UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

A stream in transition : short term morphodynamics of Fishtrap Creek following wildfire Andrews, Christie 2010

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

Item Metadata

Download

Media
24-ubc_2010_spring_andrews_christie.pdf [ 6.32MB ]
Metadata
JSON: 24-1.0069407.json
JSON-LD: 24-1.0069407-ld.json
RDF/XML (Pretty): 24-1.0069407-rdf.xml
RDF/JSON: 24-1.0069407-rdf.json
Turtle: 24-1.0069407-turtle.txt
N-Triples: 24-1.0069407-rdf-ntriples.txt
Original Record: 24-1.0069407-source.json
Full Text
24-1.0069407-fulltext.txt
Citation
24-1.0069407.ris

Full Text

A Stream in Transition: Short Term Morphodynamics of Fishtrap Creek Following Wildfire by Christie Andrews B.Sc., The University of British Columbia, 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) April 2010 c￿ Christie Andrews 2010 Abstract In August 2003, a wildfire burned through Fishtrap Creek Watershed north of Kamloops British Columbia. This high intensity fire killed almost all the trees within the burned area, including 90% of the riparian vegetation in the vicinity of our study site. The fire did not significantly alter the duration or magnitude of the peak flows within this creek, nor did it have substantial effects on the total suspended sediment concentrations. Changes in channel morphology during the first two years after the fire were minor. The first evidence of morphologic adjustment occurred in 2006 when the channel began to widen and develop very distinct channel bars - adopting a characteristically riffle pool morphology by the end of the 2006 freshet. The most dramatic channel reconfiguration occurred during the 2007 freshet, when the channel widened by as much as 15 m in places. Approximately 82% of the total volume of large wood (LW) recruited to the channel following the fire entered the channel as a result of bank erosion, and the majority of the bank erosion occurred during the 2007 flood season. The post-fire wood load in Fishtrap Creek is slightly higher than other disturbed systems, but is comparable to wood load in undisturbed rivers. LW has had significant influence on channel morphology and bed surface texture distribution. The number of LW pieces of wood in the channel is related to the channel morphology, and 80% of the pools are a result of LW. Most of the post-fire wood is suspended above the channel bed and is not currently functioning in the channel as effectively as pre-fire wood. Estimates of net erosion and deposition were made based on Digital Elevation Models (DEMs) and cross-sections located at regular intervals: a comparison of the two methods shows that cross-sectional analysis results in biased estimates of net erosion and deposition in various, identifiable circumstances, while revealing the same general pattern of channel change within the study reach. ii Table of Contents Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii Table of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vi List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix 1 Wildfire and Channel Morphodynamics . . . . . . . . . . . . . . . . . 1 1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.2 Controls on Channel Morphology and Stability . . . . . . . . . . . . . . 3 1.2.1 Sediment Supply and Transport . . . . . . . . . . . . . . . . . . 4 1.2.2 Riparian Vegetation and Large Wood . . . . . . . . . . . . . . . 8 1.3 Watershed Response Following Wildfire . . . . . . . . . . . . . . . . . . 9 1.4 Wildfire Case Studies Documenting Channel Response and LW Loading 11 1.5 Objectives and Research Questions . . . . . . . . . . . . . . . . . . . . 15 1.6 Thesis Organization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 2 Study Area Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 2.1 Geographic Location of Fishtrap Creek . . . . . . . . . . . . . . . . . . 17 2.2 Physiography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 2.3 Glacial History and Surficial Deposits . . . . . . . . . . . . . . . . . . . 19 2.4 Climate and Hydrology . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 2.5 Vegetation and Disturbance Regime . . . . . . . . . . . . . . . . . . . . 27 3 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 3.1 Study Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 3.2 Data Collection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 3.2.1 Cross-Section and Planimetric Survey . . . . . . . . . . . . . . . 30 3.2.2 Grain Size Distribution and Facies Map . . . . . . . . . . . . . . 32 iii Table of Contents 3.2.3 Subsurface Sampling and Analysis . . . . . . . . . . . . . . . . . 36 3.2.4 LW Survey and Tagging: 2004-2009 . . . . . . . . . . . . . . . . 36 4 Channel Morphodynamics . . . . . . . . . . . . . . . . . . . . . . . . . . 39 4.1 Previous Studies: Post-Fire Flows and Suspended Sediment . . . . . . . 39 4.2 Cross-Sectional Changes . . . . . . . . . . . . . . . . . . . . . . . . . . 42 4.3 Bed Material Tracer Particle Dynamics . . . . . . . . . . . . . . . . . . 47 4.4 Bank Location Changes . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 4.5 Changes in Bed Elevation, Slope and Creation of Secondary Channels . 55 4.6 Sedimentologic Response . . . . . . . . . . . . . . . . . . . . . . . . . . 62 4.7 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 5 Instream Large Wood . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66 5.1 Volumetric Estimation of LW and Error . . . . . . . . . . . . . . . . . . 66 5.2 Longitudinal Influence of LW . . . . . . . . . . . . . . . . . . . . . . . . 69 5.3 LW Recruitment and Movement . . . . . . . . . . . . . . . . . . . . . . 72 5.4 Hydraulic Influence of LW . . . . . . . . . . . . . . . . . . . . . . . . . 79 5.5 Large Wood Debris Jams . . . . . . . . . . . . . . . . . . . . . . . . . . 85 5.5.1 Accumulation Rate of LW in Jams . . . . . . . . . . . . . . . . . 85 5.5.2 Break-up and Movement of Jams . . . . . . . . . . . . . . . . . 89 5.6 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89 6 Bed Material Sediment Transport . . . . . . . . . . . . . . . . . . . . . 94 6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94 6.2 Volumetric Estimation of Erosion and Deposition . . . . . . . . . . . . 95 6.3 Evaluation of Error Using the DEM Method . . . . . . . . . . . . . . . 97 6.4 Volumetric Net Change . . . . . . . . . . . . . . . . . . . . . . . . . . . 101 6.5 Bed Material Transport Rates . . . . . . . . . . . . . . . . . . . . . . . 107 6.6 Surface Response and Morphologic Adjustment . . . . . . . . . . . . . 110 6.7 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113 6.7.1 Topographic Change, Transport Rate and Facies Distribution . 114 6.7.2 Comparing DEM and XS Method . . . . . . . . . . . . . . . . . 115 7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117 7.1 Short Term Response to Wildfire at Fishtrap Creek . . . . . . . . . . . 117 7.2 Influence of LW and Sediment Supply on Aquatic Habitat . . . . . . . 119 7.3 Management Implications and Recommendations . . . . . . . . . . . . 121 Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124 iv Table of Contents Appendices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141 Appendix A: List of Incorporated Work . . . . . . . . . . . . . . . . . . . 142 v List of Tables 1.1 Sediment Supply, Discharge Changes and LW Recruitment Following Wildfire, Prescribed Fire and Model Case Study Results . . . . . . . . . 14 2.1 Peak Discharge Records from WSC Gauging Station at Fishtrap Creek, Kamloops B.C. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 3.1 LW Diameter and Length Classes Used for Field Survey . . . . . . . . . 37 4.1 Bar Amplitudes for XS 1 to XS 11 in 2004, 2005, and 2006 . . . . . . . 47 5.1 Length and Diameter Classes and Representative Length and Diameter of Large Wood . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 5.2 Wood Load Results From Studies on Similar Sized Streams in Disturbed and Undisturbed Watersheds, Including Fishtrap Creek. . . . . . . . . . 78 5.3 Fishtrap Creek LW Old and New Jam Characteristics and Jam Types (From Abbe and Montgomery, 2003). . . . . . . . . . . . . . . . . . . . 86 6.1 Error Analysis for Three Different Surface Interpolation Methods: Linear, Biharmonic Spline and Nearest Neighbour . . . . . . . . . . . . . . . . . 98 6.2 Estimated Net Volumes of Erosion and Deposition for Sub-Reach A-D Using the XS and DEM Method for the 2007 and 2008 Freshets. . . . . 102 6.3 Estimated Net Change in Sediment Storage for Sub-Reach A-D Using the XS and DEM Method for the 2007 and 2008 Freshets. . . . . . . . . 102 vi List of Figures 1.1 Fire Extent and Watershed Delineation . . . . . . . . . . . . . . . . . . 2 2.1 Fishtrap Creek Study Site Longitudinal Profile . . . . . . . . . . . . . . 17 2.2 Fishtrap Creek Morphologic Map . . . . . . . . . . . . . . . . . . . . . . 18 2.3 Fishtrap Hypsometric Curve . . . . . . . . . . . . . . . . . . . . . . . . . 20 2.4 Fishtrap and Skull Creek Surficial Deposits and Terrain Mapping . . . . 21 2.5 Lidar Image Showing Surficial Deposits . . . . . . . . . . . . . . . . . . 22 2.6 Fishtrap and Skull Creek Longitudinal Profile . . . . . . . . . . . . . . . 23 2.7 Fishtrap Creek Hydrograph, Monthly Precipitation and Temperature . . 25 3.1 Fishtrap Creek Study Site Planimetric Profile . . . . . . . . . . . . . . . 29 3.2 Digital Gravelometer Data Process and Output . . . . . . . . . . . . . . 34 3.3 LW Orientation Classes . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 4.1 Hydrograph Showing Pre and Post-Fire Flows . . . . . . . . . . . . . . . 41 4.2 Estimates of Net Erosion and Net Deposition at all Cross-Sections . . . 44 4.3 Estimates of Net Change in Stored Sediment Based on Cross-Sectional Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 4.4 Cross-Sectional Profiles for Cross-Sections F , G, 9, and 13 . . . . . . . . 46 4.5 Cross-Sectional Profiles for Cross-Sections 1, 2, 17, and 19. . . . . . . . 48 4.6 Bed Material Tracer Path Length Distributions Grouped by Size Class . 49 4.7 Bed Material Tracer Path Length Distributions Grouped by Launch Lo- cation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 4.8 Depth-of-Burial Distributions for Bed Material Tracers Grouped by Launch Location . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 4.9 Planimetric Map of Bank Erosion Between 2006 - 2008 . . . . . . . . . . 54 4.10 Surface Elevation Differences Between 2006 - 2008 . . . . . . . . . . . . 56 4.11 Upper-reach Degradation at Cross-Section 17 . . . . . . . . . . . . . . . 57 4.12 Mid-Channel Aggradation at Cross-Section 13 . . . . . . . . . . . . . . . 58 4.13 Cross-Section 9 Log Jam Induced Aggradation. . . . . . . . . . . . . . . 59 4.14 Banks and Bar Locations in 2006, 2007 and 2008 . . . . . . . . . . . . . 60 vii List of Figures 4.15 Sub-Reach Slope Changes from 2006 - 2008 . . . . . . . . . . . . . . . . 62 4.16 Surface GSD Comparisons . . . . . . . . . . . . . . . . . . . . . . . . . . 63 5.1 Fishtrap Creek Study Site Showing LW Segment Locations and LW from 2007-2009 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 5.2 Longitudinal Profile Showing LW Jam Locations and Pool Forcing, Dis- tribution of Wood Load and LW Orientation . . . . . . . . . . . . . . . 70 5.3 Volume of LW Recruitment. . . . . . . . . . . . . . . . . . . . . . . . . . 72 5.4 Volume of LW Pre and Post Wildfire. . . . . . . . . . . . . . . . . . . . 74 5.5 Boxplot of Observed Wood Load vs. Predicted Wood Load by Bragg et al. (2000) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76 5.6 LW Blockage Ratios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80 5.7 Facies Map Categorized by Median Surface Grain Size . . . . . . . . . . 82 5.8 Relationship of Surface Texture Variability and Total LW . . . . . . . . 84 5.9 Jam Breakup and LW Movement . . . . . . . . . . . . . . . . . . . . . . 88 5.10 Root Decay and Vegetation Re-growth . . . . . . . . . . . . . . . . . . . 91 6.1 Map Showing Sub-Reach and DEM Segment Locations . . . . . . . . . . 96 6.2 Surface Interpolation Methods . . . . . . . . . . . . . . . . . . . . . . . 100 6.3 Estimates of Net Change WIthin the Study Reach. . . . . . . . . . . . . 103 6.4 2007 and 2008 Net Change per Bankfull Width . . . . . . . . . . . . . . 104 6.5 Elevation and Bank Changes Near Jam 2 (at XS 9) in 2007 and 2008 . . 106 6.6 Estimates of Bed Material Transport Rates for the 2007 and 2008 Freshet.108 6.7 Upper Reach Sub-Surface Grain Size Distribution . . . . . . . . . . . . . 111 viii Acknowledgements Accomplishment of this work could not have been achieved without the unending guid- ance, support and encouragement by my supervisor, Dr. Brett Eaton. His wide breadth of knowledge and enthusiasm for the project provided me with the strength and perse- verance towards the completion of this project. He has helped me to strive for higher goals and encouraged me to meet challenges that I had never before thought possi- ble. Dr. Dan Moore’s comments, suggestions and ideas are also highly recognized as a critical asset to this project and are greatly appreciated. The field component of this study was conducted with the help of my field assistant, Mike Van der Laan. Mike spent two summers bending over backwards to help get field work done in a timely fashion, and I thank him for his enthusiasm, humour, company and ideas. I also thank all those previously involved in this project. Tim Giles who initiated this study, along with Jeff Philips, who not only collected data for his thesis, but introduced me to the study site and also taught me the basic field techniques required to continue forward with this project. I would also like to acknowledge my research assistant and good friend Jen Andrews for numerous hours put into GIS digitizing and data entry. I would like to thank my office partner and good friend, Dave Luzi. Along with being full of great ideas, Dave was always there to provide a good laugh and never- ending support. I would also like to acknowledge all the technical GIS troubleshooting provided by Chris Adderley. I would also like to thank my family and friends for always standing by my side. Their never ending love, support and encouragement helped me persevere through the trials and tribulations of this process. A big thanks to Ryan Bonnett for enduring long endless nights, listening to me rant about thoughts and ideas, constantly providing feedback and support, and for believing in me, even when I didn’t. Finally, to my kitty Avalon, for always putting a smile on my face and keeping me company in my office all hours of the day/night. ix Chapter 1 Wildfire and Channel Morphodynamics 1.1 Introduction Fire occurs in all vegetated biomes on Earth to varying degrees and with variable impacts on vegetation and ecosystems, human health and safety, and economies (Taylor et al., 2009). An estimated 4000 fires occur in Canada each year (2000 of them in B.C.) burning almost 80,000 hectares annually (Taylor et al., 2009). Possibly as a result of the recent history of fire suppression, the 2003 fire season was one of the most severe in recorded history for British Columbia, Alberta, and the western U.S. Nearly 2,500 fires occurred in BC, burning almost 265,000 hectares. The drier, central part of the province was hit the hardest. In the Kamloops Forest Region alone, over 100,000 hectares were burned. The largest of these fires was the McLure fire (see Fig. 1.1), near Barrier B.C., which burned 26,420 hectares (Ministry of Forest and Range B.C, 2009). This fire was a very intense, stand-replacing crown fire that resulted in the loss or damage of 72 homes and nine businesses and forced 3,800 people out of their homes. Fishtrap Creek was the most severely burnt watershed in the area. Approximately two thirds of the watershed was burned, with only the western edge of the watershed remaining unaffected by the fire (Fig. 1.1). The area that was not burned is characterized by marshy, hummocky topography. Since the fire, salvage logging has taken place within the watershed, but the riparian zone adjacent to the study site has not been logged. The location of the study site for this thesis is shown in Fig. 1.1. The entire study site was burned, resulting in the death of almost 90% of the riparian vegetation. Severe, stand replacing fires seem likely to become more common as a result of climate change. Flannigan et al. (2005) predicted that increases in temperature and fire fuel will increase the area affected by forest fire by between 74 and 118% in Canada by the end of this century. Nitschke and Innes (2008) predicted a warmer, drier climate will increase the chance of fire ignition and propagation and will lengthen the fire season. In areas with climate regimes and vegetation types similar to the Kamloops region, Nitschke and Innes (2008) predicted a 30% increase in fire season length by 2070, and a 1 1.1. Introduction Figure 1.1: Location of Fishtrap Creek watershed, and with the extent of wildfire. Fishtrap Creek watershed is shown here as dark grey, while the McLure fire limits is delineated with a gray border and shaded with grey dots 95% increase in the fire severity resulting from a shift in fire behaviour from surface fire- intermittent crown fire regimes to a predominantly intermittent-full crown fire regime. This predicted increase in wildfire occurrence and severity in BC emphasizes the need for a better understanding of how rivers and aquatic ecosystems respond to these changes in order to implement appropriate management techniques. The effects of wildfire on landscapes has been studied in depth for several decades, with the majority of the research efforts in the past 10 years; however, these research efforts have mostly been geared toward the study of larger scale hillslope processes and watershed hydrology. The impact of wildfire on streams process and form has received relatively little attention. This thesis is a field study of the morphologic response of Fishtrap Creek following wildfire. This project provides a rare opportunity to perform a detailed study of the short term response of a gauged stream in order to provide insight on the temporal and spatial channel morphodynamics following a severe wildfire. The remainder of this chapter provides an in depth look at the form and function of alluvial channels, with 2 1.2. Controls on Channel Morphology and Stability emphasis placed on the influence of wildfire on channel morphology, watershed processes and large wood recruitment, distribution and movement. Case studies and model results documenting channel change following wildfire are presented and key knowledge gaps are identified. 