Chilean Andes 1 7.8 6.6-7.1 Gurnell et al. (2002) USA, UK, Italy, France 7 1.2-890 <0.10-8.1 Bragg et al. (2000) Wyoming, USA 4 9.2-14 0.40-4.5 Richmond and Fausch (1995) Colorado, USA 11 3.7-10 0.92-1.9 Wohl and Jaeger (2009) Colorado, USA 12 3.5-14 1.0-43 Nakamura and Swanson (1993) Oregon, USA 5 7.6-11 0-5.7 Unpublished data** BC, Canada 7 3.2-10 0.080-16 Andrews (2010) BC, Canada (Fishtrap Creek) 1 12 2.1 **BC Ministry of Forests, Lands and Natural Resource Operations 19 Chapter 3 Large Wood Dynamics 3.1 Introduction The quantity and distribution of wood in a stream channel is a function of re- cruitment processes, as well as subsequent entrainment, transport, and deposition (Benda and Sias, 2003; Hassan et al., 2005; Czarnomski et al., 2008; Wohl and Jaeger, 2009). Once recruited to a reach, the entrainment of an individual wood piece depends on the balance between drag forces and resisting forces, which dic- tate piece stability; piece entrainment occurs when the water depth is sufficiently large, relative to piece diameter, to create instability (Braudrick et al., 1997; Brau- drick and Grant, 2001; Bocchiola et al., 2006; Manners and Doyle, 2008). The threshold ratio of piece diameter to water depth, in turn, depends on piece shape, size, and orientation (Braudrick et al., 1997; Braudrick and Grant, 2000; Bocchiola et al., 2008). The size of wood pieces relative to channel width is such a fundamen- tal determinant of wood dynamics that the ratio of piece length to bankfull width is often used to define channel size (Church, 1992; Gurnell et al., 2002; Piegay, 2003). Following entrainment, wood transport and deposition act to maximize piece stability. On average, wood is re-deposited in positions which minimize drag force by aligning pieces with the dominant flow direction (Gippel et al., 1996; Braudrick and Grant, 2001; Webb and Erskine, 2003; Bocchiola et al., 2006). By changing piece orientation, the entrainment and deposition of large wood thereby minimizes 20 the hydraulic impact of wood over time (Gippel et al., 1992; Gippel et al., 1996). The greatest changes in the position of wood usually occur during the first large flow event following recruitment (Sweka and Hartman, 2006). In intermediate-sized streams, large wood often accumulates into wood jams, which are defined here as accumulations of at least three wood pieces with at least two points of contact (Manners et al., 2007; Czarnomski et al., 2008). Jam for- mation complicates the simple processes of wood entrainment and deposition pre- sented above, and may amplify the hydraulic and morphologic effects of wood pieces (Nakamura and Swanson, 1993). Because the effects of multiple wood pieces are not simply additive, piece-scale dynamics cannot be easily scaled up. While general models of jam formation and evolution have been proposed (e.g. Manners et al., 2007; Manners and Doyle, 2008; Bocchiola et al., 2008), the ef- fects of jam formation on channel morphology remain poorly understood. The purpose of this chapter is to investigate the dynamics of large wood in- troduced to the model experimental reach during four separate experiments. Wood entrainment and transport will first be considered at both the piece- and reach-scale. Jam characteristics will then be presented, as well as the evolution of these charac- teristics through time. To elucidate the importance of jam presence, the orientation and hydraulic impact of wood pieces will then be differentiated between individual pieces and jam members. Finally, these results will be used to determine the wood metrics which best represent the influence of large wood on channel dynamics. These proposed metrics will be used throughout the following chapters to deter- mine their relative impact on channel hydraulics and associated bed morphology. 3.2 Methods 3.2.1 Large Wood and Jam Dynamics Wood was surveyed throughout the reach following each run. Pieces were labeled with an identification number based on the cross section at which they were added, and the piece ID was used to track subsequent piece movement and orientation. Each survey included the cross section at which each piece was located, its length and piece type, its orientation, and its position relative to other pieces. Pieces were 21 defined as ‘suspended,’ if any of the piece rested on the edge of either bank. From this information it was possible to determine changes in location, orientation, and jam characteristics following each run. To standardize the wood sizes, wood length was defined in terms of its size relative to the bankfull channel width, such that: Lp∗= LpWb (3.1) Jams were defined as locations with at least three pieces, with at least two points of contact, and were noted in the wood surveys. Jam types were then de- fined using overhead photographs taken at the end of each run. Observations on piece movement, as well as jam formation and failure, were deduced using seperate overhead photographs taken at 4 second intervals throughout each run. 3.2.2 Hydraulic Effectiveness The hydraulic effects of large wood are dependent on both piece size and orien- tation (Gippel et al., 1992; Webb and Erskine, 2003). Hydraulic effectiveness is quantified by calculating a blockage ratio, which is the dimensionless ratio of the projected area of the wood piece to the area of the flow at that point. The projected area (Api) of each wood piece was calculated according to: Api = L ·D · sin(σ) ·P (3.2) where L and D are the piece length and diameter in metres, σ is the piece angle relative to the flow direction, and P is the proportion of wood submerged by the flow (Webb and Erskine, 2003). As flow depth was not directly measured, the flow area (Af ) was calculated using: Af = Q vhm (3.3) where Q is the discharge and vhm is the harmonic mean flow velocity – measured using a slug injection of a saline solution coupled with a conductivity probe – with wood in the channel. A blockage ratio (Bp) was finally calculated for each piece according to: 22 Bp = Ap Af (3.4) Due to variations in piece number and wood load between experiments, the blockage ratio was aggregated throughout the reach to provide a useful metric of reach-scale hydraulic impact for each experiment. Because flow area (Af ) and reach length were constant for all four experiments, the hydraulic impact of large wood could be compared using the total projected area (Ap) of the wood pieces for each experiment, calculated as: Ap =∑Api (3.5) To standardize these results the projected area was then non-dimensionalized by dividing by the bed area: Ap = Ap L ·Wb (3.6) where L is the length of the reach, and Wb is the bankful width of the reach in metres. 3.3 Results 3.3.1 Distance and Timing of Wood Transport Initial piece stability at the time of recruitment is related to piece density, length and diameter, rootwad presence and diameter, and piece orientation (Braudrick and Grant, 2001). It is hypothesized that wood transport will be greater for those pieces with lower initial stability, and will maximize piece stability by changing the piece orientation. While piece diameter, rootwad diameter and density were held constant throughout the experiments, piece length and rootwad presence varied according to piece type. The angle of entry also varied, as pieces were recruited either perpendicular (90◦) to flow or oblique (45◦) to flow. Surveys of large wood, conducted following each run in the four experiments in which wood was added to the reach, were used to test these hypotheses. A total travel distance for each piece was determined based on its original and final posi- 23 Figure 3.1: The effects of piece length, rootwad presence, and recruitment angle on travel velocity are shown. Each difference is significant at α = 0.05. 24 tion in the reach. Because wood sometimes re-mobilized following stabilization, travel distances were also divided by the duration of the experiment to yield a travel rate for each piece (m/hr). The results presented in figure 3.1, show that the travel rate of individual pieces was significantly influenced by piece type, with each decrease in piece size associ- ated with an increase in the travel distance per unit time. The orientation of entry also affected travel distance, with pieces recruited perpendicular to the flow mov- ing at a rate 2.4 times that of the pieces recruited at an oblique angle to the flow. While all of the relationships shown in figure 3.1 were significant (α = 0.05), the presence or absence of a rootwad exerted the greatest influence on travel distance and rate. On average, pieces lacking rootwads travelled 6.8 times the distance of those with rootwads, resulting in a travel rate that was 6.9 times higher. The orientation of wood pieces also changed significantly during the experi- ments: the percentage of all wood pieces oriented perpendicular to the flow de- creased from 26% at the time of recruitment to 13% at equilibrium, while the per- centage of pieces oriented obliquely (45◦) to the flow decreased from 74% upon entry to 53% at equilibrium. Conversely, the percentage of pieces oriented parallel to the flow increased from 0% at the time of recruitment to 35% at equilibrium (figure 3.2). Aggregate results for the experimental reach were derived by averaging indi- vidual results for each experiment. The mean distance travelled by a piece of wood, regardless of the piece type and orientation, ranged from 0.51 m (1.5Wb) in exper- iment 2 to 1.2 m (3.6Wb) in experiment 4, with an average travel distance of 0.73 m. This is equivalent to travel distances of 15 m to 37 m in the prototype system, with an average travel distance of 22 m. The associated travel rates varied from 0.012 m/hr in experiment 5 to 0.027 m/hr in experiment 4, with an average travel rate of 0.018 m/hr, equivalent to 0.1 m/hr in the prototype system. However, none of the experiments differed signif- icantly in either mean travel distance, or travel rate (α = 0.05). At a significance level of α = 0.1, only the differences in the travel rates of experiments 4 and 5 differed significantly (p = 0.078). Average wood transport distance and travel rate were not significantly related to wood load or jam characteristics. A large proportion of wood movement, ranging from 43-66% of the total trans- 25 Figure 3.2: Wood orientation at recruitment and equilibrium. port, occurred in the first run following the addition of wood to the reach, while periodic increases in downstreammovement were observed later in the experiments (figure 3.3). Pieces without rootwads appeared to exert little influence on the channel bed during transport. Following scouring and de-stabilization, these pieces usually rolled until they reached the thalweg, where water depth was then sufficient to float the piece, which then became oriented parallel to the flow direction. Pieces with rootwads, however, exerted a strong influence on channel bed, as the rootwad maintained contact with the bed throughout wood transport producing bed scour. Downstream movement of these pieces involved two steps: the piece first pivoted such that the rootwad was oriented upstream and then moved downstream parallel to the flow direction with the rootwad scouring and mobilizing sediment. The ma- jority of individual pieces with rootwads experienced at least the first stage of this process, as 76% of individual pieces with rootwads were oriented with the rootwad upstream at equilibrium (figures 3.4 and 3.5). 26 Figure 3.3: Wood transport distance over time, as a percentage of total trans- port, is shown for each of the four experiments in which wood was added to the reach. 3.3.2 Jam Frequency and Size A total of 11 jams had developed throughout experiments 2 to 5 at steady state. 9 of the 11 jams contained at least one suspended 1.2Wb (0.4 m) piece which had re- mained in situ throughout the experiment, while the remaining two jams contained at least one 1.2WB piece. Defining these large pieces as ‘key members,’ each jam was classified using the classification system proposed by Abbe and Montgomery (2003). Owing to the stability of the suspended 1.2Wb pieces (only 1 of 15 moved during the four experiments), nine of the jams were broadly classified as combina- tion jams, which contain in situ key members with additional racked large wood. The remaining two jams, which did not contain suspended pieces, were broadly defined as transport jams, as the key member had been transported some distance downstream prior to jam formation. The specific jam classifications and frequen- cies are shown in Table 3.1. The number of jams in each reach at steady state varied from 0 jams in exper- iment 2, to 6 jams in experiment 5. Jam size averaged 5.9 pieces/jam and 0.0027 27 Figure 3.4: The orientation of pieces with rootwads at steady state is com- pared for jam members and individual pieces. Figure 3.5: Pieces with rootwads often oriented parallel to the flow direction, with the rootwad upstream. m3/jam and did not differ significantly between the 3 experiments in which jams formed (experiments 3 to 5). The strong linear relationship between the average jam volume and piece number (figure 3.10; table 3.2) shows that the distribution of piece sizes in the jams was consistent throughout the experiments. 28 Table 3.1: Jam classification (after Abbe and Montgomery (2003) Classification Jam type Definition Frequency (%) Example Combination Jam Flow deflection Key member may be rotated and deflects flow. 64 Figure 3.6 Valley Jamwidth is equal to the channel width and influences the channel bottom. 18 Figure 3.7 Transport Jam Unstable Unstable accumulations of racked LW upon bar top. 9 Figure 3.8 Bar apex One or more key member down- stream; often associated with de- velopment of bar. 9 Figure 3.9 29 Figure 3.6: Flow deflection jam. Figure 3.7: Valley Jam. 30 Figure 3.8: Unstable jam. Figure 3.9: Bar apex jam. 31 Table 3.2: All relationships significant at α = 0.1 are described. Independent Variable Dependent Variable Coefficient Intercept R2 p(1) p(2) Jam size (pieces) Jam size (m3) 0.0000070 0.000040 1.0 0.64 0.00026 Suspended pieces (m−2) Jam frequency (W−1b ) 0.097 -0.0092 0.90 0.90 0.053 % pieces in jams Mean blockage ratio 0.0012 0.13 0.92 0.016 0.040 Wood load (m3/m2) Projected area 40 -0.0058 0.93 0.37 0.034 Piece frequency (m−2) Projected area 0.0016 -0.0064 0.93 0.36 0.037 Jam frequency(W−1b ) Projected area 0.049 0.0093 0.86 0.13 0.073 32 Figure 3.10: The volume of pieces in jams exhibits a strong linear relation- ship to the number of pieces in jams for the four experiments in which wood was added to the reach. Paired analyses of the wood characteristics shown in Table 3.3 were used in- vestigate jam characteristics at steady state. While wood load exhibited a positive linear relationship to jam frequency (p = 0.14), there was a stronger linear relation- ship between the number of suspended pieces and the jam frequency (figure 3.11; table 3.2). The percentage of pieces in jams, which is a measure of the trapping efficiency of the jams as well as the mobility of individual pieces, was positively related to the number of jams in the reach, though the relationship was not statisti- cally significant (p = 0.13). Jam evolution throughout the experiments primarily involved the accumulation of mobilized individual pieces. Changes in jam size, as well as the percentage of pieces in jams, are shown in Table 3.4. Increases in jam size and the percentage of pieces in these jams varied from 0-25% in the four experiments. 33 Table 3.3: Reach-scale wood characteristics for experiments 2 to 5 at steady state. Experiment Piece Wood Number of Jam frequency Pieces in Frequency Load Suspended (W−1b ) Jams (%) (m−2) (m3/m2) Pieces 2 9.7 0.00037 0 0 0 3 14 0.00053 3 0.23 86 4 19 0.00073 4 0.15 50 5 24 0.00093 7 0.38 87 Figure 3.11: Jam frequency is linearly related to the number of suspended pieces (per unit of channel bed) within the reach. 