1.2 Controls on Channel Morphology and Stability Alluvial channels are dynamic landscape features that can readily adjust aspects of their morphology, hydraulics, and sedimentology in response to altered environmental conditions or disturbances (Johnson and Simon, 1997; Buffington and Montgomery, 1999b). According to Lane (1957), the following are the most important variables that affect stream channel form: • Bank Resistance • Disturbance • Geology • Longitudinal Slope • Stream Discharge • Sediment Load • Temperature • Vegetation Changes in discharge and sediment can influence the hydraulic geometry of a chan- nel, and can alter the roughness and bed sediment size (Leopold and Maddock, 1953). Stream discharge is the most obvious factor in determining stream morphology (Lane, 1957); however, the magnitude and response to a change in discharge is highly depen- dent on climate and soil regimes. In rain dominated regimes, the frequency, timing, magnitude, duration and rate of change of flow can be significantly different than in snow, or nival, dominated regimes. Nival regimes are generally characterized by a sin- gle, long duration flood event generated during the melting of the snow pack. Rain driven regimes, on the other hand, often have multiple peak events which depend on rain events which are variable throughout the season. Flows can often be flashy, in which the timing, magnitude and duration of flows are highly dependent on the rain event and on the soil type and structure within the basin. Overland flow, generated when input exceeds the natural infiltration capacity of a soil, can result in faster and larger flow 3 1.2. Controls on Channel Morphology and Stability response in channels. In nival regimes, the influence of soil type and structure is less critical than in rain dominated regimes. The consequences of changes in long-term mean flow can result in variation in chan- nel dimensions, gradient, channel pattern, sedimentation, bank erosion rates, and chan- nel migration rates (Ashmore and Church, 2001). Montgomery and Buffington (1997) linked channel-reach morphology and channel processes to characteristic slopes within mountain drainage basins. Numerous studies have found that local changes in chan- nel gradient are a direct response to changes in discharge, sediment supply and caliber (Makin, 1948; Lane, 1957; Eaton and Church, 2004; Eaton et al., 2006). Some studies have shown that an increase in sediment delivery following disturbance can result in significant aggradation, reduced mean grain size and channel widening and/or braiding which may lead to localized changes in longitudinal gradient that can create feedback for more deposition or erosion (Madej and Ozaki, 1996; Miller and Benda, 2000; Hoffman and Gabet, 2007). 1.2.1 Sediment Supply and Transport The quantity of water and sediment introduced to the channel, the calibre of sediment, and the history and physiography of the landscape in which the river runs all govern the channel morphology of a river system (Church, 2002). The material transported through a stream can be classified based on its mode of transport. Wash load is material suspended and transported in the water column, which is seldom deposited within the channel or on the adjacent floodplain. Bed material load is the material that forms the bed and lower banks of the river. Bed material is usually transported as bedload (i.e., it travels by rolling and jumping along the bed surface (Church, 2006)), but will include minor amounts of sand carried in suspension at moderate and high flows (called suspended load). Most of the sand carried in suspension will be deposited upon the bar tops or outside the active stream channel upon the floodplain. Alluvial channel morphology is primarily an expression of the bed material load of a stream. While wash load and suspended load do not usually influence channel morphology, they can be indicators of basin management and sediment supply to the channel (Walling and Fang, 2003). The amount of sediment a stream can transport is called its transport capacity. Transport capacity is the maximum volume of sediment a channel can transport if unlimited by sediment supply (Buffington and Montgomery, 1999b) and is a function of flow magnitude and bed gradient. A channel is considered stable, or in a state of equilibrium, when the transport capacity is equal to the supply. An increase in sediment supply, in excess of the capacity of a channel, can result in depositional features, fining 4 1.2. Controls on Channel Morphology and Stability of bed material and braiding (Lane, 1957). A decrease in sediment supply can result in net erosion of material from the bed and banks, channel narrowing and/or straightening, and a reduction in bed slope via vertical degradation and/or lateral meandering (Galay, 1983; Dietrich et al., 1989; Church, 2002; Curran andWilcock, 2005; Dade, 2000; Church, 2006). In alluvial channels with coarse beds, the ability of a stream to entrain and transport sediment is a function of the shear stress exerted on individual particles by flow, the size of these particles, and the bed structure (Church, 2002)1. The reach average boundary shear stress exerted by flow is given by equation 1.1: τ = ρgdS (1.1) In this case, τ is the shear stress exerted by flow on the boundary (N/m2), ρ is the fluid density of water (kg/m3), g is the acceleration due to gravity (9.81 m/s2), d is the average water depth in m and S is the energy slope (m/m), which is usually assumed to be equivalent to the water surface slope over moderately long segments of channel. The critical shear stress, or the stress required to move a particle of a particular size, is given by equation 1.2, which is a modified version of Shields (1936) equation for sediment mobility: τc = τ∗c (γs − γw)D50 (1.2) In this case, τc is the critical shear stress (N/m2), τ∗c is the critical dimensionless shear stress defined by Shields, which ranges from 0.03 to 0.062, γw and γs are the unit weights of water and sediment, respectively (kN/m2), and D50 is the reach average median surface grain size (m). Disequilibrium in sediment supply and transport capacity, which are often responses to disturbance, can result in bed surface patchiness. Bedform adjustments, visible as textural patches or changes in surficial grain size distribution, are a response to size- selective deposition or entrainment caused by spatial variations in shear stress, sediment supply and bed slope (Buffington and Montgomery, 1999c). Surface texture response to changes in sediment supply rate has been documented by a number of researchers (Dietrich et al., 1989; Lisle et al., 1993; Buffington and Montgomery, 1999c; Nelson et al., 2009). Dietrich et al. (1989) suggested changes in sediment supply rate can result in longitudinal grain size segregation on the bed, in which coarse inactive zones and finer active zones can cause zones of congestion and areas of transition or smooth surfaces. The influence of surface hydraulic roughness elements can also result in highly vari- 1Bed structures include imbrication of the bed surface and the presence of stone clusters or nets. 2Meyer-Peter and Muller (1948) use an intermediate value of 0.048. 5 1.2. Controls on Channel Morphology and Stability able bed surface patchiness. Hydraulic roughness elements in gravel bed rivers com- monly arise from: • form drag due to bars and in-channel flow obstructions (boulders, wood debris, and bedrock projections); • skin friction and form drag caused by riparian vegetation lining the banks and protruding into the flow; and • momentum losses due to downstream changes in channel width and planform curvature (Buffington and Montgomery, 1999b). The presence of hydraulically rough features increases friction which can dissipate the total energy available for sediment transport. As a result, boundary shear stresses are reduced, which in turn lowers the transport capacity, causing substantial fining of the bed surface sediment. Bed surface patchiness and the arrangement of bed material influence the ability of a stream to entrain and transport material. The variation of grain entrainment with flow has important implications for the rates and frequency of sediment sorting and stream bed armoring (Haschenburger and Wilcock, 2003), which can further influence bed surface patchiness. Partial mobility, or selective transport, results when size frac- tions may include both grains that move and grains that remain stable under a given flow (Wilcock and McArdell, 1993, 1997; Church and Hassan, 2002). Conversely, full or equal mobility occurs when all size fractions on the bed are mobile under a given flow. It has been suggested that near full mobility of all grain sizes is responsible for the majority of geomorphic change within gravel bed channels (Hassan and Church, 2000; Church, 2006). However, evaluation of partial versus full mobility in rivers is limited by the lack of knowledge regarding the physical processes of sediment transport. Field assessment and collection of sediment transport data is costly, time consuming and often very difficult to measure (Bunte et al., 2004). Numerous sediment transport formulae have been developed to relate sediment transport to stream hydraulics for gravel bed rivers. Gomez and Church (1989) evaluated the performance of 12 bedload sediment transport formulae. Their results show that the formulae perform poorly when tested against data different from those used to calibrate them; consequently, there is currently no formula that can reliably predict or estimate bedload transport to better than one order of magnitude. Natural river variability arising from disequilibrium of flow hydraulics further con- founds the issues in applying bedload transport formula to natural rivers. Armouring, or paving, of a riverbed occurs wherever flowing water is incapable of moving all sizes of a 6 1.2. Controls on Channel Morphology and Stability widely graded sediment bed mixture (Galay, 1983). Surface armoring is the ratio of the surface median grain size, D50s , to the subsurface median grain size D50ss (D50s/D50ss). Hassan et al. (2006) suggested that sediment supply is a first order control on the de- gree of surface armoring for different river systems, followed by the shape, duration and characteristics of the flow hydrograph. Surface armoring occurs under almost all flow conditions. In ephemeral gravel bed desert streams, Hassan et al. (2006) found armor ratios on average of 1.2, while in snowmelt-fed gravel bed streams, they found an average surface armoring of 3.4, with armor ratios reaching as high as 7. Surface armoring becomes more pronounced downstream of dams or sediment traps, which char- acteristically results in a sediment starved system wherein transport capacity exceeds the supply. In this case, fine sediment winnowing occurs from increased flow and/or reduced sediment supply (Dietrich et al., 1989; Lisle and Hilton, 1992; Buffington and Montgomery, 1999a; Whiting and King, 2003). As a result of size-selective transport and subsequent armoring, sediment transport formulae can typically produce errors up to two orders of magnitude in mountain rivers, which are characterized by coarse bed materials and steep slopes (Bathurst et al., 1987; D’Agostino and Lenzi, 1999; Barry et al., 2004). The morphologic method for estimating sediment transport, also called the inverse approach, is an alternate means of evaluating sediment transport using topographic survey data. This approach has been used in numerous studies (Neill, 1987; Carson and Griffiths, 1989; Ashmore and Church, 1998; Stojic et al., 1998; Ham and Church, 2000; Eaton and Lapointe, 2001). This method uses sediment continuity, evaluated from bed scour and fill estimated from cross-sectional surveys or digital elevation models, and path length (or grain travel distance) to evaluate sediment transport. Morphologic methods are limited in three ways: i) they are limited to estimates of the transport of the sediment that interacts with the bed; as a result, these methods yield minimum estimates of bed material flux (Ashmore and Church, 1998), ii) in order to make accurate estimates of scour and fill, detailed topographic data must be collected (Lindsay and Ashmore, 2002), and iii) these data must be collected before and after each event for which sediment trans- port is to be estimated. However, given the current problems with sediment transport functions and direct field measurements of bedload transport, the morphologic method is becoming more widely used as a basis to estimate bed material transport and to improve understanding of channel deformation and development (Church, 2006). 7 1.2. Controls on Channel Morphology and Stability 1.2.2 Riparian Vegetation and Large Wood The influence of riparian vegetation on bank strength has been recognized for some time, and has been studied in relatively great depth. Flow resistance and bank strength from root reinforcement of fluvial sediments can influence channel stability to a high degree (Hickin, 1984; Naiman and and Décamps, 1997; Kiley and Schneider, 2005; Perucca et al., 2007). The effects that vegetation has on the stream environment are related to the size of the stream, the hydrologic regime, and the local geomorphology (Naiman and and Décamps, 1997). If the cohesion provided by roots is reduced, the shear stress exerted from flow on the banks may cause instability and failure of the banks (Millar, 2000, 2005; Anderson et al., 2004). In this case, banks act as key sediment sources, and can supply up to 50% of catchment sediment output (Lawler et al., 1999) Abernethy and Rutherfurd (2000) used a physically based slope stability model to evaluate the influence of vegetation roots on bank stability. They found that the presence of mature trees increases bank stability against mass failure by reinforcing bank sediments with roots. The type of riparian vegetation, and thus rooting depth and relative bank strength, is important when evaluating the bankfull width. Eaton and Giles (2009) found that vegetation represents a very important and quantifiable control on downstream hydraulic geometry relations and meandering thresholds. Eaton (2006) found that representing the effect of vegetation on bank strength using an effective cohesion form in a regime model increased the accuracy of bankfull geometry predictions for different vegetation types. Murray and Paola (2003) used a simple numerical stream- pattern model to investigate the influence of sediment stabilization by roots on the channel pattern of bed-load rivers. Their results showed that bank stability is the main cause of single-channel streams and that a reduction in riparian vegetation will result in braiding of these single-channel streams. Similarly, using laboratory experiments, Tal and Paola (2007) found bank stabilization from vegetation can organize flow and convert the planform morphology from braided to single-thread. Braudrick et al. (2009) found that sustained meander development is highly dependent on strength and cohesion provided by bank vegetation. Large wood (LW) can affect channel morphology by altering in-stream hydraulics, bed and bank scour and trapping increased sediment behind LW obstructions in small and intermediate channels (Montgomery and Buffington, 1997; Montgomery et al., 2003a). The movement of wood, as well as changes in the size and amount supplied to a stream, can have effects as great as those arising from changes in sediment supply or discharge (Montgomery et al., 2003a). LW that covers as little as 2% of the surface area of a stream can provide about 1/2 of the total roughness or flow resistance in rivers (Manga and Kirchner, 2000). LW and log jams in rivers influence the channel hydraulics 8 1.3. Watershed Response Following Wildfire by dissipating the energy available for sediment transport. These hydraulically rough elements are responsible for distinct patterns of surface fining and coarsening within the active channel. Most studies document fine sediment deposition upstream and sur- face coarsening downstream of log jams (Keller and Swanson, 1979; Hogan, 1989; Gomi et al., 2001; Brummer et al., 2006). Buffington and Montgomery (1999b) found that streams with greater roughness had systematically reduced reach-average surface grain sizes. They also found that the estimated median grain size of wood-rich channels can be up to 90% less than the predicted competent value for the bank-full stage. Individual logs create hydraulic roughness elements that can increase pool spacing and frequency of textural patches and result in localized channel widening (Montgomery and Buffington, 1997; Buffington and Montgomery, 1999a; Montgomery et al., 2003a,b). The function of LW within a channel is dependent on the size and characteristics of both the channel and the wood being recruited to the channel. The time in which a LW jam or piece remains stable and an integral part of the bed depends on the age and integrity of the jam or wood. Hogan (1989) found that log jams in disturbed areas influenced channel form the greatest within the first 10 years followed by little to no influence after 30 years. The structure of these jams, along with the distribution of LW and sedimentologic response in streams, is also very important for providing good quality aquatic habitat, particularly for fish (Montgomery et al., 1996; Buffington and Montgomery, 1999b,c; Roni and Quinn, 2001). The recruitment rate of LW following widespread death of riparian vegetation can be significantly increased in the years fol- lowing wildfire. Similarly, the movement of in-stream LW can also be influenced by fire, resulting in dynamic changes in LW function and distribution within a channel. 1.3 Watershed Response Following Wildfire Wildfire causing widespread vegetation mortality can have important implications for hydrologic and geomorphic processes. Reduction in forest cover can result in lowered interception, a larger snow pack and faster ablation rates (Winkler, 2009; Silins et al., 2009). Silins et al. (2009) reported advanced onset of the melt period by approxi- mately 14 to 24 days and an increase in the magnitude of peak flow events by approx- imately 110% due to greater snowpack accumulation following the Lost Creek Fire in the southern Rocky mountains. Similar results were documented by Winkler (2009) on the Thompson plateau near Mayson Lake, north-west of Kamloops BC. Fire can also reduce water infiltration by inducing or enhancing soil water repellency, (i.e. hydrophobicity), which can result in an increase in runoff and erosion (Scott, 1997; Pierson et al., 2001; Shakesby and Doerr, 2006). Researchers have found the 9 1.3. Watershed Response Following Wildfire influence of soil repellency on increased runoff is highest the first year after the fire and reduces significantly after the first year (Chanasyk et al., 2003; Shakesby et al., 2003; MacDonald and Huffman, 2004; Doerr et al., 2006; Shakesby and Doerr, 2006; Covert et al., 2009). The degree of water repellency and hillslope response is positively related to fire intensity (Shakesby et al., 2003). A review presented by Moody and Martin (2001) found that increased occurrence of runoff declined after the first three to four years following wildfire. Many studies have also found that ash coverage can significantly reduce the oc- currence of hydrophobic soils immediately after wildfire, thus reducing the post-fire hydrogeomorphic response (Woods and Balfour, 2008; Cerda and Doerr, 2008; Larsen et al., 2009). In fact, Cerda and Doerr (2008) found that the mean sediment yield from the ash covered plots was more than two orders of magnitudes lower than from bare plots. In this case, the reduction in runoff and thus sediment yield due to ash will remain as long as the ash remains in on the surface. While it is understood that the incidence of hillslope erosion depends on fire severity, as well as the climate and topography of the burned area, there is no definable common effect on the timing, magnitude or probability of hillslope erosion and/or runoff follow- ing wildfire. Some studies have found that moderate to high runoff events capable of evacuating fine-grained material from the hillslopes to the channel network occurs over a period of 5 to 10 years following wildfire (Legleiter et al., 2003; Moody and Martin, 2001). Moody and Martin (2009) argue that the sediment available within gullies and channel networks may be more important than slope or soil erodibility in predicting post-fire sediment yields. In rain dominated climate regimes, high intensity rain storms significantly increase the occurrence of hillslope erosion for the first 5 years following wildfire. Wondzell and King (2003) suggested that debris slides and debris flows occur more frequently after wildfire as a result of high intensity, low frequency storms and are thus contingent upon the occurrence of such storms soon after the fire. Similar results were documented by Keller et al. (1997), Spigel and Robichaud (2007) and Jordan et al. (2009). Keller et al. (1997) found that only precipitation events at least twice the normal rain events which occurred within the first two years following wildfire could potentially result in moderate to large debris flows. In snowmelt dominated regimes, Spigel and Robichaud (2007) found that the most significant debris flow event occurred during a period of rapid snowmelt, which is expected to reflect the increased snow accumulation from reduced interception. Furthermore, fire severity, or the percentage of total tree mortality within a catchment, is critical when evaluating potential erosional response following wildfire. Reduced root strength from tree mortality may result in reduced soil cohesion, which can increase the occurrence of landslides and debris flows. Shallow 10 1.4. Wildfire Case Studies Documenting Channel Response and LW Loading root systems can be weakened, increasing the susceptibility of the trees to disease and insect attacks following fire, resulting in slower growth, death and blow-down during large storms. A field study performed by Jackson and Roering (2009) in the Oregon Coast Range found that root strength decline was particularly rapid in the first year following wildfire and slowed by the fourth year, which is generally consistent with post- harvest decay trends. They also found that root strength was substantially lowered at one site only one month following the fire, suggesting that these roots may have been compromised before the fire. Benda and Dunne (1997) suggest an increase of erosion is likely to occur 4 to 8 years after following vegetation mortality due to reduced soil reinforcement and strength from decaying roots. 1.4 Wildfire Case Studies Documenting Channel Response and LW Loading It is recognized that disturbances, both natural and anthropogenic, can result in a disequilibrium of in-stream forces and variables that can result in substantial channel changes (Simon and Rinaldi, 2006). Table 1.1 presents case studies and model results that have documented the influence of wildfire on channel morphology and LW recruit- ment and movement. Channel response is highly variable in space and time. The severity of a wildfire, and the overall impact on the riparian vegetation has a significant influence on the degree of channel change. While the incidence of increased runoff and erosion is not consistent for all case studies, one or the other is always present in the studies presented in the literature. That is to say, in all reported instances of fire induced channel change, hydrologic and geomorphic agents — be it increased erosion and/or runoff — were documented to be the driving forces in channel change following fire disturbance. Given the significant impacts of LW in streams, there is relatively little documen- tation of post-fire input rates to river systems after wildfire. This may in part be due to the natural complexity in rivers and/or the variability in disturbance regimes. Most of the current research has focused on understanding the influence of anthropogenic modifications, most specifically logging, on the rate of recruitment into streams (Bilby and Ward, 1991; Ralph et al., 1994; Comiti et al., 2006; Webb and Erskine, 2003). The last five cases studies presented in Table 1.1 show documented responses of LW recruitment and movement following wildfire. It is evident that movement of LW is greatly dependent on the size of the channel, and the influence the wildfire has had on the integrity of instream wood. Marcus et al. (2002) evaluated the LW within different order streams to evaluate the distribution and movement within the channel bound- 11 1.4. Wildfire Case Studies Documenting Channel Response and LW Loading aries. Their results show that LW movement is transport-limited in the headwaters and supply-limited in the lower reaches, with a trend toward equilibrium in intermediate reaches. In the studies documenting LW movement following a wildfire, the movement of wood is often attributed to loss of structural integrity from burning of in-stream wood (Zelt and Wohl, 2004; Minshall et al., 1997). Scherer (2008) found that instream wood volumes were three times higher in streams recently disturbed by wildfire. His findings confirm that wildfire disturbance is an important mechanism for wood recruitment into streams. However, the limited number of case studies documenting short and long term LW dynamics following wildfire has resulted in a poor understanding of the timing and magnitude of wood recruitment to channels, as well as the distribution and movement of LW within small, intermediate and large channels. The contrasting results documented by Minshall et al. (1997) and Beche et al. (2005) emphasize the importance of fire sever- ity on LW recruitment. It is speculated, based on these wildfire case studies, that fire severity in the riparian zone is one of the most important variables in understanding LW recruitment magnitude and timing, as well as the distribution and potential for movement within a channel. Disturbance and channel morphologic adjustment can have adverse effects on aquatic life. The role of natural disturbances, such as wildfires, floods and mass wasting with regards to habitat function and existence are generally not well understood. In the event of fire, most studies record an immediate input of sediment initially following the fire (refer to Table 1.1). These large influxes of sediment following disturbances can be detrimental to stream functioning. In many cases, increased supply (in which transport capacity of the river is exceeded) can result in lower pool density which can have ad- verse side effects on aquatic life, including fish (Madej and Ozaki, 1996). The timing and nature of the disturbance is important when understanding influence on aquatic habitats. Kondolf (2000) found that timing of sediment transport from anthropogenic modifications result in increased sediment supply that may occur during low flows in the channel; on the other hand, when fine sediment enter the channel during higher flows (often the case in natural disturbances) stream power is adequate to move and disperse this material adequately. Most studies have found adverse side effects following anthro- pogenic disturbances on fish egg and larval mortality (McNeil and Ahnell, 1964; Adams and Beschta, 1980; Scrivener and Brownlee, 1981). Conversely, Benda et al. (2003) found that disturbances may contribute to or govern significant aspects of physical het- erogeneity and perhaps biological diversity in rivers which can be viewed as positive events in the life of a watershed. They also suggest anthropogenic modifications of streams may result in decreased physical heterogeneity while natural disturbances may benefit in the long-term from increased physical and biologic diversity. Similarly, Lisle 12 1.4. Wildfire Case Studies Documenting Channel Response and LW Loading (1989) found that scour could erode eggs laid in the bed and expose deeper levels of the bed to infiltration by fine sediment, but at the same time could allow fine sediment that is detrimental to fish to be winnowed away. 13 1.4. W ild fi re C ase S tu d ies D ocu m entin g C h an n el R esp on se an d LW L oad in g Table 1.1: Sediment Supply, Discharge Changes and LW Recruitment Following Wildfire, Prescribed Fire and Model Case Study Results Documented watershed and stream response following wildfires, prescribed fires and model case study results Research and Study Area Year ∆ sediment supply ∆ discharge Channel changes Severe Fire Sleeping Child Rainfall triggered sediment overall reduced grain size, incising Creek Basin in west-central 2000 pulses from hillslopes Intense rainfall of fans, large-scale aggradation Montana resulting in formation of new floodplains, Hoffman and Gabet (2007) widening of fans and braiding Rabbit Creek Fire Punctuated increase local steepening of the longitudinal in the Boise River basin 1995 in sediment supply Not specified profile immediately downstream Benda et al. (2003) from basin of alluvial fans and an overall increase in bed grain size distribution Yellowstone National modeled results suggest an increase Park Fires 1988 Increased Increased in the probability of sedimentation on Meyer and Wells (1997) alluvial fans Yellowstone National Model showed more extensively burnt Park Fires watersheds tend to have increased stream Legleiter et al. (2003) 1988 Increased Increased power per unit width, lower w:d ratios, and less bank failures (Noted results are highly dependent on temporal and spatial variability Painted Cave FIre Increased sediment yield Definite sequences of scour and deposition Southern California from hillslopes following with sediment limited and transport limited storm Keller et al. (1997) 1990 storm surges Storms (Rain Events) surges. Smoothing of longitudinal profile in transport limited scenario, and sediment flushing with sediment limited Prescribed (low-moderate intensity) Prescribed fire in riparian zone resulted in minor Fire, Dark Canyon Ck. Little to none Little to none changes in fine sediment concentrations within the Watershed, El Dorado California 2002 stream Beche et al. (2005) Severe Wildfire Increased suspended sediment concentrations Santa Ana RIver Periodic within the creek, sediment discharge, and California Increased (urbanization Increased (urbanization increased discharge (compounded by the Warrick and Rubin (2007) and wildfire induced) and wildfire induced) influence of urbanization in this basin LW Recruitment following wildfire and model case study results Research and Study Area Year of Fire Study description General Findings Specific Study results Yellowstone National Evaluate LW recruitment Increase in LW 1st order streams have highest LW recruitment Park, Severe wildfire and movement within 20 Input and some first and forth year after fire. 2nd and 4th order Minshall et al. (1997) 1988 streams minor movement streams have smaller and similar recruitment, and 3rd order streams have 4 times more LW inputs the first year after the fire. Movement was due primarily to in-stream wood that had been burnt Prescribed (low-moderate intensity) Evaluated LW recruitment Very little change Riparian mortality was only 4.4% and they Fire, Dark Canyon Ck. in a 1st order watershed and in-stream LW did not see