34 Table 3.4: Changes in jam frequency and size from the initial time of wood addition to the attainment of steady state are shown for experiments 2 to 5. Experiment Original Jam Final Jam Original Final Pieces Original Final Mean Frequency Frequency Pieces in in Jam (%) Mean Jam Jam Size (W−1b ) (W −1 b ) Jam (%) Size (pieces) (pieces) 2 0 0 0 0 0 0 3 0.23 0.23 71 86 5 6 4 0.15 0.15 40 50 6 7.5 5 0.38 0.46 70 87 5.2 5.3 35 Movement of entire jams occured less frequently than movement of individ- ual pieces, as it involved the de-stabilization and entrainment of key members. Throughout the experiments, however, there were multiple episodes of jam failure and formation. Pieces from both of the jams that failed, one in experiment 4 and one in experiment 5, later joined downstream jams. Neither of these jams con- tained suspended pieces, highlighting again the role of large stable pieces in jam stabilization. The jams which formed during the experiments, one in experiment 4 and two in experiment 5, were small, containing between three and five members. Overall, jam frequency remained relatively constant throughout the experiments, changing only in experiment 5. 3.3.3 Hydraulic Effectiveness The combined results from all four experiments yield an average blockage ratio of 0.20, with 59% of all pieces having a blockage ratio greater than 0.1, which is considered the threshold of hydraulic effectiveness (Gippel et al., 1992). The mean blockage ratio varied significantly (α = 0.05) between individual pieces and those located in jams (figure 3.12). The mean blockage ratio for jam members was 0.27, while that of individual pieces was 0.094. On average, individual pieces were at or slightly below the threshold for hydraulic effectiveness proposed by Gippel et al. (1992), while jam members far exceeded it. This discrepancy is largely attributable to differences in piece orientation; while individual pieces were mostly oriented parallel to the flow at equilibrium (63%), the majority of jam members were oriented either obliquely (63%) or perpendicular (18%) to the flow (figure 3.13). Despite considerable differences in jam characteristics, there was no signifi- cant difference in the average piece blockage ratio between the four experiments at steady state. Although the mean blockage ratio (Bp) varied from 0.12 in experiment 2 to 0.23 in experiment 3, the difference between experiments was not significant (α = 0.05). Paired analyses reveal that these differences in average blockage ratio were not significantly related to the reach-scale wood load (p = 0.35), but were sig- nificantly related to the percentage of pieces in jams (figure 3.14; table 3.2). Mean blockage ratio was not significantly related to jam frequency (p = 0.26). 36 Figure 3.12: Blockage ratio is compared for individual pieces and jam mem- bers. Figure 3.13: Wood orientation for individual pieces and those in jams is com- pared. 37 Figure 3.14: The mean blockage ratio is linearly related to the percentage of all pieces in jams. The total projected area of all wood pieces at steady state varied from 0.010 m2 in experiment 2, to 0.048 m2 in experiment 5, while the dimensionless projected area varied from 0.0065 to 0.031. The dimensionless projected area was signifi- cantly linearly related to wood load and piece frequency, and nearly significantly related to jam frequency (figure 3.15; table 3.2). 3.4 Discussion Three primary factors – piece size, shape, and orientation – dictated the transport of individual wood pieces in the four experiments described herein. The importance of piece length has been previously shown in numerous field studies which have suggested that piece size controls the ability of wood to become jammed within the channel (Gurnell et al., 2002; Sweka and Hartman, 2006). The results also lend additional support to the theoretical models of Braudrick et al. (1997) and Braudrick and Grant (2001), which state that entrainment is a function of a channels debris roughness, which depends on the relative diameter of the a (Dlog/dw) as well as its relative length (Lc/Wb). While the effects of variations 38 Figure 3.15: The dimensionless projected area is linearly related to both the wood load and jam frequency. in diameter were not tested in these experiments, the results support the notion that the critical roughness (Dlog/dw) for entrainment varies with rootwad presence and piece angle. Furthermore, that travel distance did not vary significantly between experiments suggests that these piece-scale wood characteristics dictate entrain- ment and transport, rather than reach-scale characteristics such as wood load or 39 jam frequency. The results also support the hypothesis that individual pieces, on average, re- deposit in positions of greater stability, such that stability increases over time; over 95% of individual pieces were oriented either parallel or oblique to the flow direc- tion at steady state. Between 43-66% of this movement and re-positioning occurred during the first run following recruitment. This pattern of wood movement, in which the greatest transport and orientation change occurs immediately following recruitment, has also been observed in field studies (Sweka and Hartman, 2006), and supports previous research suggesting that most un-fixed pieces remain stable following wood installation. Periodic increases in transport throughout the experi- ments highlight the stochastic nature of wood transport. A total of 11 jams had formed at steady state in experiments 2 to 5. The forma- tion of these jams was facilitated by the stabilization of a ‘key member,’ which was most often a suspended 1.2Wb piece; at steady state 9 of the 11 jams contained a suspended piece. These results suggest that the 1.2Wb pieces, and more specifically those stabilized by suspension on a bank, act as ‘key members,’ pieces defined by Manners and Doyle (2008) as a “relatively large tree trunk that may or may not contain branches and a rootwad which serves to accumulate additional pieces of wood.” Further, as all 1.2Wb pieces contained both branches and a root wad, these results emphasize the importance of such natural irregularities in piece stabilization and jam formation. Jam evolution largely followed the model proposed by Manners and Doyle (2008), and can be described according to the following three stages: 1. Key member recruitment: a key suspended member was placed in the chan- nel, or alternately an in-channel 1.2Wb piece became stabilized due to suffi- ciently high Llog/Wb or Dlog/dw to prevent entrainment. 2. Accumulation: pieces mobilized upstream joined the jam, increasing the jam size by up to 25% after the first major flow event. 3. De-stabilization and fragmentation: scour at the jam site increased the wa- ter depth (dw) around the key member such that Dlog/dw decreased below the threshold for entrainment. Generally the released jam members became 40 lodged in the nearest downstream jam. The similarities between the wood dynamics in the experiments and the models of jam formation described in the literature lend support to the realism of the physical model. The broad range in the percentage of pieces that accumulated in jams, which varied from 0-87%, is supported by previous studies: Dahlstrom and Nilsson (2004) reported that 0-59% of wood pieces were in jams, while Marsh et al. (2001) found that 60-85% of wood pieces were located in jams. Jam spacing varied from approximately 2.2 to 6.5 times the channel width at steady state. Thus, jam spacing was generally greater than the spacing of two to three channel widths proposed by Petts and Foster (1985), but less than the spacing of 7 times channel width reported by Assani and Petit (1995) for a drainage canal in France, and the spacing of 6.7 times the channel width reported for the prototype stream, Fishtrap Creek (Andrews, 2010). The jam spacing was similar to the absolute frequency of 2.4 jams/100 m found by Kreutzweiser et al. (2005) in 16 Boreal streams ranging from 3-7 m in channel width, though jam spacing was low compared with these results when considered in terms of channel width (spacing of 6-14 times channel width). Jam spacing is difficult to compare between systems, however, as it depends on forest type, which dictates the density and mortality rate of the riparian vegetation, as well as the decay rate, size and shape of the wood pieces. Thus, jam spacing may differ signficantly in streams with similar ratios of tree height to channel width. Jam formation significantly altered the orientation and hydraulic impact of wood pieces. As hypothesized, jams appear to lend stability to wood pieces, al- lowing pieces to deposit at orientations which would otherwise enable entrain- ment. Much like rootwads, jams increase the critical roughness (Dlog/dw) needed to entrain pieces. As a result, the average blockage ratio of jam members was sig- nificantly higher than that of individual pieces, and nearly 20% of jam members were oriented perpendicular to the flow direction at equilibrium. The average blockage ratio for all pieces, which was above the threshold for hydraulic effectiveness (Bp = 0.1) proposed by Gippel et al. (1992) for all four ex- periments, is within the range of blockage ratios computed in previous studies. In a 41 study of 512 wood structures, Kail et al. (2007) reported that 66% of the structures were above the threshold of hydraulic significance, with nearly 50% of structures having blockage ratios between 0.1 and 0.3. Using data from the Tonghi River in Australia, however, Webb and Erskine (2003) calculated an average blockage of 0.04, with only 10.4% of pieces above 0.1. This lower average blockage ratio may result from smaller average piece size, as the average wood piece was approxi- mately 2 m shorter and 0.1 m smaller in diameter than the average (scaled) wood piece in the present flume study. As expected, the blockage ratio was lower in experiment 2 where no jams de- veloped. The difference, however, was only significant with α = 0.1. Of the factors considered here, including the percentage of pieces in jams, jam frequency, and wood load, only the percentage of pieces in jams was significantly related to the average blockage ratio. Also, the total projected area was closely related to the wood load, and showed little deviance from this linear relationship despite signifi- cant differences in jam frequency between the experiments. Given the influence of jams on the mean piece blockage ratio, the lack of significant correlations between jam frequency and the average blockage ratio is likely a result of the small sample size. Further experimentation is needed to expand the sample size and determine whether a relationship actually exists. The four metrics developed throughout this chapter, which include the mean blockage ratio, total projected area, jam frequency, and percentage of pieces in jams, will be used throughout the remaining chapters, along with wood load and piece frequency, to assess the effects of large wood on channel hydraulics and bed morphology. 42 Chapter 4 Stream Hydraulics and Sediment Transport 4.1 Introduction Large wood dissipates a substantial proportion of the potential energy available to many forested streams (Heede, 1972; Keller and Swanson, 1979; Keller and Tally, 1979; Bilby and Likens, 1980; Thompson, 1995; Faustini and Jones, 2003; An- dreoli et al., 2007). If wood is considered in the assignment of the characteristic grain size or roughness length, the addition of wood decreases conveyance, reduc- ing the mean flow velocity for a given flow depth. As a result, channels with large wood may experience reduced sediment transport and bank erosion (Heede, 1972; Smith et al., 1993; Gippel, 1995; Lisle, 1995). Decreased conveyance also increases flow depth for a given discharge. Given the law of continuity, changes in velocity alone are expected to increase flow depth, as: v0 ·d0 ·w= v1 ·d1 ·w (4.1) where d and w are the flow depth in m. Thus, increases in wood load generally increase water stage, and wood has often been removed from channels as a flood remediation measure (Bisson et al., 1987; Young, 1991; Gippel, 1995). Stage 43 changes are further amplified by increases in water stage due to displacement of water by the wood itself, as well as bed aggradation from local decreases in sedi- ment transport. The distribution of stage changes influence the water surface slope, which de- termines the potential energy in the system. Given that water surface slope and flow depth both directly determine the reach-averaged shear stress, changes in these pa- rameters following the addition or removal of wood may dramatically influence the average shear stress in a reach. While total shear stress is positively related to wa- ter depth, and therefore expected to increase with wood load, the magnitude of the shear stress acting on the bed has generally been shown to decrease following the addition of wood to a channel, and increase following its removal (Heede, 1972; Beschta, 1979; Mosley, 1981; Smith et al., 1993; Lisle, 1995). The purpose of this chapter is to determine the effects of wood addition and re- moval on channel hydraulics. It is hypothesized that the addition of large wood will decrease flow velocity and the proportion of total shear stress partitioned to grain shear stress. It is also hypothesized that the addition of wood will increase the wa- ter surface elevation, the total shear stress, and the channel roughness. The removal of wood is expected increase velocity and grain shear stress, while decreasing the water surface elevation, total shear stress, and roughness. To determine the effects of wood on channel hydraulics, changes in flow ve- locity will first be considered, followed by an investigation of the effects of wood addition and removal on water stage and water surface slope. The effects of each treatment on the total shear stress, channel roughness, and shear stress partitioning will then be considered. Each of the changes in these hydraulic parameters will be compared with the wood metrics introduced in chapter 3 to determine the relative influence of each wood-related metric on channel hydraulics. 4.2 Methods 4.2.1 Flow Velocity Dilution techniques are often used to measure stream flow in small, mountainous streams where the turbulence associated with individual roughness elements and 44 the small flow depths present in low-flow conditions preclude the use of current meters (Day, 1977; Moore, 2003). Salt is commonly used as a tracer as it is inex- pensive, non-toxic, and can be measured using a conductivity meter (Moore, 2003). For this research, velocity measurements were obtained using a slug injection tech- nique. Throughout each experimental run a saline solution (5mg/L) was injected at approximately two minute intervals into the flow by a tipping bucket located at the upstream end of the flume. The bucket position (a tip registered as a value of 0.007), electrical conductivity (EC), and time were recorded at 0.1 second intervals in the program LabVIEW. Velocity measurements were obtained from the temporal profile of the con- ductivity data by calculating the travel-time of the saline pulse along the known distance (5.5 m) from the injection site to the downstream conductivity meter. Be- cause the pulses dispersed as they traveled downstream, with faster velocities in the leading segment of the pulse and lower velocities in the trailing end, it was first necessary to identify the centre of each pulse. Although the centroid of mass is commonly used, Waldon (2004) and Zimmerman (2009) suggest that the cen- troid provides an under-estimation of velocity, and instead propose the use of the harmonic mean velocity. For this study, the velocity measurements associated with three different anal- ysis methods were compared: the peak, the harmonic mean, and the centroid. The velocity associated with the peak (vp) of the saline pulse is simply: vp = x tp (4.2) where x is the travel distance between the tipping bucket and the conductivity probe and tp is the time between the slug injection and the peak concentration associated with the pulse. Harmonic mean velocity and the centroid velocity both required the calculation of the probability density function, px(t), and the time integral of concentration, Ix, at the fixed point x: px(t) = c(x, t) Ix (4.3) 45 Ix= ￿ c(x, t) ·dt (4.4) where x is the location of the downstream conductivity probe and t is the time since the injection (Waldon, 2004). The travel time of both the centroid (tc) and the harmonic mean (thm) were then calculated as: tc = ￿ px(t) ·dt (4.5) thm = 1￿ 1 t (px(t) ·dt) (4.6) and finally the centroid velocity (vc) and the harmonic mean velocity (vhm) were calculated as: vc = x tc (4.7) vhm = x thm (4.8) where x is the distance between the tipping bucket and the conductivity probe (Wal- don, 2004). The relative, or dimensionless, velocity (vr) at steady state following wood addition was also calculated for each experiment using the peak velocity values according to: vr = vp(LW ) vp (4.9) where vp(LW ) is the peak velocity at steady state with wood in the channel, and vp represents the peak velocity prior to wood addition. To obtain velocity measurements for each experimental run the conductivity data was first filtered to eliminate high frequency noise. This was accomplished using a 10 point binomial filter in the program R, which eliminated random spikes in conductivity while preserving the peak conductivity values. Both filtered and un-filtered data was then processed in MATLAB to determine 46 Figure 4.1: The filtered electrical conductivity (blue) is shown, as well as the threshold conductivity values and the polynomial threshold line (green). reach-averaged flow velocity. Threshold conductivity (ECtsd) at each injection time was first defined as: ECtsd = ECavg+6σ (4.10) where ECavg is the mean conductivity for the 200 un-filtered measurements pre- ceding an injection, and σ is the standard deviation. A 2nd order polynomial was fit to these points and used to define the threshold conductivity at any time, t (figure 4.1). The value of 6σ was chosen following sensitivity analysis, which showed that this value minimized the variance in the centroid velocity (vc) by optimally ex- cluding the integrated concentration (Ix) associated with secondary pulses, which were thought to be related to slower seepage of saline water through the styrofoam structure. Further increasing the threshold standard deviation did not produce ad- ditional reduction in variance. In several cases, however, a higher threshold was required to eliminate secondary pulses. In these cases, the minimum standard de- 47 viation needed to reduce the variance in the centroid velocity was used. Finally, the filtered conductivity values were normalized using this threshold conductivity (ECnorm = ECavg−ECtsd), allowing the subsequent definition of pulses as those segments of the temporal profile where ECnorm ≥ 0. 