significant change in LW volumes Watershed, El Dorado California 2002 compared to five unburned volumes or recruitment Beche et al. (2005) sites (1 to 7 years post fire) Moderate - Intense modeled the effects Bimodal tree Bimodal distribution suggested LW recruitment Wildfire of fire on LW recruitment recruitment distribution very high initially after the fire, followed by (Laboratory Model) N/A by simulating moderate the first 30 years a period of low recruitment, and then Bragg et al. (2000) to intense wildfire following fire high recruitment 2 to 3 decades later. Wildfire Compared the response of LW accumulations Wood loading was greater in unburned basin Park County, stream sediment and LW lower in burned due to lower ratio of Wbf to LW size - greater Wyoming 1988 in burned and unburned catchments movement in burnt streams. Wbf greater in burned Zelt and Wohl (2004) creek creek. Unburned exceeded burned creek in frequency of LW and LW accumulations Wildfire, Plateau Regions of Evaluated temporal trends in Fire important Distinct temporal trend with elevated volumes B.C., Canada > 30 ya wood loading in 38 small streams recruitment method present 30 - 50 years after fire. Wood volumes Scherer (2008) following wildfire 30-50 years after fire in streams 3x higher than older riparian stands 14 1.5. Objectives and Research Questions 1.5 Objectives and Research Questions The current knowledge of changes in LW loading, channel morphology, and sedimentol- ogy following fire is limited. Differences in climate, burn severity, and general topogra- phy result in variable channel responses. As a result, we are currently unable to predict the magnitude, timing or direction of watershed response following fire. Most published studies have focused on channel change in response to large increases in runoff and/or soil erosion, while none evaluate stream response in the absence of these exogenous hillslope variables. This study will evaluate the timing and magnitude of cross-sectional and planimetric adjustments observed in Fishtrap Creek over the five years since the fire. The timing, type and rate of recruitment of LW into the creek following the fire will be explored and compared to previously documented wildfire studies and model results. The LW loading in Fishtrap Creek will be examined and compared to studies in both disturbed and undisturbed basins, and the influence of key LW jams and pieces on the channel sedimentologic structure will be evaluated. Finally, the bed material sediment transport will be estimated using the morphologic method, in which the results will be compared and related to the channel morphology and channel sedimentology. The goal of this research is, i) to provide a conceptual model showing channel re- sponse over time, ii) to evaluate LW recruitment rates and distribution to better un- derstand sedimentologic surface response iii) to evaluate sediment transport using the morphologic method to help understand the observed morphologic response, iv) to re- late sedimentologic changes from LW loading and sediment transport patterns on the influence of fish and aquatic habitat. The following research questions will be addressed: 1. Are the timing and magnitude of post-fire channel changes different from studies presented in the literature? 2. Is LW recruitment, movement and loading following wildfire in Fishtrap Creek different than other studies? 3. How do the sedimentologic patterns relate to the LW loading and sediment trans- port patterns? 4. How does data density influence the accuracy of the morphologic method for estimating bed material sediment transport rates and for reconstructing patterns of transport within the study reach? 5. How might fire adversely impact aquatic ecosystems in systems similar to Fishtrap Creek? 15 1.6. Thesis Organization 1.6 Thesis Organization Chapter 2 begins with a detailed description of Fishtrap Creek watershed, including the location, climate, and topographic relief. The glacial influence and characteristics of Fishtrap Creek watershed surficial deposits are also described. Chapter 3 presents the initial study design, field data collection methods, and a detailed description of the data collected and analyzed for this thesis. Chapter 4 presents the post-fire morphologic channel adjustment from 2004 to 2008, with emphasis on changes since 2006. Suspended sediment concentrations and hydro- logic changes following the fire will be presented. Cross-sectional adjustments prior to the 2007 freshet are compared to the data collected after the 2007 and 2008 freshets. Planimetric maps evaluating bank change and bar distribution from 2006 to 2008 are compared and related to the timing and magnitude of channel response. In addition, changes in surface elevations, in the form of erosion or deposition between each year, is evaluated in order to define zones of net erosion and/or deposition throughout the study reach. Chapter 5 evaluates the volume of in-stream wood before and after the fire. Rate and mode of LW recruitment has been identified, and the distribution and movement have been evaluated. The hydraulic influence of pre-fire and post-fire LW on channel topography and bed surface grain size distributions has been evaluated, in which a facies map of the bed surface has been created, and the function of new and old LW has been quantified. In addition, the structural integrity, orientation and overall function of new and old jams have been evaluated in relation to their influence on channel topography and facies distributions. Chapter 6 evaluates the sediment transport using the two different applications of the morphologic method: i) traditional cross-sectional data method which evaluates year-to-year topographic change and ii) subtraction of surface DEM’s (Digital Elevation Model) created from survey points throughout the channel. Results showing patterns of channel change and sediment transport from the traditional cross-section method are gathered from Eaton et al. (in press). Finally, patterns of bed grain size texture distributions are related to source material and patterns of sediment transport. Chapter 7 provides a timeline of the response following disturbance in Fishtrap Creek, including insight into future channel adjustments and recruitment modes and quantities of LW. Effects of the channel adjustments, primarily sedimentology and LW loading, on the influence on aquatic habitat and fish spawning have been evaluated and recommendations for data collection and management techniques have been provided. 16 Chapter 2 Study Area Description 2.1 Geographic Location of Fishtrap Creek Fishtrap Creek is located approximately 50 km north of Kamloops, British Columbia (see Fig. 1.1). The study reach is 400 m long, and is located just upstream of the Water Survey of Canada gauging station (08LB024) near Westsyde Road (see Fig. 2.2). The average bed gradient of the study site is 0.02 m/m. Fig. 2.1 shows the longitu- dinal profile along the thalweg within the study reach. Bankfull widths throughout the study site are highly variable, ranging from 7.5 m to almost 25 m in 2007/2008, with an average bankfull width of approximately 12 m. Due to the complex channel morphology, bankfull depth varies across the channel. Multiple depths for each cross-section were averaged to estimate the average bankfull depth for the reach. Reach average bankfull width and depths were determined using 27 cross-sectional transects throughout the study reach. The reach average depth is approximately 0.5 m, and velocity at bankfull flow is approximately 1.23 m/s. Figure 2.1: Fishtrap Creek Longitudinal profile along 400 m study site, measured along the thalweg in 2008. 17 2.1. G eograp h ic L ocation of F ishtrap C reek Figure 2.2: Fishtrap Creek planimetric profile along 400 m study site showing 2008 bank locations, bars, avulsions, LW, jams, and the locations of the 2006 abandoned channel 18 2.2. Physiography 2.2 Physiography Characterized by highly incised streams, steep hillslopes and undulating plateau, Fish- trap Creek watershed spans nearly 165 km2 in area. The elevation ranges from 375 m asl at the mouth of the North Thompson to 1610 m asl on the North Thompson plateau. Fishtrap Creek runs 30 km, predominantly southeast, originating on the Interior plateau at 1380 m asl and ending at its confluence with the North Thompson river at 375 m asl. The creek is deeply incised in the landscape, resulting in steep, highly coupled slopes for the uppermost 70% of the stream. According to the stream size classification proposed by Church (1992), Fishtrap Creek can be classified as an intermediate stream. In this case, the bankfull width of Fishtrap Creek is far greater than the bankfull depth, (12.5 m >> 0.3 m), bankfull width is between 20 and 30 m, and boulders represent significant roughness elements in flow. Fishtrap’s largest tributary, Skull Creek, joins Fishtrap Creek approximately 15 km upstream from the confluence of Fishtrap to the North Thompson River. The relief of Fishtrap Creek is presented by the hypsometric integral shown in Fig. 2.3. The hypsometric curve for Fishtrap Creek Watershed is slightly convex in nature (Fig. 2.3). Almost 50% of the watershed area resides on the Interior plateau between 1300 and 1600 m asl. Another 30% of the basin is found between 1000 to 1300 m asl, while only 20% of the basin area is between found 300 to 1000 m asl. The area underneath the curve, or the hypsometric integral (HI), can be estimated using equation 2.1: HI = h̄− hmin hmax − hmin (2.1) where hmin = 345 m, hmax = 1610 m, and h̄ is 1300 m. The HI value for Fishtrap is approximately 0.75. This relatively high hypsometric integral value, in combination with the convex nature of the curve, indicates that Fishtrap Creek is a relatively young landscape, wherein tectonic uplift has a greater influence on the landscape than fluvial erosion and mass movement. This basin is currently not at a steady state; rather, it is in a dynamic state that is continuously evolving as gullies expand and erode into the plateau. 2.3 Glacial History and Surficial Deposits The glacial and postglacial geologic history of south-central British Columbia in general, and the Kamloops region in particular, has been well documented (Ryder et al., 1991; Tribe, 2005). Although the region has experienced several glaciations, the effects of the recent glaciation are most apparent in determining the type and distribution of the 19 2.3. Glacial History and Surficial Deposits Figure 2.3: Hypsometric curve for Fishtrap Creek. Elevation bands were categorized from the 340 m- 750 m, 750 m - 1000 m, 1000 m -1250 m, 1350 m - 1500 m, and 1500 m to the maximum basin elevation of 1610. The solid line indicates the elevation at which 50% of the basin is above or below. surficial sediments in this area (Tribe, 2005). The current topography and patterns of surficial deposits within the Fishtrap Creek basin reflect the combination of the glacial history and continuous reworking by ongoing fluvial erosion and mass movement. Glacial features as eskers are present within the basin, mostly on the north side of Skull Creek. Rapid downwasting of the Cordilleran Ice Sheet, between 15,000 and 10,000 years before present, deposited a thin veneer of basal and ablation till over about 80% of the watershed. Figure 2.4 provides a detailed map of the surficial geology of Fishtrap Creek Watershed. 20 2.3. G lacial H istory an d S u rfi cial D ep osits Figure 2.4: Fishtrap Creek basin surficial deposits and terrain mapping along with location of normal fault boundary. Source: Surveys and Mapping Branch for the Department of Energy, Mines and Resources 1976. 21 2.3. Glacial History and Surficial Deposits Ablation till was deposited by stagnating ice in several high-elevation portions of the region, while the lower elevations are mantled by poorly sorted, matrix supported basal till (Tribe, 2005). Evidence of the former meltwater channels and deltaic deposits are found in abundance within the watershed. Glaciofluvial fans and terraces are most prevalent on either side of Skull Creek and in the lowermost reaches of Fishtrap Creek (Fig. 2.4). Evidence of these terraces are shown the LIDAR image presented in Fig. 2.5 subset A. Figure 2.5: LIDAR (Light Detection and Ranging) image showing major surficial de- posits along Skull and Fishtrap Creek The mid-reach of Skull Creek has numerous terraces. In some sections, up to 10 levels of terraces created from rapid incision of Skull Creek into these glacial deposits can be identified. These deposits are generally massive to stratified sand, often with gravel and silt textures. As the glacial meltwater channel entered the North Thompson Valley, it deposited glaciofluvial material in a large fan; as a result, the lowermost reaches of Fishtrap Creek floodplain, where it joins the North Thompson River, has incised into the glaciofluvial deposits. The majority of the slopes adjacent to Fishtrap Creek have slopes that range from 30 to 60%. These slopes are mantled by a thin veneer of colluvial sediments, possibly related to mass wasting of surface deposits following deglaciation in this area. A north trending fault line runs through Fishtrap Creek just downstream of the study site (see 22 2.3. Glacial History and Surficial Deposits Fig. 2.4). This fault was active during the Paleogene extension, resulting in intrusions of volcanic and sedimentary rocks. The Eocene magmatic and tectonic event 53 to 42 Ma resulted in widespread volcanism (Tribe, 2005). Volcanic deposits are widespread, but are most evident in the southernmost part of the Fishtrap Creek Basin (Fig. 2.4). Skull Creek is a very important tributary as it provides a great deal of flow and sediment to Fishtrap Creek. Figure 2.6 presents the longitudinal profiles for Fishtrap and Skull Creek. Figure 2.6: Longitudinal Profile along the mainstem of Fishtrap Creek and Skull Creek with 5x vertical exaggeration. Fishtrap Creek has an average gradient of 0.045 m/m while Skull Creek has a bed slope of 0.048 m/m, only slightly steeper than Fishtrap Creek (Fig. 2.6). Skull Creek drains an area that is characterized by gentle to strongly sloping topography (5% to 30% grade), with many poorly drained soils and marshy depressions. The lower 500 m of Skull Creek is highly incised within the landscape. The average bed slope for the lower 500 m is 0.20 m/m (Fig. 2.6). The confluence of Skull Creek to Fishtrap Creek is characterized by a large debris flow fan (shown as subset B in Fig. 2.5). This deposit is probably the primary source of sediment for Fishtrap Creek. The sediment carried within Fishtrap is a expression of landscape and source areas. The material being supplied from Skull Creek is primarily basal till and glaciofluvial deposits with gravely sand loam, and sandy loam textures. Large boulders are most likely supplied either from debris flows, or are ‘lag’ deposits from glaciofluvial materials 23 2.4. Climate and Hydrology that cannot be transported by the current flow conditions. The median grain size for the study reach is approximately 45 mm, while the median subsurface grain size is 36 mm. Bed surface grain size distributions are a function of roughness elements which have resulted highly variable grain sizes distributions within the active channel margins. Evidence of this will be presented Chapter 5 and 6. 2.4 Climate and Hydrology Fishtrap Creek watershed is located in the dry Southern Interior Plateau of British Columbia in the rain-shadow of the Coast Mountain Range. As a result, the weather in this region is typically continental and very arid. Based on the present climate station in Kamloops (350 m asl), the region’s valley bottoms receive on average 279 mm of total annual precipitation, while the highlands adjacent to our study area receive just under 700 mm (Winkler et al., 2005), 60% of which falls as snow. Kamloops winters are generally cold, with a mean temperature for January of 4.2◦C (see Fig. 2.7). Temperatures increase rapidly in the early spring and summers are warm, with a mean July temperature of 21◦C. Most summers experience maximum daily temperatures over 38◦C, while the maximum temperature recorded at the Kamloops International airport was 40.6◦C (Environment Canada, 2009). Winds usually flow through the valley with a prevailing direction from the east, though summer months have a predominately westerly component. Strong winds are not uncommon; cross-valley wind gusts upwards of 137 km/h were recorded on March 30, 1975. These cross-valley winds may be important for increased potential of tree throw within the watershed and more importantly into streams, especially after large scale disturbance when root systems are weak. The watershed has a nival regime, in which the highest flows occur during the spring snow melt in April and May (Fig. 2.7). The flat, hummocky terrain of the Interior Plateau results in very rapid and relatively uniform melt rates in the Fishtrap Creek Basin. Table 2.1 provides the peak discharge values, including the date and time in which they were recorded for Fishtrap Creek3. The dominant discharge, or bankfull discharge, for Fishtrap Creek is estimated to be approximately 7.0 m3/s. Flows of this size range occur once every 2.5 years on average. Discharges greater than or equal to bankfull discharge are highlighted in grey in Table 2.1. Flows greater than 10 m3/s have a return period of approximately 4.8 years. The highest daily average peak flow recorded was 14.9 m3/s, which occurred on May 15, 3Fishtrap creek has been continuously monitored by theWater Survey of Canada since 1971, including periodic measurements from 1915 to 1971. 