4.2.2 Water Stage Changes in stage were determined using the water surface elevations measured following each experimental flume run. These values – measured as the distance from bank top to water surface at 0.25 m intervals – were subtracted from the bank height elevation at each measurement point to determine the water surface elevation above an arbitrary datum using MATLAB. The change in stage at each measure- ment point was computed as the difference between the water surface elevation at equilibrium without wood, and at steady state with wood in the channel. The average stage change (D) attributable to the addition of a volume of large wood, which displaces an equal volume of water, was also calculated for each experiment according to: δD= Vw Wb ·L (4.11) where Vw is the volume of wood added to the channel in m3 andWb and L are the width and length of the channel in m. Water surface slope was also calculated using the water surface elevation data. Mean slope, S, was determined according to: S= δE L (4.12) where E is the water surface elevation in metres, and L the distance over which wa- ter surface elevation was measured. Changes in slope were derived by comparing the mean slope of the water surface with wood in the channel to the mean slope prior to the addition of wood. 48 4.2.3 Shear Stress Partitioning Alterations in flow depth and velocity, as well as the water surface slope, affect shear stress in the channel. To determine the magnitude of these effects, the mean hydraulic depth (d) was first calculated in all five experiments for all steady state runs (i.e. prior to wood addition, following wood addition, and after wood removal) using the harmonic mean velocity. From flow depth, the wetted perimeter (P) and hydraulic radius (R) were subsequently determined, according to: d = Q vhm ·Wb (4.13) P=Wb+2d (4.14) R= Q P (4.15) These values, as well as the water surface slope calculated in the previous sec- tion, were then used to calculate the total reach-averaged shear stress (τ), as well as the shear stress partitioning between grain shear stress (τg) and bedform shear stress (τb) (Assani and Petit, 1995; Manga and Kirchner, 2000): τ = ρ ·g ·R ·S (4.16) τ = τb+ τg (4.17) The roughness due to particles (ng) and the total Manning’s roughness (n) were then calculated as: ng = 0.048 · (D50)1/6 (4.18) n= 1 vhm ·R2/3 ·S2/3 (4.19) where D50 is the median grain size of the surface material, and used to determine the grain shear stress. 49 τg = τ · ￿ng n ￿3/2 (4.20) The D50 used throughout this chapter was determined from a weighted average of surface grain sizes calculated from relative facies areas according to the methods described in the following chapter. 4.3 Results 4.3.1 Flow Velocity Estimated mean velocities for each of the five experiments are shown in Table 4.1. Similar to the results found byWaldon (2004) and Zimmerman (2009), the velocity associated with the peak of the saline pulse was significantly higher (α = 0.05) than that associated with the harmonic mean and temporal centroid. Averaging the data for all wood loads, the mean velocity associated with the peak (0.32 m/s) was 7% higher than the median harmonic mean velocity (0.30 m/s) and 10% higher than the median centroid velocity (0.29 m/s) (figure 4.2). The peak velocity was associated with the least variance and had the fewest outliers, while the centroid- derived velocity was the most sensitive to the value of the integrated concentration, Ix, and was therefore under-estimated wherever the threshold polynomial failed to exclude a secondary pulse. By comparing back-calculated depth with measured depth, Zimmerman (2009) found that the harmonic mean velocity was the most accurate measure of the three. Given that such calculations were not performed during this research, the peak velocity was primarily used in comparisons between different wood loads, since estimates of the peak were associated with the lowest uncertainty. Harmonic mean velocity was used to determine flow depths, which were used to calculate the flow area in chapter 3. The presence of large wood significantly reduced stream velocity in experi- ments 2 to 5, regardless of the volume of wood added or the method used (figure 4.3). The decrease in velocity measured following the addition of only 0.00037 m3/m2 (equivalent to 0.01 m3/m2 in the prototype) of wood was statistically sig- nificant for all three velocity calculation methods. 50 Table 4.1: A comparison of peak, harmonic mean and centroid velocity from the five experiments. Wood load (m3/m2) Parameter Units 0 0.00037 0.00053 0.00073 0.00093 Time Temporal peak s 15.9 17.2 17.5 18.5 18.4 Harmonic mean s 16.8 18.6 18.7 19.7 19.7 Temporal centroid s 17.3 19.2 19.2 20.5 20.3 Velocity Temporal peak m/s 0.346 0.320 0.314 0.297 0.299 Harmonic mean m/s 0.327 0.296 0.295 0.279 0.279 Temporal centroid m/s 0.318 0.286 0.286 0.268 0.272 The effects of wood addition were generally immediate, and flow velocities did not significantly change during the period following the initial wood input in the majority of the experiments. In experiment 3, however, the decrease in velocity was more gradual, with velocity consistently decreasing throughout the first 15 hours following the addition of wood. The peak-derived channel velocity varied significantly prior to wood addition between all five experiments, ranging from 0.34 m/s in experiment 3, to 0.36 m/s in experiment 2. Changes in velocity relative to this original value were therefore also considered. Peak velocity decreased by 15% in experiment 4, and by 16% in ex- periment 5. Thus, while the mean velocity following wood addition was higher in experiment 5 than experiment 4, the relative velocity decrease was actually greater in experiment 5. The reverse was true, however, in experiments 2 and 3: while the mean velocity with wood was greater in experiment 2 than experiment 3, velocity decreased by a greater margin in experiment 2 (13%) than in experiment 3 (11%). Following the removal of large wood in experiments 2 and 3, the peak flow velocity increased, but remained significantly lower (α = 0.05) than its pre-addition value. While velocity increased by 7.2% following the removal of 0.00037 m3/m2 in experiment 2, it remained 4.8% below its initial value prior to the addition of wood. Likewise, peak flow velocity increased by 8.7% following the removal of 0.00053 m3/m2 of wood in experiment 3, but remained 2.2% below its original 51 Figure 4.2: Comparison of velocity estimates derived using the peak, har- monic mean, and the centroid of the pulse in the temporal conductivity profile. Red plus signs represent outliers. value. The magnitude of the changes of both the relative and absolute velocities fol- lowing the addition of large wood were related to several of the metrics introduced in chapter 3. The mean velocity at steady state with large wood in the channel was significantly related to wood load and piece frequency (figure 4.4), as well as the total projected area, mean blockage ratio, and percentage of pieces in jams (figure 4.5; table 5.3). The change in velocity following wood addition was significantly related to wood load, piece frequency, and the mean blockage ratio (table 5.3). For all metrics, the significance of the relationships was greater for the absolute peak velocity than for the relative peak velocity. 52 Figure 4.3: Comparison of the effects of wood addition on peak velocity. The upper boxplot shows the velocity change associated with the presence or absence of wood, while the lower boxplot shows the magnitude of the peak velocity associated with varying wood loads. 53 Figure 4.4: The mean peak-derived velocity was negatively related to both wood load and piece frequency. 54 Figure 4.5: The mean peak-derived velocity following the addition of large wood is compared with the mean piece blockage ratio and the total pro- jected area. 55 Table 4.2: All significant (α = 0.1) relationships between wood metrics and flow velocity (both absolute and relative) are described. Independent Variable Dependent Variable Coefficient Intercept R2 p(1) x 10−5 p(2) Wood load (m3/m2) Velocity (m/s) -53 0.34 0.86 2.2 0.022 Wood load (m3/m2) Relative velocity -140 0.98 0.82 3.0 0.035 Piece frequency (m−2) Velocity (m/s) -0.0021 0.34 0.87 2.1 0.021 Piece frequency (m−2) Relative velocity -0.0055 0.98 0.83 2.8 0.032 % pieces in jams Velocity (m/s) -0.00038 0.33 0.66 5.0 0.096 Mean blockage ratio Velocity (m/s) -0.20 0.34 0.97 0.31 0.0025 Mean blockage ratio Relative velocity -0.50 0.98 0.77 2.8 0.051 Projected area Velocity (m/s) -1.5 0.34 0.82 2.6 0.033 56 4.3.2 Water Stage The addition of large wood increased the water surface elevation in experiments 2 to 5 (figure 4.6). The magnitude of the increase varied from 2.1 mm in experiment 2 to 9.1 mm in experiment 5, and amounted to an increase of 1.4% to 5.9% (ta- ble 4.3). Increased water stage was attributable to aggradation (increased sediment stage), as well as increased water depth related to decreased flow velocity and the displacement of water by wood. The increase in depth attributable to the displace- ment of water by the wood itself varied from 0.4 to 0.9 mm, and represented 8.5% to 17% of the depth increase. The increase in stage was not uniform throughout the channel; the magnitude of the change in surface elevation decreased downstream with net stage decrease occuring at the cross section located 0.25 m upstream of the outlet. As a result, the mean water surface slope increased following the addition of large wood in all four experiments (table 4.3). The magnitude of the increase varied from 0.8% in experiment 2, to 29% in experiment 5. Table 4.3: A comparison of the effects of wood addition on water surface elevation and water surface gradient in experiments 1 to 5. Experiment Parameter Units 1 2 3 4 5 Stage Change mm 0 2.1 6.1 8.3 9.1 % 0 1.4 4.0 5.4 5.9 Displacement mm 0 0.4 0.5 0.7 0.9 % 0 17 8.5 8.9 10 Slope m/m 0.014 0.015 0.016 0.017 0.019 % 0 0.80 6.4 16 29 The magnitude of the average change in water surface elevation for each ex- periment was significantly related (α = 0.05) to several of the reach-scale wood metrics (table 5.5). The total change in water surface elevation, both including and excluding the contribution by water displacement from the added wood, was sig- nificantly related to the wood load and piece frequency (figure 4.7), as well as the mean blockage ratio and the dimensionless projected area (figure 4.8). Increasing α to 0.1, the change in stage was also significantly related to jam frequency and the 57 Figure 4.6: Longitudinal plots of water surface elevation are shown for ex- periments 2 to 5 prior to the addition of wood (red) and following the addition of wood (blue) (vertical exaggeration = 10). percentage of pieces in jams (table 5.5). In most cases the total change in surface elevation, including the contribution from the wood volume, provided a stronger relationship. 58 Table 4.4: All significant (α = 0.1) relationships between wood metrics and water stage are described. The adjusted stage change represents the total stage change excluding the contribution from water displacement by wood. Independent Variable Dependent Variable Coefficient Intercept R2 p(1) p(2) Wood load (m3/m2) Total stage change (mm) 0.00011 -0.35 0.94 0.73 0.063 Wood load (m3/m2) Adjusted stage change (mm) 0.0097 -0.35 0.93 0.73 0.0083 Piece frequency (m−2) Total stage change (mm) 0.42 -0.40 0.94 0.71 0.0065 Piece frequency (m−2) Adjusted stage change (mm) 0.38 -0.39 0.93 0.71 0.086 Jam frequency (W−1b ) Total stage change (mm) 18 2.2 0.72 0.25 0.070 Jam frequency (W−1b ) Adjusted stage change (mm) 16 1.9 0.71 0.26 0.072 % pieces in jams Total stage change (mm) 0.079 1.6 0.75 0.38 0.059 % pieces in jams Adjusted stage change (mm) 0.073 1.4 0.76 0.39 0.055 Mean blockage ratio Total stage change (mm) 37 -0.59 0.83 0.76 0.033 Mean blockage ratio Adjusted stage change (mm) 33 -0.59 0.82 0.74 0.0025 Projected area Total stage change (mm) 310 0.21 0.97 0.75 0.0020 Projected area Adjusted stage change (mm) 280 0.14 0.97 0.83 0.0025 59 Figure 4.7: The change in water stage (dE), measured in mm, is compared to wood load and piece frequency. The change in mean water surface slope, as well as the absolute value of the surface slope with wood in the channel, were also related to wood load and piece frequency (figure 4.9), as well as the jam frequency and dimensionless total pro- jected area (figure 4.10). The metrics consistently explained a greater proportion of the variance of the absolute value of the slope than in the change (% increase) in 60 Figure 4.8: The change in water stage (dE), measured in mm, is compared to both mean blockage ratio and dimensionless projected area. surface slope (table 5.7). Neither the water surface slope nor the change in water surface slope were significantly related to the percentage of pieces in jams or the mean piece blockage ratio. 61 Figure 4.9: The mean water surface slope at steady state following the addi- tion of wood in experiments 2 to 5 was linearly related to the wood load and piece frequency added to the reach. 62 Figure 4.10: The mean water surface slope at steady state following the ad- dition of wood in experiments 2 to 5 was linearly related to the jam frequency and dimensionless projected area. 63 Table 4.5: All significant (α = 0.1) relationships between wood metrics and water surface gradient are described. Independent Variable Dependent Variable Coefficient Intercept R2 p(1) x 104 p(2) Wood load (m3/m2) Gradient (m/m) 5.5 0.013 0.95 0.80 0.0047 Wood load (m3/m2) Relative gradient 300 0.95 0.82 3.1 0.034 Piece frequency (m−2) Gradient (m/m) 0.00021 0.013 0.94 0.97 0.0056 Piece frequency (m−2) Relative gradient 0.012 0.95 0.81 3.4 0.038 Jam frequency (W−1b ) Gradient (m/m) 0.0095 0.014 0.81 1.9 0.037 Jam frequency (W−1b ) Relative gradient 0.57 1.0 0.83 1.0 0.033 % pieces in jams Gradient (m/m) 0.000038 0.014 0.65 6.1 0.098 Mean blockage ratio Gradient (m/m) 0.017 0.013 0.66 0.0017 0.095 Projected area Gradient (m/m) 0.16 0.013 0.97 0.28 0.0023 Projected area Relative gradient 8.9 0.96 0.86 1.6 0.02364 4.3.3 Shear Stress Partitioning The partitioning of shear stresses, as well as the Manning’s roughness factor, for each experiment are summarized in Table 4.6. In each of the experiments, the Manning’s roughness was greater with wood in the channel than prior to addition or following removal. Conversely, the ratio of grain shear stress to total shear stress was lower with wood in the channel. Table 4.6: A comparison of Manning’s roughness coefficient and shear stress from the five flume experiments. Experiment Treatment 1 2 3 4 5 n Pre-LW 0.019 0.020 0.022 0.022 0.021 With LW - 0.025 0.027 0.029 0.031 Post-LW - 0.022 0.023 - - τg/τ Pre-LW 0.69 0.65 0.57 0.60 0.61 With LW - 0.51 0.41 0.36 0.35 Post-LW - 0.58 0.57 - - Manning’s roughness parameter (n) increased from 21% in experiment 2 to 43% in experiment 5. Compared to the changes in Manning’s n, the median sur- face grain size (D50) and the associated grain roughness (ng) varied little with the addition of wood; ng decreased by 0.8% to 1.8% in experiments 3 to 5, and actu- ally increased by 3.4% in experiment 2. Thus, increases in roughess caused nearly equivalent decreases (see equation 4.20) in the ratio of grain shear stress to total shear stress. Similar to flow velocity, Manning’s n and τg/τ did not recover fully to the pre-addition values following the removal of large wood. By increasing flow depth (and the associated hydraulic radius, R) and mean wa- ter surface slope, the addition of wood actually increased the total shear stress in the channel, while also altering the partitioning of shear stress between grain shear stress and bedform shear stress.The magnitude of the decrease in the proportion of total shear stress partitioned by grain shear stress ((τg/τ)/(τg0/τ0)) was signif- icantly (α = 0.05) related to several of the wood metrics presented in the previous 65 Figure 4.11: The ratio of the shear stress partitioning prior to and following the addition of wood is compared with wood load and piece frequency. sections (table 5.9). Wood load, piece frequency, and total projected area were most strongly related to this ratio, while blockage ratio was more closely related to the absolute ratio of grain shear stress to total shear stress (τg/τ) with wood in the channel (figure 4.11). All four metrics, however, were significantly related to both 66 the magnitude of the change in the shear stress partitioning, and the absolute value of the ratio at equilibrium following the addition of wood (table 5.9). Figure 4.12: The relative increase in Manning’s n at steady state following the addition of wood was related to the wood load and piece frequency. 67 Table 4.7: All significant (α = 0.1) relationships between wood metrics and shear stress partitioning are described. Independent Variable Dependent Variable Coefficient Intercept R2 p(1) x 104 p(2) Wood load (m3/m2) τg/τ -380 0.66 0.93 3.7 0.0079 Wood load (m3/m2) Relative partitioning -0.015 0.66 0.98 3.3 0.0071 Piece frequency (m−2) τg/τ -470 0.98 0.94 0.29 0.0012 Piece frequency (m−2) Relative partitioning -0.018 0.98 0.98 0.21 0.00081 % pieces in jams τg/τ -0.0027 0.58 0.66 30 0.097 Mean blockage ratio τg/τ -1.4 0.68 0.94 3.2 0.0060 Mean blockage ratio Relative partitioning -1.6 0.99 0.84 8.4 0.029 Projected area τg/τ -11 0.63 0.88 7.3 0.019 Projected area Relative partitioning -13 0.94 0.90 2.9 0.014 68 Table 4.8: All significant (α = 0.1) relationships between wood metrics and Manning’s n are described. Independent Variable Dependent Variable Coefficient Intercept R2 p(1) x 104 p(2) Wood load (m3/m2) n 12 0.020 0.97 1.0 0.0019 Wood load (m3/m2) Relative n 460 1.0 0.98 0.23 0.0011 Piece frequency (m−2) n 0.00048 0.020 0.98 0.92 0.0016 Piece frequency (m−2) Relative n 0.018 1.0 0.98 0.17 0.00080 % pieces in jams n 0.000085 0.023 0.66 17 0.094 Mean blockage ratio n 0.044 0.019 0.90 10 0.014 Mean blockage ratio Relative n 1.5 1.0 0.75 13 0.056 Projected area n 0.35 0.021 0.93 3.1 0.0084 Projected area Relative n 13 1.1 0.88 2.6 0.019 69 Similarly, the same four metrics were also significantly related to both the change in the Manning’s roughness (n) following the addition of wood, and the absolute value of Manning’s roughness with wood in the channel, except for the mean piece blockage ratio which was only significantly related to the absolute value of Manning’s roughness (table 4.8). Wood load and piece frequency were most strongly related to the magnitude of the change in Manning’s roughness fol- lowing the addition of wood, while blockage ratio and total projected area were more strongly related to the absolute value of Manning’s n with wood in the chan- nel (figure 4.12). 4.4 Discussion The results presented in this chapter indicate that large wood significantly alters stream hydraulics. The addition of wood lowered the peak-derived flow velocity by 11% to 14% in experiments 2 to 5. Previous authors have found increases in bankfull velocity ranging from 20% to 83%following de-snagging (Erskine, 1994b; Reinfelds et al., 1995). Increases of reach-averaged flow velocity from 52% to 380% have been measured at low flow following wood removal (Mason et al., 1990; Shields and Smith, 1992), as well as local increases of 250% (MacDonald and Keller, 1987). While the changes in flow velocity measured in this study appear modest in comparison to these results, it is difficult to compare between studies as the effects of wood on flow velocity are inversely related to discharge; at high flows previous studies suggest there may be no difference in flow velocity (Mason et al., 1990; Shields and Smith, 1992). Furthermore, field results generally reflect local velocities derived from point measurements rather than true reach averages. As predicted, wood input was associated with increases in water surface ele- vation. The water stage increases produced by the addition of large wood ranged from 1.4% to 5.9% of the flow depth, and were attributable to both aggradation of sediment and increased flow depth. The stage increases appear to be within the range of previously measured values: stage increases of up to 9% were measured in the Tamut River, Australia, in reaches with blockage ratios of 0.3-0.4, while debris with blockage ratios below 0.1 did not produce significant stage changes (Gippel et al., 1992; Gippel et al., 1996). Gippel et al. (1996) also measured smaller increases 70 of 0.8% to 1.5% in a similar flume experiment modeled on the Thompson River, Australia, while finding stage increases of only 0.2% in an undisturbed reach of the Thompson River with a median blockage ratio of only 0.004. The importance of blockage ratio in producing stage increase is supported by the present research, as the dimensionless projected area of large wood explained more than 97% of the variance in stage increase. Wood load and piece frequency, which are both closely related to the projected area, explained more than 93% of the variance in stage. A greater proportion of the stage increase occurred in the upper portion of the reach, producing an increased water surface gradient following the addition of wood. Slope increases combined with increases in flow depth and the associated hydraulic radius to produce increases in total shear stress following the addition of large wood. Similarly, Manga and Kirchner (2000) found that total shear stress increase with decreasing log spacing. As expected, channel roughness significantly increased following the addition of large wood. Increases of Mannings n from 21% to 43% were of a similar magni- tude to those measured by previous investigators. In a survey of previous research, Shields and Gippel (1992) found increases in n following the addition of large wood ranging from 11% in the Murray-Darling Basin Commission, to 100% in the River Yare, U.K. (Watts and Watts, 1990). Brooks et al. (2004) measured increases of n from 0-33% following the addition of engineered log jams in a reach of the Williams River, Australia. While Gippel et al. (1996) suggest, once again, that frictional resistance and Mannings n are dependent on flow blockage, the value of Mannings n was most closely related to wood load and piece frequency in the current study. While total reach-averaged shear stress increases with wood load, less shear stress is borne by the channel bed (Manga and Kirchner, 2000). The results show that the ratio of grain shear stress to total shear stress decreased following the ad- dition of large wood by 21% to 43%, suggesting that the proportion of shear stress taken up by bedform shear stress and large wood shear stress increased concomi- tantly. Similarly, Manga and Kirchner (2000) showed that the proportion of shear stress taken up by large wood was inversely related to log spacing, while Asssani and Petit (1995) measured decreases in the proportion of grain shear stress of 73% 71 following wood removal in Ru des Waidages, France. Changes in the proportion of grain shear stress in the present study were closely related to the amount of wood in the channel, with the wood load and piece frequency explaining more than 98% of the variation. The addition of large wood significantly altered the flow velocity, water stage and water surface gradient, as well as the channel roughness, shear stress, and shear stress partitioning in the experiments presented herein. Changes in these param- eters were generally most closely related to wood density (both piece frequency and wood load) and the total projected area of wood in the channel, highlighting the influence of wood orientation and blockage ratio on channel hydraulics. While the removal of wood had the opposite effect on each of the measured hydraulic parameters, the parameters did not entirely return to the pre-addition state. 72 Chapter 5 Channel Morphology 5.1 Introduction The morphologic effects of in-stream wood have been extensively studied over the past four decades. Large wood causes variations in sediment transport which create a stepped profile, as sediment is generally deposited and stored upstream of obstructions and scoured downstream of them (Keller and Tally, 1979). Overall in-stream wood has been shown to increase sediment retention and storage in a reach; a large proportion of the storage in forested channels is directly attributable to wood accumulations, even where the proportion of total elevation lost to large wood is low (Keller and Swanson,1979; Bilby, 1981; Megahan, 1982; Thompson, 1995; Brooks et al., 2004, 2006; Andreoli et al., 2007). Sediment storage upstream of large wood accumulations may mitigate the effects of sporadic sediment inputs from hillslopes by preventing rapid sedimentation of downstream reaches (Hassan et al., 2005). Large wood-related variations in sediment transport are reflected in channel morphology. Scour associated with concentrated flow downstream of large wood increases pool frequency (Keller and Tally, 1979; Beechie and Sibley, 1997; Jack- son and Sturm, 2002; Faustini and Jones, 2003). As a result, a significant per- centage of pools in high gradient streams are associated with wood accumulations (Andrus et al., 1988; Montgomery et al., 1995; Erskine and Webb, 2003). Scoured material is deposited upstream of subsequent flow obstructions and alters bar am- 73 plitude; as wood load increases the increasingly stepped profile may force a transi- tion from plane-bed to riffle-pool morphologies, potentially increasing habitat di- versity (Montgomery and Buffington, 1997; Abbe and Montgomery, 2003; Brooks et al., 2004; Eaton et al., 2010). The alternating regions of high and low transport capacity associated wood ob- structions also produce local variations in the surface grain size distributions. Large wood steps promote the deposition of fine sediment upstream of obstructions, while coarsening the bed downstream where flow is concentrated, thereby enhancing fa- cies complexity (Buffington and Montgomery, 1999b; Andrews, 2010). The net effect of wood addition is generally a decrease in surface grain size (Manga and Kirchner, 2000), which can create a negative feedback by enhancing sediment ex- port from storage reservoirs (Lisle and Church, 2002). Experimental removal of wood steps increases sediment transport, releasing stored sediment and coarsening the bed upstream of the removed wood, while si- multaneously infilling downstream pools (Beschta, 1979; Bilby and Likens, 1980; Bilby, 1981; Mosley, 1981; Smith et al., 1993; Lisle, 1995; Gurnell and Sweets, 1998). By increasing sediment transport from upstream storage reservoirs and en- hancing sedimentation in downstream scour pools, wood removal therefore reduces bed complexity (Bilby and Ward, 1991). Interestingly, however, pool frequency may be maintained in some channels in the absence of large wood through in- creased bar formation (Smith et al., 1993; Lisle, 1995). Thus, when roughness elements are removed, streams may compensate by increasing roughness through sedimentary bedforms to maintain equilibrium. Alternatively, this suggests that a threshold level of wood loading may be required prior to the onset of significant morphologic change in a wood-depleted channel. Experimental installation of en- gineered log jams (ELJ) has been reported increase sediment storage and reduce surface grain size (Brooks et al., 2004, 2006). Despite extensive research, the relationship between wood volume and the ex- tent of morphologic change in a channel remains largely unknown. While it is in- creasingly understood that large wood enhances hydraulic complexity, and thereby produces a diversity of aquatic habitats, the amount of wood needed to produce these favourable changes has not been reliably quantified. The primary objective of this chapter is to quantify the effects of varying amounts of added in-stream large 74 wood on channel morphology. Changes in sediment transport and storage will first be presented, followed by the effects of wood addition on channel morphology and surficial sedimentological characteristics. Throughout the chapter these changes will be compared to the six wood-related metrics presented in chapter 3. 5.2 Methods 5.2.1 Sediment Transport and Storage Sediment output was collected in a 0.177 mm mesh bag located at the stream ta- ble outlet and removed at 15 minute intervals. After removal, the mesh bag was inverted over a 0.025 mm sieve and rinsed to transfer all fine particles. The out- put samples were then oven-dried at 250 ◦C, and weighed to determine the average sediment output per unit time. The dried output samples were later combined into five hourly samples, and a sub-sample of 250-350 g from each hour was sieved. Each size class was then weighed to determine the grain size distribution of the output. Sediment storage was first determined at each 15-minute interval by calculat- ing the difference between the cumulative input and output values. Volumes were derived from dried weights using a specific dry sediment weight of 2650 kg/m3. Vc = Σ(Input−Out put) (5.1) A transport-storage coefficient, K, measured in units of time−1, was then de- rived from the empirical linear transport-storage relationship produced in experi- ments 3 and 4 following wood removal according to methodology employed by Lisle and Church (2002): ω = KV (5.2) The theoretical volume of stored sediment at each sample time (t) following wood removal was reconstructed using an exponential decay function such that: V −VBL = (V0−VBL)exp(−Kt) (5.3) 75 where VBL is a reference volume of stored sediment (cm3) calculated from the transport-storage relation as the storage volume when output is zero, and V0 is the volume of sediment stored at the time of wood removal. Morphologic estimates of the change in sediment stage, defined as the mean bed elevation, and sediment storage were calculated from laser images of the bed elevation. The images were collected following each experiment at 0.05 m intervals using a laser and camera mounted on a mobile cart. To optimize the quality of the output, the laser images were taken after all surface water has drained from the stream table. The images were further processed in MATLAB to eliminate radial distortion as well as wood pieces. The spatial view of the camera was first determined by digitizing an image of a calibration template composed of dots separated by 18.7 mm along the x and z axes. The calibration map produced by the digitization of each point on the calibration template was then used to correct and re-shape each laser image. Actual x and z coordinates (above an arbitrary datum) of a point on the left and right bank were determined for each cross section and used to rotate the image. Background noise was reduced by eliminated patches containing fewer than 8 pixels and the bed of each cross section was then re-defined by manually en- tering points along the bed of the resulting skeleton image of the laser photograph. This manual process was used to further eliminate noise in the image as well as to remove wood from each profile and define the lateral extents of the channel bed. Changes in sediment stage (δE), as well as morphologic estimates of sediment storage (Vm), were calculated according to: δEbed = E0−EE (5.4) Vm = Σ(δEbed) ·Wb ·Li (5.5) where E0 is the initial bed elevation at a cross section in metres, EE is the elevation at a cross section at equilibrium, Wb is the bankfull channel width, and Li is the distance between cross sections. 