24 2.4. Climate and Hydrology Figure 2.7: Fishtrap Creek hydrograph, and Kamloops average monthly precipitation and average monthly temperature normals. Runoff calculated using total monthly mean daily discharges for each month. 25 2.4. Climate and Hydrology Table 2.1: Peak Discharge Records from WSC Gauging Station at Fishtrap Creek, Kamloops B.C. Water Survey of Canada Flow Data for Fishtrap Creek (Station I.D 08LB024) Date Instantaneous Daily Average Peak Max Peak (m3/s) Flow (m3/s) Pre-Fire Flows 1971 n/a 10.6 May 22, 1972 n/a 12.7 May 24, 1973 3.77 3.37 May 8, 1974 8.72 8.35 June 2, 1975 12 10.3 1976 n/a 10.3 April 26, 1977 4.11 3.91 April 28, 1978 6.03 5.89 May 6, 1979 4.33 4.18 April 29, 1980 3.31 3.09 May 1, 1981 3.78 3.67 May 17, 1982 9.64 9.23 1983 n/a 7.95 May 30, 1984 9.55 8.29 May 19, 1985 9.27 8.86 May 20, 1986 4.13 4.05 May 1, 1987 11.7 10.2 April 22, 1988 3.6 3.43 May 2, 1989 5.85 5.37 1990 n/a 10.9 May 10, 1991 3.98 3.89 April 30, 1992 3.88 3.75 May 14, 1993 10.5 9.45 April 22, 1994 6.01 5.95 May 12, 1995 9.67 9.39 May 18, 1996 10.8 10.2 May 15, 1997 15.3 14.9 May 3, 1998 10.1 9.4 May 25, 1999 13.5 11.7 April 28, 2000 5.05 4.74 May 14, 2001 3.13 2.97 May 22, 2002 8.68 8.38 May 2, 2003 4.2 4.09 Post-Fire Flows April 15, 2004 6.3 4.56 April 25, 2005 10.1 8.93 April 30, 2006 8.8 7.18 May 9, 2007 8.02 6.61 May 17, 2008 8.08 6.83 26 2.5. Vegetation and Disturbance Regime 1997, while the lowest daily recorded peak flow, 2.97 m3/s, occurred on May 14, 2001. The largest flow following the McLure fire occurred in 2005 with a daily average peak flow of 8.93 m3/s. 2.5 Vegetation and Disturbance Regime The lower elevations in the Fishtrap Creek basin are in the Interior Douglas-Fir Biogeo- climatic Ecosystem Classification (BEC) zone, while the higher parts of the basin are in the Montane Spruce BEC zone (MoFR, 2009). Insect disturbance is important for both BEC zones. Since the early 1990’s, Interior Douglas-Fir forests, and to a smaller extent Montane Spruce forests, have been influenced by the western spruce budworm (WSB) (Campbell et al., 2005). Lodgepole Pine stands found within the Montane Spruce BEC zone have been also been affected by MPB (Mountain Pine Beetle) infestations. In the Fishtrap Creek basin, salvage logging has occurred in response to MPB infestation primarily along the western most region of the Interior Plateau (see Fig. 1.1). Campbell et al. (2005) found WSB in the Interior Douglas-Fir forests to have a return period of nearly 50 years. These BEC zones area also frequently influenced by forest fires. Stand replacing fires typically have a return interval of 100 to 300 years in the South-Central Interior. However, smaller surface fires occur on average every 25 years (MoFR, 2009a). Insect in- festations and fires are both highly dependent on climatic controls. The climatic regime has shifted and is continually shifting, as a result, prediction of future occurrence of these disturbances is difficult. It is expected, however, given the magnitude of distur- bance generated by fire compared to that of insect infestation, that fires are responsible for the most significant disturbance in this region. 27 Chapter 3 Methods 3.1 Study Design The study design employed for this project combines a number of quantitative and qualitative data collection techniques to evaluate short term morphologic adjustment of Fishtrap Creek following wildfire. An empirical approach, combining various sources of field data, was used to evaluate a 400-m length of stream since 2004. We have carried out detailed surveys of the channel study reach, evaluated LW recruitment and movement, and taken repeated photos of the reach from established photo locations during numerous field visits to the site. Bed surface textures have been estimated for the entire stream bed for this study to better understand the influence of large woody debris and sediment transport on bed sedimentology. A digital elevation model of the bed was constructed from a survey conducted in 2008 and will be used to evaluate net erosion and deposition over the past three years, and to estimate sediment transport rates at different locations within the study reach. 3.2 Data Collection Water Survey of Canada has been monitoring flows continuously since 1971, with some data records dating back to 1915. Tim Giles, a field geomorphologist from the Ministry of Forest and Range, surveyed the first 11 cross-sectional transects within the mid reach of Fishtrap Creek in 2004 (see Fig. 3.1). The cross-section locations were surveyed each year, and the large woody debris within the channel was documented. In 2006, the study reach was expanded to more than double its original length, adding an additional 8 cross-sections in the uppermost reach (cross-section 12 - 19) and 8 in the lowermost reach (cross-section A - H). Figure 3.1 shows the study reach along with the locations of these cross-sectional transects. Only the middle part of the reach, ranging from cross-sections 1 - 11 have been monitored from 2004 to 2008. Cross-sections A - H and 12 - 19 have been surveyed since 2006. Significant photo and written documentation supplemented numerous field visits to the field site during the spring and summer from 2004 to 2008, in which changes 28 3.2. Data Collection Figure 3.1: Fishtrap Creek study reach planimetric map showing 2008 channel bound- aries, thalweg location, bars, cross-sections installed in 2004 initially after the fire (light grey) and the additional cross-sections added in 2006 (black), and tracer stone launch lines (hatched black lines). 29 3.2. Data Collection in LW loading, channel morphology and watershed behaviour were recorded. Surface samples of bed surface and subsurface textures were also collected in 2006. Tracer stones were launched in 2006 and 2007. Philips (2007) described the lab and field techniques for creating and launching of tracer stones. Magnetic tracers were planted inside different sized stones. The stone size ranged from 22 to 91 mm and were chosen to bracket the estimated median grain size of the bed surface. These stones were painted four different colours and each colour group was placed at one of the launch line A, B, C and D (see Fig. 3.1) before the 2006 and 2007 freshet. These launch lines are also used to delineate sub-reach boundaries, in which each sub-reach (sub-reach 1-4) spans the distance between each subsequent launch line starting from the downstream end of the study reach to the upstream end. During high flows, these stones were displaced and subsequently located with a magnetic locator during low flows. When the stones were recovered, the size, distance travelled, and burial depth was recorded. The travel distance is used to estimate the average path length for bed material, which is used to estimate sediment transport rates (as described in Chapter 6). 3.2.1 Cross-Section and Planimetric Survey Survey Configuration Cross-sectional and planimetric surveys were performed using a Leica TCR 805 Power Total Station. The accuracy for the angular measurement is approximately 7 arc- seconds, which is estimated to result in a positional error on the order of 0.5 mm at a distance of approximately 15 m. Four benchmarks were created and surveyed in 2004 and 2006. In addition, left, right and occasionally center bank pins were estab- lished for each cross-sectional transect within the study reach. These benchmarks and cross-sectional pin locations were then used to triangulate the total station position (called a free station) in real space. For each free station, we required at least 3 known survey location for accurate triangulation. For the free-stations, the estimated horizon- tal positional error for the survey was no more than 1 cm and elevational error was no more than 1 mm. The total station was strategically set up at locations within the channel and floodplain, such that the maximum number of cross-sectional surveys could be surveyed with one free-station location. All cross-sectional and planimetric surveys were performed during low flows, ranging from 0.3 to 0.7 m3/s. 30 3.2. Data Collection Cross-Sectional Survey Once the free-station location was chosen and the cross-section pins were located, a 50 meter tape was strung from left bank to right bank in order to provide a straight line for the surveyors to follow. Survey points on the floodplain were spaced 1 to 3 m apart, while measurements in the channel were spaced no more than 30 cm apart in order to gather sufficient information of bed topography. Water depths were also recorded within the active channel. Morphologic features, including banks, vegetation edge (defined as the boundary between well developed vegetation, such as trees and shrubs, and poorly developed vegetation such as grasses), bankfull width (often edge of vegetation), waters edge, bank bottom, and bank top were recorded when possible. Due to the complex morphology of Fishtrap Creek, bankfull width was often hard to define. Field visits during peak flows, along with photos, helped provide a sense for the location of the bankfull width boundaries. Topographic Surveys Additional topographic surveys of the study site have been collected in 2006, 2007 and 2008 in order to map the key morphologic features and to create Digital Elevation Models (DEM’s) of the study reach. These surveys involve detailed mapping of LW pieces and jams, right and left bank location, thalweg location, as well as morphologic features including bars, riffles and pools within the active channel. Survey points were taken at increments ranging from 1 - 3 m apart depending on the complexity of the morphology. Bar tops, or the highest elevation of the deposit, along with the tip and tail of bars were identified and surveyed. Similarly, the deepest section of the pools were identified along with the tip and tail locations. The thalweg, identified as the fastest and deepest flow within the channel, was surveyed at approximately 1 meter increments for areas of more complex morphology, and at 2 to 3 meter increments on relatively featureless bed segments. LW was surveyed at each end in order to determine its orientation and length within the bankfull channel widths. Actively flowing avulsions within the floodplain during 2008 were also measured, however only the bank locations were surveyed. In order to generate a high precision DEM of the surface in 2008, bed elevations were surveyed at approximately 1 m spacing within the 400 meter study reach. Points were generally located where there was a distinct break in slope. As a result, in complex topographic areas, measurements were often taken at less than 1 m spacing, while in simple topography the spacing increased, but rarely exceeded 1 to 2 m. Nearly 4000 survey points were collected and used to generate the DEM for 2008. 31 3.2. Data Collection 3.2.2 Grain Size Distribution and Facies Map Surface grain size distributions have been collected at several representative locations within the channel since 2006. In 2008 the grain size distribution of the entire bed was mapped. The bed surface material in 2006 and 2007 was measured using a grid- by-number methodology, called Wolman sampling (Wolman, 1954). When possible, locations visited in 2006 were re-visited in 2007; however, channel changes during the 2007 freshet (discussed further in Chapter 4) made this impossible in some places. For these samples, large, morphologically uniform features were chosen. A 30 m tape was placed along a straight line and stones were sampled at spacing equal to two times the size of the largest stone. This spacing was chosen to avoid potential spatial autocorrela- tion and to ensure the independence of the grain size measurements (Rice and Church, 1998; Wolman, 1954). To ensure consistency in the data, 50 cm spacing was used for all Wolman samples. The b-axis (or intermediate axis) of each stone was measured using templates that measured half-phi size classes. Material less than 5 mm in diameter was classified as fines. Where possible, at least 100 stones were collected. Areas where suitable samples sizes could not be collected were avoided. As a result, Wolman sam- ples were limited by the ability to collect an adequate number of stones within a single morphologic unit. Grain Size Sampling for Facies Mapping In 2008, a more extensive sampling of the surface bed material was performed. All mea- surements were taken during flows of ∼0.3 m3/s, when the bed was considered stable. A facies map, which identifies distinguishable units of homogeneous bed surface texture, was created (refer to Fig. 5.7 in Chapter 5). The textural patch of each polygon was determined by classifying the relative abundance of the three primary grain sizes (sand, gravel and cobble) using the tertiary phase I and phase II figures from Buffington and Montgomery (1999a). The repeatability of facies mapping is very important, so identi- fying textures of sediment is a way to standardize measurement techniques (Buffington and Montgomery, 1999a). While in the field, the location of these polygons were drawn on a small scale planimetric map using banks, bar and LW locations within the creek as reference locations. The error associated with the location of the polygons is estimated to be ±1 m, limiting the size of sedimentologic units to no less than 1 m2. These units were then delineated using polygon shape files in ArcGIS 9.3. Measurements of surface grain size distributions for each facies unit were collected in the field and processed and analyzed in the lab. Wolman samples were performed on all large, morphologically similar units. In addition, digital photos of the bed were taken 32 3.2. Data Collection and processed using the Sedimetrics Digital Gravelometer (version 1.0). The Digital Gravelometer program is a tool for rapidly measuring the surface grain-size distribution of gravel using advanced image-processing techniques that identify and measure the size of grains in digital photographs (Sedimetrics Digital Gravelometer, 2009). While there are limitations in transforming a three dimensional object onto a plane surface, this technique is considered adequate for gravel and larger material when high accuracy is not required (Church et al., 1987). A frame of reference and control points are required in order for the program to evaluate GSD for a digital photo. Wood frames reinforced with L-brackets to ensure the frame remained square were constructed in the lab. Four screws with painted white tips were placed on inner most corners of the frame. The distance was measured between these screws in the x and y direction, and these points were used as control points for the processing of these images (Graham et al., 2005). Photos of surface material within the frame were taken using a Nikon Coolpix 8400 digital camera with a resolution of 8 megapixels. The camera was held 1 to 1.5 m over top of the sampling area, approximately parallel the the surface. Dry surfaces with little vegetation or debris were chosen. To ensure shadows from the angle of the sun did not interfere with photo quality, the entire sample area was artificially shaded by using a large fabric covered frame. Nearly 180 photos were taken of the study reach. At least one photo and/or Wolman sample was taken for each facies unit (with grains large enough to sample) to ensure quality and to evaluate the potential difference between the output from the digital gravelometer and Wolman samples. The surface digital photos were imported as JPEG images into the digital gravelome- ter program and converted to grey scale. Figure 3.2 shows an example of the steps from photo import to data output (A - D). The internal workings of the digital gravelometer, including the image processing, analysis and derivation of grain-size distributions, are presented in detail by Graham et al. (2005). Grains within the sample area often had multi-coloured bands of different minerals within the rock. This presented a problem for image processing by the Digital Gravelometer program. Individual grains with differing geology were often delineated into 4 or 5 stones by the program, while small size grains of similar geology and colour were often lumped together as one. In order to reduce this error, the grains that were causing problems in the analysis were delineated using using solid colour polygons in Adobe Illustrator (see A in figure 3.2). This technique was used on approximately 70% of the images taken. Given that modification of these images can reduce image quality (Sedimetrics Digital Gravelometer, 2009), the precision of this data processing technique may be less accurate than a Wolman sample. Therefore, in the event of major discrepancy between the grain size distribution provided from Wolman samples and that of the digital gravelometer, Wolman samples were preferred. 