76 5.2.2 Channel Morphology Changes in channel morphology were assessed using longitudinal profiles along the channel thalweg – or minimum cross-sectional bed elevation – constructed from laser images. Pools were defined according the criteria of Montgomery et al. (1995) as sections with a maximum residual depth greater than 25% of the bankful depth, and a width or length of at least 10% of the bankfull channel width. This definition is similar to that employed by Kreutzweiser et al. (2005), who de- fined pools as having a minimum width of at least 10% of the bankful width, and a residual depth of at least10 cm. Bankfull depth (d) was calculated for each of the longitudinal profiles analysed according to: d = Q v ·Wb (5.6) Residual depths were calculated from the profiles of minimum bed elevation using the method described by Lisle (1987). Pools were identified as locations where the difference in elevation between the minimum point of the topographic depres- sion and the maximum height of the successive riffle was greater than 25% of the calculated bankfull depth. Approximate pool lengths were then measured as the horizontal distance between adjacent riffles. Overhead photographs were used to determine the length of pools located near the first or last surveyed cross section, to assure that seperate pools were not grouped together, and to assess the widths of the smallest pools. Pools were then defined as log-affected or non-log-affected according to the methods of Webb and Erskine (2003). Non-log-affected pools were generally lo- cated along the inside of the channel bends, and were usually present prior to the addition of wood. Log-affected pools were often located in straight channel seg- ments in the proximity of wood pieces, and were generally in locations which did not contain pools prior to wood emplacement. 5.2.3 Sedimentological Characteristics The bed of the channel was mapped manually at the end of each run to determine facies boundaries. Facies maps, which differentiate between texturally distinct 77 patches, were generated for the entire channel following each run. Photos were then taken of a 6 inch by 8.5 inch segment of each facies, and used to determine its approximate grain size. To determine grain size, Wolman samples were first conducted on twenty of the facies photos by measuring grain sizes in Adobe Illustrator along a super-imposed grid. Each measured facies was then classified according to the facies categories in Table 5.1. The remaining facies images were then classified visually by referencing the photographs of the measured facies. Examples of each category are shown in figures 5.1-5.5. Later Wolman sampling of ten of these facies revealed that nine of ten were correctly classified. Table 5.1: Facies grain size categories, as well as the characteristic grain size of each category, are defined. Category Size Range (mm) Characteristic Size (mm) Very Fine 0.35-0.50 0.43 Fine 0.50-0.71 0.60 Medium 0.71-1.0 0.85 Coarse 1.0-1.4 1.2 Very Coarse 1.4-2.0 1.7 The proportion of the reach occupied by each facies (Fp) was determined using a transparent dot sheet. Weighted averages of surface grain size were calculated using the facies areas and the characteristic grain size of the category. D50 = Σ(Fp ·D50p) (5.7) 5.3 Results 5.3.1 Sediment Transport and Storage The ratio of sediment output to input over time is shown in figure 5.6, for each experiment. In experiments 2 to 5 sediment transport increased briefly, relative to sediment input, following the addition of wood. In experiments 3 to 5, sediment transport then decreased for a prolonged period, ranging from approximately 2000 78 Figure 5.1: A photograph of a very fine facies (D50 < 0.5 mm) is shown. Figure 5.2: A photograph of a fine facies (D50 = 0.53 mm) is shown. 79 Figure 5.3: A photograph of a medium facies (D50 = 0.88 mm) is shown. Figure 5.4: A photograph of a coarse facies (D50 = 1.18 mm) is shown. 80 Figure 5.5: A photograph of a very coarse facies (D50 = 1.58 mm) is shown. minutes in experiment 3 to over 3000 minutes in experiment 5, equivalent to 180 to 270 hours of morphologically active flows in the prototype system (or 7 to 10 years). Immediately following wood removal, the ratio of sediment output to input increased 3.7-fold in experiment 3 and 4.7-fold in experiment 4, before returning to equilibrium within a single run (300 minutes, or one year of flow). The removal of wood in experiment 2, in which the reach did not experience prologonged aggra- dation, had little apparent effect on sediment transport. The grain size of the transported material varied throughout the experiments, but was not dependent on wood presence or absence, or wood volume. Instead, both the median grain size (D50) and the D90 were positively related to the rate of sediment transport (figure 5.7). Thus, grain size increased briefly during the periods of scour following wood addition and removal, and decreased during the aggradational phase associated with wood addition. Net sediment storage, sediment stage change (the change in the mean cross sectional bed elevation), and the storage efficiency of the added wood for each ex- periment is shown in Table 5.2. According to the cumulative data (VC), the volume 81 Figure 5.6: The ratio of sediment output to sediment input is shown over time for the five experiments. Black arrows indicate the addition or removal of wood or sediment in experiments 1 (top) to 5 (bottom). 82 Figure 5.7: The D50 and D90 of the sediment output are compared to rate of sediment output. of sediment (m3) stored for each cubic metre of added wood ranged from 3.3 to 8.0. The volume of sediment stored per piece of wood ranged from 0.00012 m3 in experiment 2 to 0.00030 m3 in experiment 4, which is equivalent to sediment storage of 3.3 to 8.2 m3 per wood piece in the prototype system. Table 5.2: Storage values calculated using both the cumulative and morpho- logic methods are compared for experiments 2 to 5, as well as sediment stage change. Experiment Parameter Unit 2 3 4 5 Sediment Storage (Vc) cm3 1850 4890 9110 6320 Storage/Wood Volume m3/m3 3.3 6.1 8.0 4.4 Storage/piece m3/piece 0.00012 0.00023 0.00030 0.00017 Sediment Stage Change mm -0.60 2.9 7.7 2.9 Sediment Storage (Vm) cm3 -850 4450 11900 4510 83 Despite minimal sediment storage, the mean sediment stage actually decreased by 0.55 mm in experiment 2. Increases in sediment stage ranged from 2.9 mm in experiments 3 and 5 to a maximum of 7.7 mm in experiment 4. When inte- grated over the entire reach, these stage increases were approximately equivalent to changes in sediment storage (Vm) of -0.853 cm3 to 11900 cm3. That sediment stage decreased despite aggradation of sediment throughout the reach in experi- ment 2 highlights the inadequacy of these point measurements, which fail to cap- ture variability in sediment stage occuring between the cross sections. The dependence of storage estimates on the method used reveals the uncer- tainty associated with both the cumulative and morphologic methods. Estimates based on cumulative differences between input and output likely represent the up- per bound of sediment storage, as losses of sediment at the outlet and periodic decreases in input produce an over-estimate of storage volume. With the excep- tion of experiment 4, the cumulative estimates of sediment storage exceeded the estimates based on sediment stage (figure 5.8). The morphologic method is also innacurate, however, as it does not account for local differences in channel width and is based on point measurements. The total volume of storage calculated using the cumulative method was sig- nificantly related to piece frequency, and nearly significantly to wood load, mean blockage ratio, and total projected area (figure 5.9; table 5.3). The sediment stage and the morphologically-derived sediment storage volumes, however, were not sig- nificantly related to any of the wood metrics considered in this analysis. 84 Figure 5.8: Estimates of storage derived using the morphologic method and the cumulative method are compared for each experiment, with a 1:1 line shown in red. 85 Table 5.3: All significant (α = 0.1) relationships between wood metrics and sediment storage, as well as storage effi- ciency, are described. Independent Variable Dependent Variable Coefficient Intercept R2 p(1) x 104 p(2) Wood load (m3/m2) Sediment storage (cm3) 8900000 -89 0.76 0.96 0.052 Piece frequency (m−2) Sediment storage (cm3) 340 -150 0.77 0.94 0.049 Mean blockage ratio Sediment storage (cm3) 3086 -380 0.70 0.86 0.076 Projected area Sediment storage (cm3) 250000 490 0.75 0.78 0.059 Wood load (m3/m2) Storage efficiency (m3/m3) 11000 -0.16 0.98 0.78 0.0085 Wood load (m3/m2) Storage efficiency (m3/piece) 11000 -0.18 0.98 0.78 0.0098 Piece frequency (m−2) Storage efficiency (m3/m3) 0.42 -0.17 0.98 0.79 0.010 Piece frequency (m−2) Storage efficiency (m3/piece) 0.44 -0.18 0.98 0.79 0.011 Mean blockage ratio Storage efficiency (m3/m3) 31 -0.037 0.89 0.98 0.055 Mean blockage ratio Storage efficiency (m3/piece) 32 -0.060 0.90 0.97 0.054 Projected area Storage efficiency (m3/m3) 330 0.43 0.97 0.53 0.013 Projected area Storage efficiency (m3/piece) 340 0.42 0.98 0.53 0.012 86 Figure 5.9: The volume of stored sediment (derived using the cumulative method) is linearly related to the piece frequency. Using data from all five experiments, the volume of sediment stored per m3 of wood added – a measure of the storage efficiency of the large wood – was related to the mean piece blockage ratio (p = 0.060), as was the volume of sediment stored per piece of wood (p = 0.057). When the final experiment – which did not proceed to steady state – was removed from the data, there was a strong linear relationship between wood load and the volume of sediment stored per m3 of added wood, as well as the volume of sediment stored per piece (figure 5.10). Both measures of storage efficiency were also significantly (α = 0.05) related to the piece frequency and total projected area, while the strength of the relationship to mean piece block- age ratio decreased when the final experiment was excluded (table 5.3). A transport-storage relation was calculated for each of the experiments in which wood was removed from the reach, using the storage values calculated from the cu- mulative difference between input and output,VC. While a strong positive relation- ship existed for the period of degradation following wood removal in experiments 3 (p = 1.4e−12) and 4 (p = 2.8e−8), there was a weaker negative relationship in experiment 2 (p = 0.020), where the removal of wood appeared to have little effect 87 Figure 5.10: Storage efficiency, measured as the volume of sediment stored per m3 and per piece of added wood, is compared with wood loading for experiments 1 to 4. 88 Figure 5.11: The volume of sediment storage is compared to the transport rate for the degradational phase following wood removal in experi- ments 2 to 4. The red lines represents the linear trendlines used to derive the values of K and VBL. 89 on sediment transport (figure 5.11). The relations developed for experiments 3 and 4 (figure 5.11) were used to determine the transport-storage coefficient (K) as well as the reference storage vol- ume (VBL). These values were used to predict the volume of storage based on the exponential decay function shown in equation 5.3. The reconstructed curve con- sistently under-estimated the actual storage volume in experiments 3 and 4 (5.12). This suggests that degradational model of Lisle and Church (2002), which was de- rived from experimental sediment feed reduction, does not capture the dynamics of sediment evacuation following wood removal. 5.3.2 Channel Morphology The results presented in Table 5.4) show that pool frequency and spacing changed significantly following the addition of large wood. Pool number increased by 20- 230% in experiments 2 to 5. The increase in pool frequency following wood ad- dition produced a concomitant increase in the total length of channel occupied by pools, which increased by13-91%. The longitudinal profiles of channel depth, shown in figure 5.13), reveal that most pool addition occured in the upper half of the reach. The number of pools at steady state, as well as pool spacing (measured in multiples of the bankfull channel width), was significantly related to all of the wood-related metrics used in the analysis, and most strongly to the blockage ratio (table 5.5). The percentage of the reach occupied by pools at steady state was also significantly related to all of the wood metrics, with the strongest relationship to the percentage of pieces in jams (figure 5.14). The change in pool number and spacing, however, was not related to any of the metrics, while the change in total pool length was significantly related to wood load, piece frequency, and the dimensionless projected area of the added wood (table 5.5). 90 Table 5.4: Pool number, spacing, and total pool length are shown for all ex- periments prior to treatment, following wood or pulse addition, and fol- lowing wood removal. Experiment Parameter Unit 1 2 3 4 5 Pool Number Pre-LW 5 5 9 3 6 With LW 5 6 11 10 12 Post-LW - 6 - 5 - Pool Spacing Wb Pre-LW 2.6 2.6 1.5 4.4 2.2 With LW 2.6 2.2 1.2 1.3 1.1 Post-LW - 2.2 - 2.6 - Total Pool Length % Pre-LW 50 42 71 41 49 With LW 42 50 80 69 93 Post-LW - 31 - 49 - 91 Figure 5.12: Actual values of sediment storage (derived using the cumulative method) are shown in blue for the degradational period following wood removal in experiments 3 and 4. Predicted values of sediment storage, derived using an exponential decay function, are shown in red. 92 Figure 5.13: Longitudinal profiles of the minimum channel depth (thalweg) prior to wood addition, following wood addition, and following re- moval are shown for experiments 2 to 5 (vertical exaggeration = 7). 93 Table 5.5: All significant (α = 0.1) relationships between wood metrics and pool spacing, as well as total pool length and the percentage of log-affected pools, are described. Independent Variable Dependent Variable Coefficient Intercept R2 p(1) p(2) Wood load (m3/m2) Pool spacing (Wb) -1800 2.6 0.84 0.0024 0.030 Piece frequency (m−2) Pool spacing (Wb) -0.068 2.6 0.83 0.0025 0.030 Jam frequency (W−1b ) Pool spacing (Wb) -3.0 2.2 0.7 0.0038 0.073 % pieces in jams Pool spacing (Wb) -0.015 2.3 0.89 0.00093 0.016 Mean blockage ratio Pool spacing (Wb) -6.8 2.7 0.94 0.00055 0.0059 Projected area Pool spacing (Wb) -52 2.5 0.91 0.00086 0.012 Wood load (m3/m2) % pool 52000 40 0.79 0.023 0.043 Wood load (m3/m2) Relative pool length 1200 0.76 0.90 0.010 0.013 Piece frequency (m−2) % pool 2.0 40 0.78 0.0025 0.047 Piece frequency (m−2) Relative pool length 0.045 0.75 0.90 0.011 0.014 Jam frequency (W−1b ) % pool 110 49 0.93 0.0012 0.0080 % pieces in jams % pool 0.47 46 0.94 0.0017 0.0072 Mean blockage ratio % pool 190 37 0.78 0.038 0.049 Projected area %pool 1600 41 0.91 0.0050 0.011 Projected area Relative pool length 36 0.85 0.82 0.014 0.035 Wood load (m3/m2) % log-affected 7000 3.5 0.87 0.74 0.020 Piece frequency (m−2) % log-affected 2.7 3.0 0.88 0.77 0.018 94 Figure 5.14: Pool spacing and the percentage of the reach length occupied by pools are compared to the percentage of pieces in jams. Pool depth and length were also measured for each pool. Mean pool length and depth were 0.36 m and 0.014 m, respectively, for the aggregated results of all five experiments at steady state. The average pool length and depth did not differ significantly (α = 0.05) between the five experiments prior to, or following, wood addition. Wood addition alone did not appear to control pool width or depth, as 95 the average depth and length of the pools also did not change significantly within each experiment following the addition or removal of wood, except in experiment 4 where mean pool length decreased by 50% (p = 0.0264) following wood addition. The number of log-affected and non-log-affected pools were also determined for each experiment at steady state. The percentage of pools that were log-affected varied from 33-70%, and was significantly related to both wood load and piece frequency (figure 5.15; table 5.5). Figure 5.15: The percentage of log-affected pools is compared to the piece frequency. Non-log-affected pools generally formed at the inside of meander bends and were significantly longer (0.39 m or 1.1Wb) than the log-affected pools (0.28 m or 0.81Wb; p = 0.011), which formed due to log-related scour (figure 5.16). The difference in the depth of log-affected (0.012 m) and non-log-affected (0.015 m) pools was not significant. These pool dimensions were equivalent to widths of 8.4 m and 12 m, and depths of 0.36 and 0.45 m in the prototype reach. Following the removal of wood in experiments 2 and 4, total pool length de- creased in both experiments, despite no change in the number of pools in experi- ment 2. Similarly, while the number of pools did not change following the release 96 Figure 5.16: Mean pool length and depth are compared for log-affected and non-log-affected pools. of the sediment pulse in experiment 1, which was designed to mimic upstream wood removal or jam failure, the total percentage of the reach length occupied by pools also decreased. These results suggest that in-filling of pools resulted from the release of stored sediment upstream, but was not sufficient to eliminate entire 97 log-induced pools. The variability of the bed elevations within each cross section was also con- sidered (table 5.6). The addition of large wood significantly increased the average cross-sectional bed variability in experiments 2 to 5, while the release of the sedi- ment pulse in experiment 1 decreased the variance. Conversely, wood removal in experiments 2 and 4 significantly decreased the depth variability. The final value of the variance with wood in the reach was not significantly related to any of the wood metrics considered in this analysis, however, while the change in variance was weakly related to the mean blockage ratio of the wood pieces (table 5.7). Table 5.6: A comparison of the mean cross-sectional variance in bed eleva- tion, mean bed gradient, and median bed gradient – all derived using the minimum sediment stage – is shown for all treatment stages. Experiment Parameter Unit 1 2 3 4 5 Depth Variability m Pre-LW 0.037 0.048 0.028 0.057 0.037 With LW 0.033 0.052 0.053 0.099 0.051 Post-LW - 0.029 - 0.066 - Mean Bed Gradient m/m Pre-LW 0.018 0.020 0.017 0.022 0.021 With LW 0.017 0.019 0.014 0.018 0.021 Post-LW - 0.020 - 0.019 - Median Gradient m/m Pre-LW 0.017 0.016 0.015 0.014 0.014 With LW 0.0087 0.012 0.0057 0.0058 0.0089 Post-LW - 0.0082 - 0.016 - 98 Table 5.7: All significant (α = 0.1) relationships between wood metrics and bed morphology, as well as gradient, are described. Independent Variable Dependent Variable Coefficient Intercept R2 p(1) x 104 p(2) Wood load (m3/m2) Mean gradient (m/m) 5.6 0.013 0.92 1.8 0.0090 Wood load (m3/m2) Relative mean gradient 300 0.96 0.86 1.8 0.023 Piece frequency (m−2) Mean gradient (m/m) 0.00021 0.013 0.92 1.9 0.0093 Piece frequency (m−2) Relative mean gradient 0.011 0.96 0.85 2.0 0.026 Jam frequency (W−1b ) Relative mean gradient 0.57 1.0 0.91 0.30 0.011 Mean blockage ratio Relative mean bed variance 3.4 0.89 0.69 310 0.081 Mean blockage ratio Median thalweg gradient (m/m) -0.047 0.017 0.92 12 0.0096 Mean blockage ratio Relative median thalweg gradient -2.5 1.0 0.87 19 0.021 Projected area Mean gradient (m/m) 0.16 0.013 0.89 2.3 0.017 Projected area Relative mean gradient 8.8 0.97 0.93 0.45 0.0077 99 The median gradient of the thalweg changed dramatically following treatment in all five experiments; following the addition of large wood in experiments 2 to 4 the median gradient decreased by 23-63%, while the median gradient decreased by 49% following the sediment pulse release in experiment 1. The median bed slope at steady state, as well as the change in median bed slope following wood addition, were only significantly related to the mean piece blockage ratio (figure 5.17; table 5.7). Figure 5.17: The median bed gradient is negatively related to the mean block- age ratio. Mean and median bed gradients were also determined based on the mean bed elevation for each cross section. Similar to the water surface gradient, the mean bed gradient increased in experiments 2 to 5 following the addition of large wood, and decreased following wood removal. The steady state mean bed gradient calculated from the mean sediment stage was significantly related to wood load and piece frequency (figure 5.18), as well as the projected area of the added wood (table 5.7). The change in mean bed slope was linearly related to these same variables, as well as the jam frequency (figure 5.19). 100 Figure 5.18: The mean bed gradient at steady state, derived from the mean sediment stage in each cross section, is compared to the wood load and piece frequency. 5.3.3 Sedimentological Characteristics Changes in facies distribution and surface grain size following the addition of large wood in experiments 2 to 5, as well as following the release of the sediment pulse 101 Figure 5.19: The relative mean bed gradient, derived from the mean sediment stage in each cross section, is compared to the dimensionless projected area of the wood and the jam frequency. in experiment 1, are shown in Table 5.8. The release of the sediment pulse in experiment 1 caused an overall coarsening of the bed surface, attributable to a decrease in the percentage of the bed covered by fine and very fine facies, as well 102 as a concomitant increase in very coarse facies. Similarly, the addition of large wood in experiment 2 caused an increase in the area-weighted average surface D50; the addition of 0.00037 m3/m2 of large wood produced a decrease in the area of all facies classes except the very coarse category, which increased in size from 2.3% to 38% of the surface area. Table 5.8: Changes in facies size following the addition of wood or a sedi- ment pulse are compared for each experiment, as well as the change in the median grain size and facies number. Experiment Parameter 1 2 3 4 5 % Change in Facies Size Very Fine -4.0 -1.1 0 1.2 5.9 Fine -9.2 -1.4 9.6 19 18 Medium 21 -22 10 -1.2 -5.9 Coarse -17 -11 -29 -15 -25 Very Coarse 9.0 35 8.6 -3.3 7.7 D50E /D50 1.1 1.2 0.95 0.90 0.90 FaciesE /Facies 1.0 1.7 1.9 1.6 1.8 The addition of greater volumes of wood in experiments 3 to 5 was associated with a decrease in the surface D50 of 4.5% to 10%. While the percentage change in the proportion of the bed occupied by the very fine fraction increased with wood load, changes in the other facies categories were less consistent. In experiment 4 the proportion of the bed occupied by the coarser facies (from medium to very coarse) decreased, while the surface area of the finer (fine and very fine) facies increased. In experiment 3 and 5, however, the surface area of the very coarse facies increased, likely due to localized scour around wood pieces and jams. The addition of wood in experiments 2 to 5 increased the number of facies in the reach, while the release of the sediment pulse in experiment 1 had no effect on facies number. The number of facies in the reach in the absence of large wood was fairly consistent; prior to wood addition the reach contained 14 facies in experi- ment 3 and 16 facies in experiments 2,4, and 5. The reach contained 15 facies in experiment 1 both before and after the release of the sediment pulse. There was 103 also little variation in facies number following wood addition, with the number of facies ranging from 26 in experiments 3 and 4, to 28 in experiment 5. The proportional increase in facies number following wood addition varied from 63% in experiment 4, to 86% in experiment 3. Of the wood-related metrics considered in this analysis, the increase in facies number was only significantly related to the mean piece blockage ratio (table 5.9). Using α = 0.1, the absolute number of facies at steady state was significantly related to piece frequency, wood load, and mean piece blockage ratio (figure 5.20). While the median grain size at steady state was not significantly related to any of the wood-related metrics, the change in surface grain size was significantly related (α = 0.1) to the total projected area of the added wood, as well as the percentage of wood pieces in jams (figure 5.21; table 5.9). Figure 5.20: The relative facies number at steady state is compared to the mean piece blockage ratio of the added wood. 104 Figure 5.21: The proportional change in the median surficial grain size (D50) is compared with the percentage of piece in jams and the dimensionless projected area of the added wood. 105 Table 5.9: All significant (α = 0.1) relationships between wood metrics and facies number, as well as median surficial grain size (D50), are described. Independent Variable Dependent Variable Coefficient Intercept R2 p(1) x 103 p(2) Wood load (m3/m2) Number of facies 12000 18 0.69 8.2 0.081 Piece frequency (m−2) Number of facies 0.48 18 0.70 8.1 0.077 Mean blockage ratio Number of facies 47 17 0.75 8.8 0.057 Mean blockage ratio Relative number of facies 1.1 3.2 0.84 4.6 0.029 Mean blockage ratio Relative median thalweg gradient -2.5 1.0 0.87 19 0.021 % pieces in jams Relative surficial D50 -0.0028 1.1 0.73 0.28 0.065 Projected area Relative surficial D50 -9.5 1.2 0.70 0.46 0.079 106 5.4 Discussion The addition of large wood caused hydraulic and morphologic changes in the ex- perimental reach. Increased flow resistance caused immediate decreases in sedi- ment transport, which were associated with significant sediment storage in exper- iments 3 to 5, and minimal storage in experiment 2. The storage efficiency of the added wood varied from 3.3 to 8.0 m3 per m3 of wood, and from 3.3 to 8.2 m3 per piece. The lower end of these results, found in experiments 2 and 5, was similar in magnitude to the average value of 3.5 m3 of sediment storage per m3 of wood reported by Brooks et al. (2004) following the introduction of engineered log jams in the Williams River, Australia, but was significantly higher then the average stor- age volume of 0.8 m3 per obstruction reported by Megahan (1982) in seven Idaho streams. The storage volumes in experiments 3 and 4 far exceeded the magnitude of the sediment storage in both studies. Discrepancies in sediment storage may be attributable to differences in the time since wood recruitment, as well as stream size. Experiments 2 to 5 represent simu- lated timescales of five to ten years, while the data reported by Brooks et al. (2004) reflect only 2 years of post-recruitment sediment accumulation. Channel width must also be taken into consideration, as it limits the volume of sediment that can accumulate upstream of recruited wood in a given time period. When bankfull width is considered the stream table results, with sediment storage ranging from 0.32-0.79 m3/piece· cw−1, are similar to those reported by Megahan (1982). Local differences in sediment transport altered the channel morphology and produced an increasingly stepped profile with a lower median gradient, similar to the results of previous field studies (Nakamura and Swanson, 1993; Faustini and Jones 2003). Pool spacing also changed significantly following wood addition. Prior to wood addition, pool spacing varied from 1.5 to 4.4 times the bankful chan- nel width, which is equivalent to a pool frequency in the prototype system of 2.3 to 6.5 pools per 100 m. This pool spacing was similar to the range of pool spacings of 2 to 5.5 channel widths, as well as the frequencies of 3-13 pools per 100 m, pre- viously reported in streams with low wood loadings (Richmond and Fausch, 1995; Montgomery et al., 1995; Beechie and Sibley, 1997; Kreutzweiser et al., 2005). Following the addition of large wood, pool spacing decreased to between 1.1 107 and 2.6 channel widths, equivalent to a pool frequency of 4.4 to 8.8 pools per 100 m in the prototype system, while the percent of the reach length occupied by pools increased by up to 91%. Post-addition pool spacing was similar to the spacing of < 1 channel width reported by Montgomery et al. (1995) and Webb and Eskine (2005), as well as the spacing of 2-3 channel widths reported by Beechie and Sibley (1997) for reaches with moderate wood loadings. While the pool spacing at steady state was significantly related to all of the wood-related metrics, the change in pool spacing was not related to any of the metrics used in this analysis, suggesting that the pool frequency is controlled by wood, and is relatively insensitive to the intital state of the system. In accordance with previous studies, a large percentage of the pools were classified as log-affected at steady state following wood addition. Previous authors have reported values of 48-90% for field studies (Andrus et al., 1988; Montgomery et al., 1995; Richmond and Fausch, 1995; Beechie and Sibley, 1997; Webb and Erskine, 2003). These results suggest that a large proportion of the increase in total pool length in the reach resulted from the creation of relatively small, log-affected scour pools, rather than the extension of pre-existing pools. Morphologic changes were accompanied by adjustments in the surface texture of the reach. Surface grain size co-varied with the channel slope; increased het- erogeneity in the bed gradient produced significant increases in facies complexity as fine sediment deposited in lower gradient segments, and scour revelead coarser sediment where flow was concentrated. As a result, the number of facies increased by 63 to 86%, though the number of facies at steady state with wood (26-28) was fairly constant regardless of the volume of wood added. These results suggest that the addition of relatively small volumes of wood (0.00037 m3/m2, equivalent to 0.011 m3/m2 in the prototype) can produce dramatic increases in facies num- ber, while additional wood does not significantly alter facies complexity. Despite significant changes in facies complexity, decreases in surface grain size following wood addition were minor in experiments 3 to 5, while the increase in very coarse facies due to wood-related scour actually increased the median grain size in exper- iment 2. Changes in the D50, which ranged from 5-10% were small compared with the 22% decreases reported by Assani and Petit (1995). Increased mean bed gradient, rather than textural adjustment, seems to be the 108 primary mechanism through which steady state was re-attained following wood addition. The strong linear relationship between mean bed gradient and wood load parallels the relationship between wood load and water surface gradient described in chapter 4. It appears that progressive aggradation in the upstream end of the reach increased the potential energy in the system, allowing sediment transport capacity to recover despite sustained increases in flow resistance. As reported in previous field studies, wood removal caused a dramatic increase in sediment transport, and rapidly evacuated much of the sediment stored in the reach (Beschta, 1979; MacDonald and Keller, 1987; Smith et al., 1993; Gurnell et al., 2002). The rates of increase in sediment transport, which varied from 3.7-fold in experiment 3 to 4.7-fold in experiment 4 were similar in magnitude to the 4-fold increase reported by Smith et al. (1993). Despite these increases in sediment trans- port, however, the sediment stored following wood addition was not fully evacuated in experiment 3 or 4. These results suggest that the system maintains a ‘memory’ of its previous state. Further, the changes in sediment storage following wood re- moval were not adequately described by the exponential model developed by Lisle and Church (2002), which was based on experimental feed reduction, emphasizing the importance of other factors such as sediment supply on reach-scale storage. Sustained sediment storage may also have been attributable to bar preservation following wood removal. These results appear to support the findings of previous authors that bar development or preservation offsets the effects of resistance loss following wood removal (Heede, 1972; Smith et al., 1993; Lisle, 1995). Interest- ingly, at steady state following wood removal neither channel roughness (n) nor channel gradient had returned to their pre-addition steady state values, showing that multiple steady state energy balance configurations are possible in a single reach. As suggested from previous field studies, increased sediment transport con- tributed to the infilling of scour pools following wood removal, and altered the sedimentological characteristics of the reach (Beschta, 1979; Lisle, 1995; Gur- nell and Sweet, 1998). While pool length decreased, however, pool frequency did not change, suggesting the infilling was insufficient to completely eliminate log- affected scour pools. Similarly, while facies complexity decreased, the number of facies remained significantly greater than the pre-addition values. These results 109 highlight the hysteresis in the response of the system to wood removal; the effects of wood removal were more rapid than the effects of wood addition, and produced a significantly different reach morphology. 