33 3.2. Data Collection Figure 3.2: Example showing the Digital Gravelometer input, processing and output data. A) is the initial photo with select stones that have been painted; B) shows the transformation of the image to grey-scale along with the proposed area delineation based on user calibration; C) shows the final delineation of stones based on area; D) shows the output generated. 34 3.2. Data Collection For facies located within the flowing channel (≥ 5 cm deep), Wolman samples and digital gravelometer analysis could not be employed, and the minimum, maximum and median grain sizes were estimated. Comparisons were also made in the field to similar units to confirm the estimates. Where sedimentological units were comprised primarily of sand size fractions, bag samples were collected and taken back to the lab for sieve analysis. Sieve analysis is a frequency-by-weight measure. Samples were dried, and or- ganics were burnt off in a sediment oven at 175◦C for at least 24 hours. A representative sample, such that the largest grain size was less than 1% of the total weight, was used for sieve analysis in the lab. Square-hole mesh wire sieves were used to evaluate the associated weight for each size fraction for the coarse gravels to very fine sand and silts. Larger grain sizes (≥ 22 m) were templated and weighed separately. A successful sieve sample has a final weight no less than 1% of the total initial sample weight. With the exception of one sample, the maximum error for this analysis was 0.02 %, while in most cases error was ≤ 0.01% Grain Size Analysis for Facies Mapping Cumulative frequency distributions for Wolman and sieved sand samples were con- structed4 and grain size associated with the 16th, 50th and 84th percentiles (D16, D50, D84) were calculated for each facies unit using the following equation from Bunte and Abt (2001): φx = (x2 − x1) ￿ yx − y1 y2 − y1 ￿ + x1 (3.1) where φx is the particle size of the xth percentile, y2 and y1 are the cumulative frequency percent values above and below the desired frequency and x2 and x1 are the particle size in φ-units associated with y2 and y1. The percentiles were then imported as attributes for each facies polygon in ArcGIS. Facies maps were categorized by these percentiles and a sorting coefficient (see equation 3.3). The D16, D50, and D84 values from the digital gravelometer output were similar to Wolman samples half of the time. The other half of the results seemed to be systematically higher than the results from the Wolman samples. This may in part be due to the influence of sand in the photos. The digital gravelometer has a difficult time differentiating sand particles, especially when wet. This may cause clumping of this sediment into larger stones which may cause the distribution of sediment to become skewed towards more coarser grain sizes. Particle-size distributions of fluvial sediment tend to roughly approximate normal distributions when particle sizes are expressed in φ-sizes (Bunte and Abt, 2001). These 4The digital gravelometer output data included cumulative frequency distributions, so no further analysis was required 35 3.2. Data Collection φ-sizes are calculated using the following equation: φ = − ￿ log(Di) log(2) ￿ (3.2) where Di is the diameter of the ith grain size. The sorting coefficient, which gives an indication of the spread of the data between 2 standard deviations, can be calculated as follows: Si = |φ84 − φ16| (3.3) where φ84 and φ16 are the φ-size diameters for the 84th and 16th percentile and Si is the absolute sorting coefficient. 3.2.3 Subsurface Sampling and Analysis Four bulk samples of the subsurface sediment in the study reach were taken in late August 2008, when flows were less than 0.2 m3/s. This technique requires substantial quantities of sediment to be excavated from the bed, which results in severe disturbance of the bed surface layer; as a result, this sampling technique was reserved for the end of the field season so there was no interference with bed surface measurements. The volume of material removed depends on the largest clast within the subsurface layer. Church et al. (1987) suggest a sample size in which the largest clast is no more than 1% of the total weight of the bulk sample. In this case, the sample size ranged from 400 to 600 kg. To ensure only subsurface material was being sampled, the surface layer was removed to the depth of the deepest-lying exposed grain prior to sampling (Church et al., 1987). All of the clast sizes, ranging from coarse cobbles (256 mm) to coarse gravels (22 mm) were sieved and weighed in the field, and a representative sample of the finer matrix material was analyzed in the lab. This sample was weighed in the field, dried and weighed again in the lab, and a water correction factor was determined for all the fine sediment. 3.2.4 LW Survey and Tagging: 2004-2009 In 2006, 2007 and 2008, the position of all the large wood5 (LW) within the channel bankfull boundaries was surveyed using a Total Station. Diameters and lengths of all surveyed LW were estimated in the field and given a class value according to Table 3.1. The range of class values were determined before field work commenced. They were defined based on the characteristics of the LW in channel and within the riparian 5For the purpose of this study, LW is defined to be any piece that is larger than 10 cm in diameter and 2 m in length 36 3.2. Data Collection Table 3.1: LW Diameter and Length Classes Used for Field Survey Class I.D Class Range Diameter 1 10 - 20 cm 2 20 - 40 cm 3 40 - 80 cm 4 80 - 160 cm Length 1 2 - 4 m 2 4 - 8 m 3 8 - 16 m 4 16 - 32 m floodplain. It is expected that these classes provided the best coverage of the LW within the stream. Decayed and severely burnt wood was also noted; however, only 4% of the total LW within Fishtrap Creek was listed as moderate to highly decayed or burned. In 2009, rather than surveying the position of large wood using the Total Station, a hip chain was used to delineate 10 m segments of stream reaches progressing from the lower end to the upper end of the study reach. The orientation, size, relative proportion of wood within the channel bankfull boundaries (presented as a fraction) were recorded, and a figure identifying each LW piece within the segment was drawn. Only the diameter and length classes, as well as a drawing showing its general structure and orientation, were recorded for LW jams. The number of pieces of a particular size was tallied to determine the total number of pieces per jam. Given that the freshet flows in 2009 were not large enough to change the channel morphology in 2009, and were incapable of moving significant wood within the channel boundaries, the results presented in this study are based on the synthesis of data collected from 2007-2009. Individual LW pieces and structural LW pieces in jams were given unique tag ID’s during 2008 and 2009. The tag ID specified the year in which they were tagged. Notes on year of recruitment, recruitment type and size also supplemented these ID’s. De- termining the year and type of recruitment, as well as the movement of individual pieces, required a field visit and collaboration with Mr. Tim Giles (MoFR). In addition, comparison of LW surveys and photos gathered from 2006 and 2007 allowed for a ret- rospective evaluation of new and/or moved wood. While the field site was not visited prior to the fire in 2003, we have assumed that the LW volume documented in autumn 2004 is representative of the pre-fire wood loads. 37 3.2. Data Collection Figure 3.3: LW orientation classes used for field data collection. The location of the circles at the end of each log segment refers to the “highest” point of the LW (which in most cases was the location of the root wad). LW Recruitment and Orientation In most cases, the recruitment of new trees to the channel following the fire was easily distinguishable into two modes: tree throw and bank erosion. Trees recruited from bank erosion were often associated with massive root wads, bank irregularities and local bank widening. Trees recruited by wind throw were usually smaller in diameter with no evidence of root wads. The orientation of wood in relation to flow was recorded and grouped into classes using the schematic shown in Fig. 3.3. The circles at the end of each line represent the higher end of the LW piece (generally corresponding to root wads when present). All orientations were recorded looking upstream. Parallel wood was grouped as 12 or 6 depending on the location of the root wad, perpendicular wood was classed as 3 or 9, while angled upstream and downstream wood was classed as 1.5 or 10.5 and 4.5 or 7.5 respectively. Angles that varied in between these points were classed according to the nearest class. 38 Chapter 4 Channel Morphodynamics This chapter evaluates the changes to the morphodynamics throughout the 5 years of study. This includes an analysis of changes in the channel morphology based on repeated surveys and of the bed material transport dynamics based on magnetic tracers placed in the stream for the 2006 and 2007 freshets. 4.1 Previous Studies: Post-Fire Flows and Suspended Sediment Most previous field studies of stream channel change have focused on the effects of varying the supply of water and/or sediment to a given reach: that is, they have focused on the exogenous drivers of channel change. These studies have generally shown that a decrease in sediment supply will result in channel narrowing and/or vertical incision for single-thread streams (Surian and Rinaldi, 2003), along with the development of a coarse surface pavement in single-thread channels with a gravel bed (Liebault and Piegay, 2001), and a reduction in the braiding index for braided streams (Chew and Ashmore, 2001). Increases in sediment supply tend to produce channel widening, aggradation, avulsions, and often a reduction in the sediment calibre (Lisle, 1982; Church, 1995; Pitlick, 1993; Sloan et al., 2001). Such increases also tend to produce a braided channel pattern, as well. Others have focused on the effects of variations in peak flow (e.g. Baker, 1977; Gardner, 1977; Church, 1995; Eaton and Lapointe, 2001), in some cases over relatively long timescales (Brewer and Lewin, 1998; Jones and Harper, 1998). However, large increases in peak flows are often accompanied by large increases in sediment supply, and thus produce similar effects, such as widening, aggradation, and changes in channel pattern (Lisle, 1982; Pitlick, 1993; Brown et al., 2001). The effect of an increase in stream discharge is strongly dependent on the recent history of flows and the channel state, and in some cases, large increases in discharge have produced relatively minor morphologic changes (Gardner, 1977; Desloges and Church, 1992; Eaton and Lapointe, 2001). Reductions in flows typically result in a gradual narrowing of the channel, usually accompanied by the progressive encroachment of riparian vegetation (Church, 1995; 39 4.1. Previous Studies: Post-Fire Flows and Suspended Sediment Gaeuman et al., 2005), and a reduction in the level of geomorphic activity (e.g. Church, 1995; Jones and Harper, 1998). Interestingly, some long-term studies have been able to associate oscillations in the channel morphology with cyclic changes in the flow regime and/or the sediment supply rates (Brewer and Lewin, 1998; Coulthard et al., 2000; Brown et al., 2001). There have been a few studies of the effect of endogenous drivers of channel change. Wathen and Hoey (1998) examined the impact of an endogenous sediment wave gener- ated by a large flood on the local sediment transport rates in the years after the flood at Allt Dubhaig, Scotland. Gaeuman et al. (2005) observed the effect of changing ripar- ian vegetation type and density on the channel pattern of the Duchesne River, Utah, while Brooks et al. (2003) documented a dramatic change in the morphology of the Cann River, Australia, as a result of disturbance of the riparian vegetation following European settlement. Millar (2000) also observed a change in channel morphology of a steep mountainous stream in British Columbia following removal of the riparian forest. These kinds of studies are retrospective investigations and have not been able to unam- biguously link the observed channel change to changes in endogenous drivers because the the exogenous changes are not known. Studies that explicitly consider forest fires have focused on erosion of hydropho- bic soils by overland flow (Inbar et al., 1998; Benavides-Solorio and MacDonald, 2001; Moody and Martin, 2001; Cerda and Lasanta, 2005; Desilets et al., 2006; Moody et al., 2008), the generation of debris flows from hydrophobic soils (Wells, 1987; Cannon and Reneau, 2000), or the aggradational landforms formed by post-fire increases in sediment supply to fluvial networks (Meyer and Wells, 1997; Benda et al., 2003; Legleiter et al., 2003). Furthermore, most of these studies tended to focus on fires that produce either a large increase in peak flows or a dramatic increase in sediment supply to the stream channel (see Wondzell and King, 2003). Thus, with the exception of some work looking at the large woody debris loads following a forest fire (Barro et al., 1988; Zelt and Wohl, 2004), the potential influence of the endogenous drivers, such as changes to the riparian vegetation, have not been well documented. At Fishtrap Creek, neither significantly elevated peak flows nor increased sediment loads have been detected since the fire (see Owens et al., 2006; Petticrew et al., 2006; Eaton et al., 2009). Figure 4.1 shows the daily average flows from March 20th (Julian day 80) to July 20th (Julian day 200) for both pre-fire and post-fire periods. In general, there seems to be an earlier onset of peak flows following the fire, however the peak flows are not noticeably different. Moore and Eaton (2009) found that the post-fire peak flows were close to or less than values predicted from the pre-fire regression, which may be attributed to the de-synchronization of melt between the disturbed and undisturbed 40 4.1. Previous Studies: Post-Fire Flows and Suspended Sediment areas within the basin. They also found that streamflows following the fire exhibited an earlier rise compared to pre-fire years, most notably during 2006 and 2007. Eaton et al. (2009) also found that the timing of the maximum instantaneous discharges occurred on average two weeks earlier after the fire. The earlier melt time and resultant peak flows are comparable to other snowmelt dominated regimes following wide-scale vegetation disturbance (e.g. Silins et al., 2009). It seems the early onset of flow noted in 2006 and 2007 was not evident in 2008. It is speculated that this delay may be the result of cooler spring conditions, which delayed the typically early melt seen in the previous two years. The 2008 freshet also differs in the absence of multiple peaks, rather the flow is characterized by one elongated peak (Fig. 4.1). Figure 4.1: Flow hydrograph showing daily maximum flows pre and post-fire for Fishtrap Creek. Post-fire discharges are shown by solid lines and dashed lines, while the daily average pre-fire flows from 1971 to 2003 are shown with dots for Fishtrap Creek. (source: Water Survey of Canada, Station 08LB024). Suspended concentration data for Fishtrap Creek were not collected prior to the fire. As a result, suspended concentrations in nearby Jamieson Creek are used as an index of the typical pre-fire sediment concentrations. Jamieson is not strictly speaking a control stream; however, comparisons can be made because the local climate, bedrock, surficial deposits and vegetation are similar. Results presented by Petticrew et al. (2006) show 41 4.2. Cross-Sectional Changes suspended sediment concentrations did not exceed normal levels, relative to Jamieson Creek. Data from Eaton et al. (2009) show that the first year after the fire, Jamieson and Fishtrap have almost identical ranges of suspended sediment concentrations. The largest recorded peak in suspended concentration in 2004 was 99 mg/L. In 2005, how- ever, the maximum recorded peak in 2005 was 650 mg/L, substantially higher than values recorded in Jamieson Creek (Eaton et al., 2009). The magnitude of this peak in suspended sediment concentrations during 2005 is indicative of channel change from bank erosion, but not of significant hillslope erosion (Eaton et al., 2009). Compared to previous studies evaluating sediment supply via hillslopes, the magnitude of response of suspended concentration following a severe wildfire was relatively small. The lack of hillslope erosion may in part be due to the limited amount of rainfall this region experienced in the years following the fire, or from the ash layer present on the surface immediately after the fire. Most wildfire case studies show erosion events as a result of rainfall intensities or large snowpack cover and earlier than normal snow melt. However, the earlier snow melt and resultant earlier peak flows documented at Fishtrap Creek did not result in erosion events in or around our study site. Despite this, Fishtrap Creek has become unstable since the fire. Eaton and Giles (2009) showed that changes in riparian vegetation could be sufficient to destabilize the stream channel, while Eaton (2008) used a rational regime model to demonstrate that Fishtrap Creek is much less sensitive to changes in discharge than either bank vegetation type or instream large woody debris volumes. Thus, the role of the boundary conditions within the reach, which constitute endogenous drivers of channel change, appear to be critical to understanding the changes documented at Fishtrap Creek. 