110 Chapter 6 Conclusions Despite four decades of research, the study of large wood remains immature; the domain has largely failed to progress beyond phenomenological studies and field experiments involving complete wood removal. The objective of the present study was to advance the discipline by investigating the effects of variations in large wood on channel hydraulics and bed morphology. This goal has been accomplished through five stream table experiments involving the addition of four different wood loads. The results presented in this thesis yield novel insight into the effects of large wood on channel processes, with important implications for future watershed management and stream restoration practices. 6.1 Large Wood Dynamics The addition of large wood to the experimental reach produced significant hy- draulic and morphologic changes, and largely re-produced natural in-stream wood dynamics. Following recruitment, wood transport and deposition resembled the patterns reported in previous field and model studies. In general, wood pieces re-oriented in positions which minimized the drag force acting on the piece, and maximized stability. The majority of this re-positioning occurred during the first run following recruitment, which is consistent with the field observations made by Sweka and Hartman (2006). The total travel distance was dependent on piece length, root wad presence, and original orientation of the wood pieces (replicating 111 the results reported by Hilderbrand et al., 1997; Braudrick et al., 1997; Braudrick and Grant, 2001; Gurnell et al., 2002). Numerous jams formed in the reach in experiments 3 to 5. Jam presence en- abled pieces to remain in otherwise unstable positions, and thereby exert a greater influence on channel hydraulics and morphology. While 81% of jam members were oriented perpendicular or obliquely to the flow direction at steady state, only 37% of pieces that were not part of a jam were similarly positioned. As a result, the blockage ratio of jam members – a measure of the hydraulic effectiveness of the pieces – was significantly higher than that of the individual pieces. While the average blockage ratio of the individual pieces was slightly below the threshold of hydraulic significance of 0.1 proposed by Gippel et al. (1992), the mean blockage ratio of the jam members was nearly three times the threshold. The process of jam formation followed the conceptual model of jam develop- ment proposed by Manners and Doyle (2008), with pieces progressively accumu- lating behind a large, stable, key member. Jam frequency was linearly related to the number of suspended pieces in the reach, further highlighting the importance of large, stable pieces in jam initiation. The positioning of individual pieces in the reach was also similar to previously reported results from natural systems (e.g. Hickin, 1984), with 76% of individual rootwad-containing pieces oriented parallel to the flow, with the rootwad upstream. The similarities between the processes of wood transport and deposition in these experiments and the natural processes reported from field studies suggest that the model reliably reproduced the natural dynamics of in-stream large wood. These results lend credibility to the realism of the model, and suggest that it presents a useful tool for assessing the impacts of in-stream large wood on channel hydraulics and morphology. 6.2 Channel Hydraulics and Morphology The addition of large wood induced immediate changes in channel hydraulics, as well as progressive changes in channel morphology. The addition of a relatively low volume of wood (equivalent to 0.011 m3/m2 in the prototype) caused an im- mediate and significant decrease in flow velocity, attributable to increased flow 112 resistance. The decreases in velocity were low, however, when compared to the magnitude of the increases reported in field studies involving wood removal (Mac- Donald and Keller, 1987; Mason et al., 1990; Shields and Smith, 1992; Erskine, 1994b; Reinfelds et al., 1995). While this discrepancy may be partially attributed to differences in measurement methods, as results from field studies are based on point measurements which reflect local velocities rather than true reach averages, it may also reflect the lower resistance provided by the modeled wood pieces, which did not contain the large branch networks often present in naturally recruited wood. The hydraulic changes induced by wood addition also affected the channel morphology. Similar to results from previous field studies, where flow was concen- trated localized scour produced numerous log-affected pools which significantly increased the pool frequency in the reach (Keller and Tally, 1979; Beechie and Sib- ley, 1997; Jackson and Sturm, 2002; Abbe and Montgomery, 2003; Faustini and Jones, 2003). The number of pools at steady state was most closely related to the average blockage ratio of the pieces, while the percent of the reach occupied by pools was most closely related to the percentage of pieces in jams. These results suggest that the most hydraulically effective pieces – which are often jam members – are the most effective agents of pool formation. Similar to previous field observations, decreased flow velocity upstream of ob- stacles induced localized sediment storage (Keller and Swanson, 1979; Mosley, 1981; Megahan, 1982; Thompson, 1995; Abbe and Montgomery, 2003; Brooks et al., 2004, 2006; Andreoli et al., 2007). Overall, the greater variability in sediment transport produced an increasingly stepped channel profile and the median slope of the channel bed decreased following wood addition. The grain size distribution of the surface material co-varied with the bed slope. Increasing heterogeneity in the channel slope – with increased gradients over log steps and decreased gradients upstream of obstructions – was accompanied by a divergence of surface grain size. The addition of large wood increased the proportion of fine facies in areas with low bed gradients, and increased the proportion of coarse facies in high-gradient seg- ments. Despite the increased heterogeneity in the surface texture, however, changes in the median grain size were relatively minor in these experiments compared with previous field experiments (Assani and Petit, 1995; Manga and Kirchner, 2000). The addition of large wood also decreased the potential energy available to 113 transport sediment in the reach. Changes in water surface slope, which increased by 0.8-29% in experiments 2 to 5, appear to be the primary mechanism through which this energy loss was recovered. The combination of greater flow depth and water surface slope increased the total shear stress at steady state following wood addition, and enabled the transport capacity to recover despite sustained decreases in flow velocity, and increases in flow resistance. The adjustment of water surface gradient was strongly related to the wood load, piece frequency, and total projected area of the wood, suggesting that the dominant channel response to wood addition may be reliably predicted for this system. Further, it suggests a possible avenue for incorporating the effects of large wood on channel dynamics in analytical models. Wood removal had the opposite effects on channel processes, causing decreases in flow resistance while increasing the flow velocity and sediment transport rate. While the reach attained a new steady state relatively rapidly, the system did not recover to the pre-addition steady state; flow resistance remained above the pre- addition values, and velocity remained below. It appears that a new steady state sediment transport rate was possible, despite continued decreases in flow velocity, due to sustained increases in water surface slope, which also did not return to pre- addition values. Similar to previously reported results, changes in channel morphology were dramatic following wood removal, with rapid evacuation of stored sediment oc- curring immediately (Beschta, 1979; Smith et al., 1993; Lisle, 1995; Gurnell and Sweet, 1998). As with the hydraulic response, however, sediment storage remained above the pre-addition values following removal, and decreases in storage volume were significantly lower than predicted by the model of Lisle and Church (2002). Several of the log-induced pools also remained, sustaining the increases in pool frequency associated with large wood. These results support the assertion that channel morphology may be sustained following wood removal by increased bar formation and associated flow resistance, or in this case bar preservation (Heede, 1972; Heede, 1985; Lisle, 1995). 114 6.3 Implications The results of these stream table experiments have important implications for wa- tershed management. The hydraulic and morphologic changes induced by the addi- tion of large wood increased potential fish habitat. The increase in pool frequency, which occurred in all four experiments in which wood was added to the reach, created areas of potential refuge for fishes from high velocities at high flows, and would likely provide protection from drought at lower flows. Similar to previous results from field studies, the increased availability of coarse and very coarse fa- cies following wood addition increased the availability of spawning habitat for salmonids (Bisson et al., 1987; Floyd et al., 2009; Nagayama and Nakamura, 2010). Overall, the increased diversity of the bed morphology and flow velocity – evidenced by greater variance in cross-sectional depth as well as greater facies complexity – increased the quality of the aquatic habitat. The favourable changes in the hydraulic and morphologic parameters were pos- itively related to the volume and number of pieces of wood added to the system, supporting previous suggestions that stream restoration attempts benefit from max- imizing the volume of wood added to the reach (Sweka and Hartman, 2006; Na- gayama and Nakamura, 2010). Interestingly, however, the increase in facies com- plexity was insensitive to the volume of wood added, suggesting that the addition of even a small volume of wood can dramatically increase the variability in the surface texture, and possibly increase spawning habitat. The dynamics of large wood and the processes of jam formation also have important implications for stream restoration. The results of these experiments suggest that jam formation, which increases the hydraulic effectiveness of the added pieces, requires the addition of large, intact pieces, containing root wads and branches, as suggested by previous authors (e.g. Kail et al., 2007; Nagayama and Nakamura, 2010). Further, the development of numerous jams with a spacing sim- ilar to that observed in natural systems suggests that less costly ‘soft engineering’ techniques, involving the installation of un-fixed wood, are effective at simulating natural patterns of wood deposition. Important conclusions must also be drawn from the hysteresis in the response to wood addition and removal. While changes in the channel morphology following 115 wood addition occurred over a simulated period of 5 to 10 years, the morphologic adjustment following wood removal generally occurred during a single large flow event. Further, the channel morphology at steady state was not the same following wood removal as it had been prior to addition, suggesting a morphologic memory in the system. These results have implications for future research, as the majority of previ- ous experiments studies involve the removal of large wood from stream reaches (e.g. Beschta, 1979; Heede, 1985; Smith et al., 1993; Lisle, 1995; Gurnell and Sweet, 1998). The hysteresis in the length of the response to addition and removal suggest that morphologic change following wood addition occurs over a signifi- cantly longer period than is suggested by these previous studies, and that moni- toring should continue for at least 10 years following wood addition. Meanwhile, the hysteresis in the steady state bed morphology suggests that changes in the mor- phologic characteristics of the bed following wood addition can not be reliably predicted using results from previous field studies. 6.4 Future Research Field studies involving wood removal provide a poor analogue for the expected changes in channel morphology following stream restoration projects, as well as natural disturbances which increase wood recruitment. To account for the hys- teresis in stream response to wood manipulation, informed decisions regarding the impacts of large wood installation require further research into the effects of wood addition on channel dynamics. Given that wood addition is a common tool for stream restoration in North America (Bernhardt et al., 2005; Kail et al., 2007; Nagayama and Nakamura, 2010), improved monitoring and reporting of existing restoration projects is one way in which to improve our current understanding of these effects. The realism of the wood dynamics reproduced in the stream table model sug- gest that laboratory experiments may also provide a practical tool for further in- vestigation of the effects of large wood on channel processes. Previous flume stud- ies, however, have focused primarily on wood entrainment and transport processes (Braudrick et al., 1997; Braudrick and Grant, 2001; Bocchiola et al., 2006) or the 116 effects of individual wood pieces on local morphology (Cherry and Beschta, 2001), and yield little insight into the broader changes in channel morphology which ac- company reach- or watershed-scale alterations in wood load. The current results show that natural reach-scale wood processes can be reliably modeled in a labora- tory experiment. The strong relationships between the wood load, piece frequency, and projected area of the wood and many of the hydraulic and morphologic vari- ables also suggest that analytical models may provide a viable avenue for future experimentation. 