4.2 Cross-Sectional Changes Changes in channel cross-sections have been analyzed quantitatively by calculating the volumes of net erosion and net deposition occurring over the course of each freshet. Net deposition and net erosion were identified as areas with an overall increase or decrease in depth at any given point in the cross-section. In this case, both net deposition and net erosion can occur within the cross-section at the same time at different points along the cross-section. Net change (m3/m) for each cross-section is the sum of net erosion and net deposition for each point along the entire cross-section. In this section, we describe the cross-sections using the following terms, based on the observed changes in stored sediment volumes: (i) static (wherein the cross-sectional profile remains virtually the same from year to year, with combined volumes of net erosion and net deposition < 1 m3/m), 42 4.2. Cross-Sectional Changes (ii) stable (wherein the cross-sectional profile does not change substantially, despite combined volumes of net erosion or deposition that range between 1 and 2 m3/m), (iii) dynamically stable (wherein erosion and deposition are approximately equal so that the bed morphology is modified and/or the channel migrates across the floodplain, and the combined volumes of net erosion and deposition range between 2 and 4 m3/m), and (iv) unstable (for which either net erosion/deposition is greater than net deposition/erosion by at least 1 m3/m, such that the mean bed elevation, the cross-sectional shape change and/or the channel migrates laterally. Unstable cross-sections are referred as either degrading or aggrading sections, depending on whether erosion or depo- sition is dominant.) The calculated volumes of net erosion and net deposition are presented graphically in Fig. 4.2, which includes a series of lines that define the static, stable, and dynamically stable cross-sections using the definitions above. The net changes in volume of stored sediment are presented in Fig. 4.3, which includes two lines that define the unstable aggrading and unstable degrading cross-sections. The analyses for the 2005 and 2006 freshets are restricted to the original study reach, which only includes XS 1 (at the downstream end) to XS 11 (upstream end). During the 2005 freshet, the channel banks remained unchanged at all of the surveyed cross- sections and changes to the bed morphology were modest, in comparison to changes documented in subsequent years. The cross-sections towards the upstream end of the reach (from XS 5 to XS 11) were generally static or stable (Fig. 4.2). The exception to this pattern was the net degradation occurring around a LWD jam at XS 9 (see Fig. 4.4), and the resultant net aggradation downstream of the jam at XS 7. The downstream end of the reach (XS 1 to XS 4) was primarily an area of net degradation. The cross- sectional profiles for two cross-sections in this area of net degradation (XS 1 and XS 2) are shown in Fig. 4.5. A visual comparison of the profiles for 2004 and 2005 shows that the observed changes during the 2005 freshet mainly comprised the development and modification of small in-channel bars. A LWD jam located immediately downstream of XS 1 (see Fig. 3.1) may have influenced the patterns of erosion and deposition at XS 1 and XS 2 during the 2005 freshet; but since it is located downstream of the original study reach, we do not have evidence to support this conjecture. The magnitude of the changes during the 2006 freshet were similar to those occur- ring in 2005. The jam near XS 9 continued to release stored sediment, producing net degradation at XS 8 and XS 9. The lower part of the study reach became laterally ac- tive, with bank erosion (and compensating bar deposition) occurring at the dynamically 43 4.2. Cross-Sectional Changes 10 0 10 1 10 0 10 1 10 0 10 1 0 50 100 150 200 250 300 350 400 450 10 0 10 1 10 0 10 1 10 2 10 0 10 1 10 2 St at ic St ab le D yn am ic al ly  S ta bl e St at ic St ab le D yn am ic al ly  S ta bl e St at ic St ab le D yn am ic al ly  S ta bl e N et  e ro si on  (m 3/ m ) N et  d ep os iti on  (m 3 / m ) N et  e ro si on  (m 3/ m ) N et  d ep os iti on  (m 3 / m ) N et  e ro si on  (m 3/ m ) N et  d ep os iti on  (m 3 / m ) Distance D/S (m) 19 18 17 16 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 AC BDEFGH Cross Section No. Changes during 2007 freshet Changes during 2008 freshet Changes during 2005 and 2006 freshets (2005) (2006) Figure 4.2: Estimates of net erosion and net deposition of bed material at each cross- section in the study reach. Quantities are expressed as volumes per unit length of channel (m3/m). Distances are measured from the upstream end of the study reach. 44 4.2. Cross-Sectional Changes 0 50 100 150 200 250 300 350 400 450 15 10 5 0 5 10 15 Distance D/S (m) N e t c h a n g e  ( m 3 /m ) 16.0 2004 05 2005 06 2006 07 2007 08 Net Erosion Net Deposition 19 18 17 16 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 AC BDEFGH Cross Section No. 1 -1 Figure 4.3: Estimates of net change in stored sediment (net deposition plus net erosion) along the study reach. The volumes are expressed as volumes per unit length of channel (m3/m). Distances are measured from the upstream end of the study reach. 45 4.2. Cross-Sectional Changes stable XS 3 and XS 4. The lower-most part of the reach aggraded (i.e., XS 1 and XS 2). During this freshet, the channel morphology generally switched from a plane-bed type channel to a riffle-pool channel in many locations, exhibiting (on average) higher bar tops and a deeper channel thalweg. Table 4.1 displays the estimated bar amplitude (defined to be the vertical distance from the thalweg to the bar top) for 2004, 2005, and 2006 and the percent change since 2004. Average bar amplitude increased by 63% be- tween 2004 and 2006. These changes in channel shape occurred in association with only limited bank erosion. Only XS 3, 4 and 10 experienced bank erosion during the 2006 freshet, and in all of those cases the recorded bank retreat was ≤ 1.5 m: the bankfull channel widths for all other cross-sections remained unchanged. 666 667 668 669 XS 13 663 664 665 666 XS 9 2004 2005 2006 2007 2008 661 662 663 664 XS F 661 662 663 664 XS G 5 10 m0 Horizontal scale E le v a ti o n  ( m  a s l) E le v a tio n  (m  a s l) Figure 4.4: Cross-sectional profiles for cross-sections F , G, 9, and 13. These cross- sections correspond to areas dominated by deposition during the study. For the 2007 freshet, we calculated net changes for the extended study reach includ- ing all cross-sections (from XS A at the downstream end to XS 19 at the upstream, see Fig. 4.2). The most pronounced change occurred in at the upper half of the extended study reach, between XS 9 and XS 19. The failure of a LWD jam (see Fig. 3.1) near XS 14 during the 2007 freshet increased the channel gradient between XS 14 and XS 19, and the channel responded by degrading (as shown in the profiles for XS 17 and XS 18 in Fig. 4.5). The volumes of net degradation occurring at each of XS 16, 17, 18, and 19 are one order of magnitude larger than the largest values recorded in 2005 and 2006. Most of the eroded material was redeposited immediately downstream of the 46 4.3. Bed Material Tracer Particle Dynamics Table 4.1: Bar Amplitudes for XS 1 to XS 11 in 2004, 2005, and 2006 XS No. Bar amplitude (m) % change 2004 2005 2006 2004 to 05 2004 to 06 1 0.80 0.41 0.91 -49 +13 2 0.48 0.54 0.67 +13 +38 3 0.32 0.35 0.77 +7 +138 4 0.76 0.62 0.57 -18 -53 5 0.21 0.20 0.46 -6 +114 6 0.16 0.33 0.53 +102 +227 7 0.54 0.60 0.88 +12 +64 8 0.66 0.79 0.85 +20 +29 9 0.35 0.96 1.21 +176 +246 10 1.08 1.08 0.46 0 -55 11 0.39 0.24 0.17 -38 -70 former LWD jam location as part of a coarse fan that formed within the channel (see Fig. 4.4) and in a sediment wedge — close to 2 m thick at its thickest point — upstream of the LWD jam at XS 9 (also shown in Fig. 4.4). The end result was net aggradation at all of the cross-sections between the jam at XS 9 and the jam that failed at XS 14. Downstream of XS 9, the cross-sections tended to be either stable or dynamically stable, except for XS 1, 2 and 3 which degraded, primarily as a result of bank erosion (note that while the locus of bank erosion shifted downstream from XS 3 and XS 4 in 2006 to XS 1 and XS 2 in 2007, it was part of an ongoing morphodynamic adjustment initiated in 2006). Bank erosion was also recorded at XS H in 2007 (these observations are based on detailed examination of the cross-sectional data, not all of which is shown). In 2008, the morphologic changes within the study reach were similar to (but less extreme than) the changes during the 2007 freshet (Fig. 4.2). Many of the cross-sections in the upper part of the reach (i.e., XS 15 to XS 19) continued to degrade, but at a slower rate than during the 2007 freshet. In response to this continued degradation, XS 8 to XS 14 aggraded as both the fan at XS 14 and the jam at XS 9 continued to trap much of the incoming sediment. The cross-sections in the study reach below XS 8 were generally stable or dynamically stable, though net aggradation occurred between XS E and XS H and net degradation occurred between XS A and XS C. 4.3 Bed Material Tracer Particle Dynamics Bed material tracers were placed in the stream in 2006 and 2007 in order to estimate the mean path length for bed material transport (see Chapter 6) and to gain insight into the event-scale bed material transport dynamics. The four locations at which the 47 4.3. Bed Material Tracer Particle Dynamics 666 667 668 669 XS 17 668 669 670 671 XS 18 661 662 663 664 XS 1 661 662 663 664 XS 2 2004 2005 2006 2007 2008 5 10 m0 Horizontal scale E le v a ti o n  ( m  a s l) E le v a tio n  (m  a s l) Figure 4.5: Cross-sectional profiles for cross-sections 1, 2, 17, and 18. These cross- sections correspond to areas dominated by erosion during the study. tracers were launched are shown in Fig. 3.1, with line A near the downstream end of the study reach and line D at the upstream end. In 2006, the tracer recovery rates were relatively high for the tracers launched at lines A, B, and C (83%, 83%, and 76%, respectively) and somewhat lower for line D (68%). The reasons for the lower recovery rate at line D are unknown, but may be related to the characteristic depth of erosion and deposition in this part of the study reach; but since the cross-sections in that part of the channel were only established during the summer low flow period in 2006, this speculation cannot be confirmed or refuted. In 2007, the recovery rates for lines A and B were similar to the rates for 2006 (73% and 76%, respectively), but substantially lower for lines C and D (63% and 44%, respectively). The lower recovery rates for lines C and D are likely attributable, in part, to the depth of aggradation immediately downstream of those launch lines: net aggradation in those areas commonly exceeded a depth of 1 m, which is the outer limit for detection of a magnetic tracer using the magnetic locator available to us. The displacement of the tracers in the stream during the 2006 and 2007 freshets has been analyzed with respect to tracer size and to the location at which the tracers were originally placed in the stream. The path length distributions, grouped according to tracer size class (i.e. 32-45 mm, 45-64 mm, or 64-91 mm), are presented in Fig. 48 4.3. Bed Material Tracer Particle Dynamics 4.6. In the left-hand panels of the figure, the probability that the path length for an individual tracer will equal or exceed a given value is plotted. For 2006, particle size clearly had an effect on tracer path lengths. The probability of a tracer moving at least 200 m (an arbitrary distance chosen for the purpose of comparison) had a mean value (based on all tracer sizes) of about 20%; the probability of travel distances exceeding 200 m for the smallest size class was about 35%; and the probability for the largest class was under 10%. Box plots of the path lengths (shown in the right-hand panels of Fig. 4.6) tell a similar story as the median, upper quartile, and maximum values all decline systematically with increasing tracer size. Finally, the mean path lengths decline systematically from 135 m for the 32-45 mm size range to just 64 m for the 61-91 mm range. 0.2 0.4 0.6 0.8 1 32 45 45 64 64 91 all data 100 200 300 400 500 600 0 200 400 600 0.2 0.4 0.6 0.8 1 32 45 45 64 64 91 100 200 300 400 500 600 Pr ob ab ili ty  (L s >  X ) Step length (m ) Downstream distance, X (m) Size class (mm) 2006 freshet 2007 freshet Figure 4.6: The path length distributions are presented for data grouped according to the size class of the tracers for 2006 (top row) and 2007 (bottom row). Data are presented as a probability distribution and using a standard box plot. The outliers in the box plot are those data that fall farther than 1.5 times the interquartile range from the median value. However, the same pattern did not hold for the 2007 freshet. During that event, while there were differences between the probability distributions and the box plots 49 4.3. Bed Material Tracer Particle Dynamics for each size class, there is no discernible systematic variation. The probability of a particle moving 200 m or more lies between 20% and 40% with an average value of 30% for all sizes. The lower, median, and upper quartile values for the path length distributions for each size class were all systematically greater in 2007 than in 2006 (Fig.4.6). Interestingly, the maximum path lengths do not seem to have changed. The change in tracer behaviour for the 2007 freshet (namely the absence of any significant size selectivity) indicates that, with respect to particle mobility, the two events were different. An analysis of the same data, this time grouped according to the initial launch line, gives insight into the spatial variations in path length. The probability distributions and box plots of path lengths for 2006 and 2007 are shown in Fig. 4.7. For 2006, the variability of the path length probability distributions for the tracers grouped by launch line was about as great as it was for the distributions for tracers grouped by size class. The longest path lengths were associated with tracers placed at line C, just upstream of the LWD jam at XS 9; and the next largest path lengths were associated with line D. Tracers launched at lines A and B had similar (to each other), relatively short, path lengths. The path length distributions for 2007 were similarly quite variable. In particular, the tracers originating at line D behaved quite differently than did the others. While the median path length for tracers from line D was similar to the medians for the other lines, the lower quartile is much higher and the upper quartile is much lower, giving a tighter and (but for one outlier) nearly symmetrical distribution, indicating that nearly all of the tracers placed at line D were eroded and then deposited in the aggradational zone. The other three lines were relatively similar to each other, having median and upper quartile path lengths noticeably higher than for the 2006 freshet. Notably, for 2007, the spatial variability of the path length distributions was greater than the size class related variability. Finally, the burial depth distributions were analyzed, grouped by launch line6. The burial depth distributions vary substantially, depending on where the tracers originated (Fig. 4.8). In 2006, tracers from line A were far more likely to be found upon the surface, or quite close to it. In contrast, tracers from line B were most likely to be deeply buried. Because tracers from line B also moved only a short distance downstream, the deeper burial depths are clearly associated with the growth of the point bar near XS 2 and 3 (shown in Fig. 3.1). If the upper quartile is taken as an approximate representation of the thickness of the active layer, then it varies from about 15 cm (i.e., ∼ 3 times the surface D50) to about 30 cm (∼ 6 times D50). The thickness of the active layer indicates 6Analysis of burial depths grouped by size class indicated that all sizes behaved similarly. 