117 Bibliography Abbe, T.B., Montgomery, D.R., 2003. Patterns and processes of wood debris accumu- lation in the Queets river basin, Washington. Geomorphology 51, 81-107. Andreoli, A., Comiti, F., Lenzi, M.A., 2007. Characteristics, distribution and geomor- phic role of large woody debris in a mountain stream of the Chilean Andes. Earth Surface Processes and Landforms 32, 1675-1692. Andrews, C.A.E., 2010. A stream in transition: short term morphodynamics of Fish- trap Creek following wildfire. Masters thesis, University of British Columbia. Andrus, C.W., Long, B.A., Froehlich, H.A., 1988. Woody debris and its contribution to pool formation in a coastal stream 50 years after logging. Canadian Journal of Fisheries and Aquatic Sciences 45 (12), 2080-2086. Assani, A.A., Petit, F., 1995. Log-jam effects on bed-load mobility from experiments conducted in a small gravel-bed forest ditch. Catena 25, 117-126. Bash, J.S., Ryan, C.M., 2002. Stream restoration and enhancement projects: is anyone monitoring? Environmental Management 29, 877-885. Beechie, T.J., Liermann, M., Pollock, M.M., Baker, S., Davies, J., 2006. Channel pattern and river-floodplain dynamics in forested mountain river systems. Geo- morphology 78 (1-2), 124-141. Beechie, T.J., Sibley, T.H., 1997. Relationships between channel characteristics, woody debris, and fish habitat in Northwestern Washington. Transactions of the Ameri- can Fisheries Society 126, 217-229. Beechie , T.J., Sear, D.A., Olden, J.D., Pess, G.R., Buffington, J.M., Moir, H., Roni, P., Pollock, M.M., 2010. Process-based principles for restoring river ecosystems. Bioscience 60, 209-222. Benda, L.E., Sias, J.C., 2003. A quantitative framework for evaluating the mass bal- ance of in-stream organic debris. Forest Ecology and Management 172, 1-16. Bernhardt, E.S., Palmer, M.A., Allan, J.D., Alexander, G.,Barnas, K., Brooks, S., Carr, J., Clayton, S., Dahm, C., Follstad-Shah, J., Galat, D., Gloss, S., Goodwin, P., Hart, D., Hassett, B., Jenkinson, R., Katz, S., Kondolf, G.M., Lake, P.S., Lave, R., Meyer, J.L., O’Donnell, T.K., Pagano, L., Powell, B., Sudduth, E., 2005. Synthesizing US river restoration efforts. Science 308, 636637. Beschta, R.L., 1979. Debris removal and its effects on sedimentation in an Oregon Coast range stream. Northwest Science 53 (1), 71-78. Bilby, R.E., 1981. Role of organic debris dams in regulating the export of dissolved and particulate matter from a forested watershed. Ecology 62 (5), 1234-1243. Bilby, R.E., Likens, G.E., 1980. Importance of organic debris dams in the structure and function of stream ecosystems. Ecology 61 (5), 1107-1113. Bilby, R.E., Ward, J.W., 1991. Characteristics and function of large woody debris 118 in streams draining old-growth, clear-cut, and second-growth forests in South- western Washington. Canadian Journal of Fisheries and Aquatic Science 48 (12), 2499-2508. Bisson, P.A., Bilby, R.E., Bryant, M.D., Dollogg, C.A., Grette, G.B., House, R.A., Murphy, M.L., Koski, K.V., Sedell, J.R., 1987. Large woody debris in forested streams in the Pacific Northwest, Past, present, and future, in Salo, E.O. and Cundy, T.W. (Eds), Streamside Management, Forestry and Fisheries Interactions, Institute of Forest Resources, Contribution 57, University of Washington, Seattle, Washington, 143-190. Bocchiola, D., Rulli, M.C., Rosso, R., 2006. Flume experiments on wood entrainment in rivers. Advances in Water Resources 29, 1182-1195. Bocchiola, D., Rulli, M.C., Rosso, R., 2008. A flume experiment on the formation of wood jams in rivers. Water Resources Research 44, doi:10.1029/2006WR005846 Bragg, D., 2000. Simulating catastrophic and individualistic large woody debris re- cruitment for a small riparian system. Ecology 81(5), 1383-1394. Braudrick, C.A., Grant, G.E., Ishikawa, Y., Ikeda, H., 1997. Dynamics of wood trans- port in streams: a flume experiment. Earth Surface Processes and Landforms 22, 669-683. Braudrick, C.A., Grant, G.E., 2001. Transport and deposition of large woody debris in streams: a flume experiment. Geomorphology 41 (4), 263-283. Brooks, A.P., Gehrke, P.C., Jansen, J.D., Abbe, T.B., 2004. Experimental reintroduc- tion of woody debris on the Williams River, NSW: geomorphic and ecological responses. River Research and Applications 20, 513-536. Brooks, A.P., Howell, T., Abbe, T.B., Arthington, A.H., 2006. Confronting hysteresis: wood based river rehabilitation in highly altered riverine landscapes of south- eastern Australia. Geomorphology 79, 395-422. Buffington, J., Montgomery, D., 1999b. Effects of hydraulic roughness on surface textures of gravel-bed rivers. Water Resources Research 35(11), 3507-3521. Cadol, D., Wohl, E., Goode, J.R., Jaeger, K.L., 2009. Wood distribution in neotropical forested headwater streams of La Selva, Costa Rica. Earth Surface Processes and Landforms 34, 1198-1215. Cedarholm, C.J., Bilby, R.E., Bisson, P.A., Bumstead, T.W., Fransen, B.R., Scarlett, W.J., Ward, J.W., 1997. Response of juvenile coho salmon and steelhead to place- ment of large woody debris in a coastal Washington stream. North American Journal of Fisheries Management 17, 947-963. Cherry, J., Beschta, R.L., 2001. Coarse woody debris and channel morphology, a flume study. Journal of the American Water Resources Association 25(5), 1031- 1036. Church, M., 1992. Channel morphology and typology. The rivers handbook 1, 126- 143. 119 Crispin, V., House, R., Roberts, D., 1993. Changes in instream habitat after the re- structuring of a coastal Oregon stream. North American Journal of Fisheries Management 13, 96-102. Czarnomski, N.M., Dreher, D.M., Snyder, K.U., Jones, J.A., Swanson, F.J., 2008. Dynamics of wood in stream networks of the western Cascades Range, Oregon. Canadian Journal of Forest Resources 38, 2236-2248. Dahlstrom, N., Nilsson, C., 2004. Influence of woody debris on channel structure in old growth and managed forest streams in central Sweden. Environmental Management 33 (3), 376-384. Eaton, B.C., Giles, T., 2009. Assessing the effect of vegetation-related bank strength on channel morphology and stability in gravel-bed streams using numerical mod- els. Earth Surface Processes and Landforms 34 (5), 712-724. Eaton, B.C., Moore, R.D., Giles, T.R., 2010. Forest fire, bank strength and channel instability the “unusual” response of Fishtrap Creek, British Columbia. Earth Surface Processes and Landforms 35 (1), 1167-1183. Erskine, W.D., 1994b. River response to accelerated soil erosion in the Glenelg River catchment, Victoria. Australian Journal of Soil and Water Conservation 7 (2), 39-47. Erskine, W.D., Webb, A.A., 2003. Desnagging to resnagging: new directions in river rehabilitation in southeastern Australia. River Research and Applications 19, 233-249. Fausch, K.D., Northcote, T.G., 1992. Large woody debris and salmonid habitat in a small coastal British Columbia stream. Canadian Journal of Fisheries and Aquatic Sciences 49, 682-693. Faustini, J.M., Jones, J.A., 2003. Influence of large woody debris on channel morphol- ogy and dynamics in steep, boulder-rich mountain streams, western Cascades, Oregon. Geomorphology 51, 187-205. Fetherston, K.L., Naiman, R.J., Bilby, R.E., 1995. Large woody debris, physical pro- cess, and riparian forest development in montane river networks of the Pacific Northwest. Geomorphology 13, 133-144. Floyd, T.A., Macinnis, C., Taylor, B.R., 2009. Effects of artificial woody structures on Atlantic salmon habitat and populations in a Nova Scotia stream. River Research and Applications 25, 272-282. Gippel, C.J, O’Neill, I.C., Finlayson, B.L., 1992. The hydraulic basis of snag man- agement. Land and Water Resources Research and Development Canberra, Aus- tralia. Gippel, C.J., 1995. Environmental hydraulics of large woody debris in streams and rivers. Journal of Environmental Engineering 5, 288-295. Gippel, C.J, O’Neill, I.C., Finlayson, B.L., Schnatz, I., 1996. Hydrdaulic guidelines for the re-introduction and management of large woody debris in lowland rivers. 120 Regulated Rivers: Research and Management 12 (2-3), 223-236. Gurnell, A.M., Sweet, R., 1998. The distribution of large woody debris accumulations and pools in relation to woodland stream management in a small, low gradient stream. Earth Surface Processes and Landforms 23, 1101-1121. Gurnell, A.M., Piegay, H., Swanson, F.J., Gregory, S.V., 2002. Large wood and fluvial processes. Freshwater Biology 47, 601-619. Hakala, J.P., Hartman, K.J., 2004. Drought effect on streammorphology and brook trout (Salvelinus fontinalis) pop- ulations in forested headwater streams. Hydrobiologia 515 (1-3), 203-213. Hassan, M.A., Church, M., Lisle, T.E., Brardinoni, F., Benda, L., Grant, G.E., 2005. Sediment transport and channel morphology of small, forested streams. Journal of the American Water Resource Association 41 (4), 853-876. Hassan, M.A., Hogan, D.L., Bird, S.A., May, C.L., Gomi, T., Campbell, D., 2005. Spatial and temporal dynamics of wood in headwater streams of the Pacific North- west. Journal of the American Water Resources Association 41 (4), 877-902. Heede, B.H., 1972. Influences of a forest on the hydraulic geometry of two mountain streams. Water Resources Bulletin 8 (3), 523-530. Heede, B.H., 1985. Channel adjustments to the removal of log steps, and experiment in a mountain stream. Environmental Management 9 (5), 427-432. Hickin, E., 1984. Vegetation and river channel dynamics. Canadian Geographer 28 (2), 111-126. Hilderbrand, R.H., Lemly, A.D., Dolloff, C.A., Harpster, K.L., 1997. Effects of large woody debris placement on stream channels and benthic macroinverte- brates. Canadian Journal of Fisheries and Aquatic Sciences 54, 931-939. Jackson, C.R., Sturm, C.A., 2002. Woody debris and channel morphology in first- and second-order forested channels in Washingtons coast ranges. Water Resources Research 38 (9), doi:10.1029/2001WR001138 Johnson, S.E., Rodgers, J.D., Solazzi, M.F., Nickelson, T.E., 2005. Effects of an increase in large wood on abundance and survival of juvenile salmonids (On- corhynchus spp.) in an Oregon coastal stream. Canadian Journal of Fisheries and Aquatic Sciences 62, 412-424. Kail , J., Hering, D., Muhar, S., Gerhard, M., Preis, S., 2007. The use of large wood in stream restoration: experiences from 50 projects in Germany and Austria. Journal of Applied Ecology 44, 1145-1155. Keller, E.A., Swanson, F.J., 1979. Effects of large organic material on channel form and fluvial processes. Earth Surface Processes 4, 361-380. Keller, E.A., Tally, T., 1979. Effects of large organic debris on channel form and fluvial processes in the coastal redwood environment, in Rhodes, D.D. and Williams, G.G. (eds) Adjustments of the Fluvial System, Santa Barbara, California, 169- 195. 121 Kreutzweiser, D.P., Good, K.P., Sutton, T.M., 2005. Large woody debris character- istics and contributions to pool formation in forest streams of the Boreal Shield. Canadian Journal of Forest Research 35 (5), 1213-1223. Lester, R.E., Boulton, A.J., 2008. Rehabilitating agricultural streams in Australia with wood: a review. Environmental Management 42, 310-326. Lienkaemper, G.W., Swanson, F.J., 1987. Dynamics of large woody debris in streams in old-growth Douglas-fir forests. Canadian Journal of Forest Resources 17 (2), 150-156. Lisle, T.E., 1987. Using “residual depths to monitor pool depths independently of discharge. Res. Note PSW-394. Berkeley, CA. Pacific Southwest Forest and Range Experiment Station, Forest Service, U.S. Department of Agriculture. Lisle, T.E., 1995. Effects of coarse woody debris an its removal on a channel affected by the 1980 eruption of Mt. St. Helens, Washington. Water Resources Research 31 (7), 1797-1808. Lisle, T.E., Church, M., 2002. Sediment transport-storage relations for degrading, gravel bed channels. Water Resources Research 38 (11), doi:10.1029/2001WR001086 MacDonald, A., Keller, E.A., 1987. Stream channel response to the removal of large woody debris, Larry Damm Creek, northwestern California, in Erosion and Sedi- mentation in the Pacific Rim, IAHS Publication 165, Washington, D.C., 405-406. Manga, M., Kirchner, J.W., 2000. Shear stress partitioning in streams by large woody debris. Water Resources Research 36, 2373-2380. Manners, R.B., Doyle, M.W., Small, M.J., 2007. Structure and hydraulics of natural woody debris jams. Water Resources Research 43, doi:10.1029/2006WR004910 Manners, R.B., Doyle, M.W., 2008. A mechanistic model of woody debris jam evo- lution and its application to wood-based restoration and management. River Re- search and Applications 24, 1104-1123. Marsh, N., Rutherford, I., Jerie, K., 2001. Predicting pre-disturbance loading and distribution of large woody debris, in Rutherford, I., Sheldon, F., Brierley, G., Kenyon, C. (Eds) Proceedings of the Third Australian Stream Management Con- ference, Brisbane: CSIRO Publications, 391-395. Mason, R.R., Simmons, C.E., Watkins, S.A., 1990. Effects of channel modifications on the hydrology of Chicod Creek Basin, North Carolina, 1975-1987, in US Ge- ological Survey Water Resource Investigations Report 90-4031, Raleigh, N.C. Megahan, W.F., 1982. Channel sediment storage behind obstructions in forested drainage basins draining the granitic bedrock of the Idaho Batholith, in Swanson, F.J. et al. (Eds), Sediment Budgets and Routing in Forested Drainage Basins, USDA Forest Service, General Technical Report, PNW-GTR-141, Pacific North- west and Range Experiment Station, 114-121. Montgomery, D.R., Buffington, J.M., 1997. Channel-reach morphology in mountain drainage basins. Geological Society of America Bulletin, 596-611. 122 Montgomery, D., Buffington, J., Smith, R., Schmidt, K., Pess, G., 1995. Pool spacing in forest channels. Water Resources Research 31 (4), 1097-1105. Moore, R.D., 2003. Introduction to salt dilution gauging for streamflow measurement: part 1. Streamline 7(4), 20-23. Mosley, P.M., 1981. The influence of organic debris on channel morphology and bedload transport in a New Zealand forest stream. Earth Surface Processes and Landforms 6, 571-579. Nagamaya, S., Nakamura, F., 2010. Fish habitat rehabilitation using wood in the world. Landscape and Ecological Engineering 6, 289-305. Naiman, R.J., Balian, E.V., Bartz, K.K, Bilby, R.E., Latterell, J.J., 2002. Dead wood dynamics in stream ecosystems, in USDA Forest Service General Technical Re- port, PSW-GTR-181. Nakamura, F, Swanson, F.J., 1993. Effects of course woody debris on morphology and sediment storage of a mountain stream system in Western Oregon. Earth Surface Processes and Landforms 18, 43-61. Petts, G., Foster, I., 1985. Rivers and landscape. Edward Arnold, London, 274 pp. Piegay, H., 2003. Dynamics of wood in large rivers, in Gregory, S.V., Boyer, K.L., Gurnell, A.M. (Eds), The ecology and management of wood in world rivers, Symposium 37, American Fisheries Society, Bethesda, 109-133. Reinfelds, I., Rutherford, I., Bishop, P., 1995. History and effects of channelization on the Latrobe River, Victoria. Australian Geophysical Studies 33, 60-76. Richmond, A., Fausch, K., 1995. Characteristics and function of large woody debris in subalpine Rocky Mountain streams in northern Colorado. Canadian Journal of Fisheries and Aquatic Sciences 52 (8), 1789-1802. Robison, E., Beschta, R., 1990. Characteristics of coarse woody debris for several coastal streams of southeast Alaska, USA. Canadian Journal of Fisheries and Aquatic Sciences 47 (9), 1684-1693. Roni, P., Quinn, T., 2001. Density and size of juvenile salmonids in response to place- ment of large woody debris in western Oregon and Washington streams. Cana- dian Journal of Fisheries and Aquatic Sciences 58 (2), 282-292. Shields, F.D., Gippel, C.J., 1995. Prediction of effects of woody debris removal on flow resistance. Journal of Hydraulic Engineering 121 (4), 341-354. Shields, F.D., Smith, R.H., 1992. Effects of large woody debris removal on physical characteristics of a sand-bed river. Aquatic Conservation 2 (2), 145-163. Smith, R.D., Sidle, R.C., Porter, P.E., 1993. Effects on bedload transport on experi- mental removal of woody debris from a forest gravel-bed stream. Earth Surface Processes and Landforms 18, 455-468. Solazzi , M.F., Nickelson, T.E., Johnson, S.L., Rodgers, J.D., 2000. Effects of in- creasing winter rearing habitat on abundance of salmonids in two coastal Oregon streams. Canadian Journal of Fisheries and Aquatic Sciences 57, 906-914. 123 Sweka, J.A., Hartman, K.J., 2006. Effects of large woody debris addition on stream habitat and brook trout populations in Appalachian streams. Hydrobiologia 559, 363-378. Thompson, D.M., 1995. The effects of large organic debris on sediment processes and stream morphology in Vermont. Geomorphology 11, 235-244. Waldon, M.G., 2004. Estimation of average stream velocity. Journal of Hydraulic Engineering 130 (11), 1119-1122. Warren, D.R., Kraft, C.E., 2003. Brook trout (Salvelinus fontinalis) response to wood removal from high-gradient streams of the AdirondackMountains (N.Y., U.S.A.). Canadian Journal of Fisheries and Aquatic Sciences 60, 379-389. Webb, A.A., Erskine, W.D., 2003. Distribution, recruitment, and geomorphic signif- icance of large woody debris in an alluvial forest stream: Tonghi Creek, south- eastern Australia. Geomorphology 51 (1-3), 109-126. Wohl, E., Jaeger, K., 2009. A conceptual model for the longitudinal distribution of wood in mountain streams. Earth Surface Processes and Landforms 34 (3), 329- 344. Young, W.J., 1991. Flume study of the hydraulic effects of large woody debris in lowland rivers. Regulated Rivers, Research and Management 6 (3), 203-211. Zimmerman, A.E., 2009. Experimental investigations of step-pool channel formation and stability. Doctoral dissertation, University of British Columbia. 124

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

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"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0072311/manifest

Comment

Related Items