50 4.3. Bed Material Tracer Particle Dynamics 0.2 0.4 0.6 0.8 1 Line A Line B Line C Line D all data 100 200 300 400 500 600 0 200 400 600 0.2 0.4 0.6 0.8 1 Line A Line B Line C Line D 100 200 300 400 500 600 Pr ob ab ili ty  (L s >  X ) Step length (m ) Downstream distance, X (m) 2006 freshet 2007 freshet Tracer launch line Figure 4.7: The path length distributions are presented for data grouped according to launch location for 2006 (top row) and 2007 (bottom row). Data are presented as a probability distribution and using a standard box plot. 51 4.4. Bank Location Changes that Fishtrap Creek had a relatively active channel bed during the 2006 freshet. 0.2 0.4 0.6 0.8 1 Line A Line B Line C Line D all data 0.1 0.2 0.3 0.4 0.5 0.6 0 0.1 0.2 0.3 0.4 0.5 0.6 0.2 0.4 0.6 0.8 1 Line A Line B Line C Line D 0.1 0.2 0.3 0.4 0.5 Pr ob ab ili ty  (b ur ia l d ep th  >  Z ) Burial depth (m ) Burial depth, Z (m) 2006 freshet 2007 freshet Tracer launch line Figure 4.8: The depth-of-burial distributions are presented for data grouped according to launch location for 2006 (top row) and 2007 (bottom row). Data are presented as a probability distribution and using a standard box plot. The burial depths were somewhat greater on average during the 2007 freshet; but based on the range of upper quartiles, the active layer was not much deeper, varying from about 20 to 30 cm. One striking feature is that the tracers from line D, which were not buried very deeply in 2006, were the most deeply buried in 2007. Tracers from line A were again the least deeply buried, and were often found on the bed surface. 4.4 Bank Location Changes The formation of riffle-pool morphology is often a response of accelerated sediment delivery, in which the bars typically grow at the expense of pools. Amplitude increases provide evidence of this channel shift in 2006. On the other hand, reduced complexity of bedforms, often from the filling of pools, is often the result of significant inputs of sediment; this was the case in 2007 and to a lesser degree in 2008. 52 4.4. Bank Location Changes One of the most notable changes in the stream following the fire was the widespread occurrence of bank erosion. Changes in bank locations in 2006 were significant, in that the material supplied from bank erosion resulted in bedform adjustment throughout the river. However, they were minor compared to the bank erosion that occurred in 2007. Figure 4.9 shows the changing bank locations between 2006 and 2007, 2007 and 2008 and the cumulative change from 2006 and 2008. Following the 2007 freshet, the most significant bank changes occurred in the uppermost reach, specifically at cross-sections 13, 14, 17 and 18. Bank changes at cross-sections 17 and 18 are primarily a result of bank failures from bank undercutting. The 2007 freshet resulted in over 15 m of bank widening in this area. The vegetated bar in the upper reach near XS 18 has been excavated and the secondary channel on the right bank was abandoned in 2007 from the significant scour in this reach. No bank widening occurred around XS 16, and only minor widening occurred around XS 15 (≤2 m). XS 13 and 14 saw upwards of 18 m of bank changes in the form of aggradation (see Fig. 4.12). Bank changes from XS 9 down to the end of the study site during 2007 are modest. Most of the changes present in the middle reach on the left bank near cross-section 9 are local LW induced changes. In this case, the changes are generally ≤0.5 m. The lower reach shows only minor changes in bank locations (≤2 m). The 2008 freshet resulted in minor bank changes in the upper reach, while the middle and lower reach (from XS 9 downstream) saw more notable changes. Again, LW-forced bank widening is evident on the left and right bank between XS 9 and 12, but to a greater extent than in 2007. LW-induced undercutting and bank erosion is also evident on the left bank between XS 3 and 4. In the lower reaches, bank location changes in 2008 are greater than what was observed in 2007; however, most of these changes are a result of bar building. There is a distinct difference in the magnitude of bank changes upstream versus downstream of of XS 9. The cumulative changes from 2006 to 2008 show the majority of the bank change in the uppermost reach, followed by the middle and lower reaches. XS 15, 7, A, B and C have all seen very insignificant changes from 2006 to 2008, while XS 18, 17, 14, 13, 10 and 9 have all seen remarkable changes in bank locations. 53 4.4. B an k L ocation C h an ges Figure 4.9: Change in bank location from 2006 - 2007, 2007 - 2008 and the cumulative bank change from 2006 - 2008. The are shaded with a light grey shows the extent of the difference between each year, while the striped polygons show the locations of vegetated islands and/or bars.54 4.5. Changes in Bed Elevation, Slope and Creation of Secondary Channels 4.5 Changes in Bed Elevation, Slope and Creation of Secondary Channels Bed Elevation Evaluation of the bed surface change in elevation for the entire study reach was per- formed using a surface contour map that was generated using the 2008 survey points, in combination with the survey points from the 2006 survey (Fig. 4.10). A surface raster was created from the 2008 points using ArcGIS 9.3. The 2006 X and Y coordinates were located on this 2008 raster surface and the associated 2008 elevation was recorded. The data collected in the GIS attribute table had the original data in X,Y, Z06 for 2006 along with a new Z08 value which corresponded to the 2006 X and Y points that was projected on the 2008 surface. Z06 was subtracted from Z08 to find the change in elevation from 2006 to 2008 such that negative values are associated with erosion and positive values are associated with deposition. Bed elevation changes in 2008 were similar to the changes observed in 2007, but of smaller magnitude, so the 2008 map provides a good indication of the patterns also observed in 2007. The bank boundaries in this figure are associated with the 2006 banks, and the figure provides a clear pattern of the aggradation and degradation throughout the study reach, but not of the bank erosion patterns discussed above. 55 4.5. C h an ges in B ed E levation , S lop e an d C reation of S econ d ary C h an n els Figure 4.10: Surface contour map showing the locations of erosion and deposition between 2006-2008. Contours show 0.5 m intervals, with darker grey being erosion and lighter grey being aggradation. 56 4.5. Changes in Bed Elevation, Slope and Creation of Secondary Channels In the uppermost reach, degradation is prevalent, and there is evidence of more than 1.5 m of scour. Local aggradation dominates the middle reach, reaching upwards of 1.8 m in depth in some areas. In the lowermost reach, there is little change in the bed elevation. There is evidence of bar building, primarily along the left bank in the lower reaches; however, these are feature-specific and localized changes. Figure 4.11 shows the cross-sectional profile for cross-section 17 in the upper reach following the 2006, 2007 and 2008 freshet. Figure 4.11: Cross-sectional changes in cross-section 17 following the 2006, 2007 and 2008 freshet showing largest region of degradation. In 2006, the active channel only occupied approximately 7 m. However, in 2007, the channel expanded more than three times its original width. Almost 12 m of bank erosion on the right bank caused this channel to expand from 7 m to nearly 20 m wide. The active channel degraded by almost 1.0 m, while the bank retreated and scoured material to a depth of 1.2 m. This impressive degradation during 2007 resulted in complete abandonment of the secondary channel on the right floodplain (refer to Fig. 4.9). The morphology, while still in transition, has progressed from a plane featureless bed in 2006 to one that has greater lateral oscillation in 2007. During bankfull flows, the mid- channel bar, located from 7.0 to 15 m across the channel, is just barely submerged. The input of LW from bank erosion, in combination with the relatively low flows following the 2007 freshet, has allowed the bed to remain relatively stable. This is evident in the minor bed elevation changes seen following the 2008 freshet (Fig. 4.11). Cross-section 13, located on what we now call ‘the fan’, is one of the most morpho- logically interesting locations within the study site (Fig. 4.12). In 2006, the morphology 57 4.5. Changes in Bed Elevation, Slope and Creation of Secondary Channels Figure 4.12: Cross-sectional changes for cross-section 13 following the 2006, 2007 and 2008 freshet showing largest region of aggradation. of this area was typically more variable than in any other cross-section in the channel. This may in part be due to the confluence of the secondary channel that was active prior to 2007. The bankfull channel in 2006 spanned from 7 to 19 m and the thalweg was located at approximately 8 m across the channel. The 2007 freshet resulted in aggradation upwards of 1.5 m, and can be observed throughout XS 13. This deposition that occurred during the 2007 freshet caused the old active channel to aggrade by over one meter and the thalweg to shift to the left by 3 m (refer to Fig. 4.12). In 2008, the changes were similar, but less dramatic than in 2007. The thalweg continued to shift left by another 1.5 m, and the mid-channel bar deposit has built up on average by approximately 0.3 m (Fig. 4.12). During bankfull flows, this cross-section is nearly 25 m wide, while the average depth is only 0.1 m. The pattern of erosion and deposition seen in the upper and mid-reaches are not evident to the same degree in the lower reach (see Fig. 4.10). The reduced morphologic adjustment in the lower reach is a result of the log jam located at XS 9 (at approximately 694950 east and 5667520 north on Fig. 4.10). This log jam has become a sediment trap for most of the material transported from the above reaches, which has resulted in a sediment-starved system in the lowermost reaches. Figure 4.13 shows the cross-section across this large log jam. In 2006, minor aggradation and bar building is evident in this cross-section. However, the addition of new LW (discussed further in Chapter 5), and the introduction of sediment from scour in the uppermost segment during 2007, resulted in approximately 1.8 m of aggradation. In 2008, the bar deposit grew horizontally; 58 4.5. Changes in Bed Elevation, Slope and Creation of Secondary Channels however, with the exception of some in-filling on the left bank, there has been very little vertical bar building. The majority of this cross-section is bar deposits forced by the LW accumulation. In 2006, the channel was more uniform throughout. During low flows, the active channel on the right is now the only channel that is occupied. In 2006, the thalweg was located almost 8 m to the left, at approximately 11 m across the channel. Figure 4.13: Cross-sectional changes at cross-section 9 following the 2006, 2007 and 2008 freshet showing the aggradation associated with a large log jam. 59 4.5. C h an ges in B ed E levation , S lop e an d C reation of S econ d ary C h an n els Figure 4.14: Location of banks and bars from 2006 - 2008. 60 4.5. Changes in Bed Elevation, Slope and Creation of Secondary Channels Using the planimetric surveys from 2006, 2007 and 2008, we can relate the upstream and/or downstream progression of bar features to the changes in surface elevations between 2006 and 2008. Figure 4.14 shows the planimetric surveys from 2006, 2007 and 2008. These data were collected each year during low flows in late July; however, the flows were not identical. The discharge during the field survey was highest in 2008 and lowest in 2006, so the surveyed boundaries of the bars evident in 2008 is a conservative comparison to the size and extent of the bars surveyed in 2006. All vegetated islands present in 2006 (shown in Fig. 4.14 as white polygons) were scoured and/or buried in sediment. The entire study reach has seen an increase in the size and frequency of gravel bars. The middle reach, from the fan segment (XS 14) to the location of the jam at XS 9, has been characterized by a significant increase in the size and number of bars due to the increased aggradation in 2007. In the narrow reach upstream of the fan (between XS 13 and 16), the small bar present on the left bank (in the main channel) in 2006 was eroded following the break up of a small jam slightly downstream of this bar (discussed further in Chapter 5). In 2008, while the bars are typically in the same location, some of them have become more patchy. This may in part be due to the slightly higher discharge during the time of survey, and/or due to ongoing sediment re-working and movement following the widespread changes in 2007. Due to the limited sediment that has been transported downstream of the jam at XS 9, bar sizes and locations did not change significantly since 2006. In 2008, however, bar elongation and widening is evident on the right bank bar around XS 3, 4 and 5 and on the bank near bar near XS E and F. Changes in Slope and Avulsions The degradation and bank erosion that occurred in the uppermost part of the study site during 2007 has resulted large quantities of sediment supplied to the channel. Another response to the widespread scour and subsequent deposition during 2007, is local changes in channel gradients. The change in slope for each sub-reach is presented in Fig. 4.15. The longitudinal profiles presented in this figure represent the thalweg locations for 2006, 2007 and 2008. The sub reaches are divided based on tracer launch lines (refer to figure 3.1). at reach 4 there has been a steady reduction in slope, in which the average slope (calculated from 330 to 450 m upstream) in 2006, 2007 and 2008 was 0.027 m/m, 0.022 m/m and 0.018 m/m respectively. In reach 3, we see a similar scenario. The average slope in 2006, calculated from the upstream end of the log jam (located at 220 m upstream), was 0.034 m/m. In 2007 and 2008 however, the slope reduced by almost 1/3, to only 0.013 m/m. 61 4.6. Sedimentologic Response Figure 4.15: Sub-reach slopes showing longitudinal profile along the thalweg from 2006 - 2008. The aggradation, and thus substantial reduction in slope, has resulted in the for- mation of numerous secondary channels within Reach 3 (refer to figure 3.1). While this secondary channel was noted in 2006 and 2007, it has now become a permanent, continuously flowing secondary channel. There is another secondary channel that flows only during high flows from reach 2 to reach 1; however, this is most likely a wood in- duced feature since no significant aggradation has occurred in this area (see discussion in Chapter 5). 4.6 Sedimentologic Response Comparing Wolman samples taken during 2006, 2007 and 2008 allows for estimation of the changes in grain size distribution in areas where significant deposition and erosion was observed. In the uppermost reach, the surface coarsened from 2007 to 2008 (see Fig. 4.16 inset A). No Wolman samples were taken in 2006, so the bed surface grain size distribution before the major channel changes in 2007 cannot be compared. The coarsening of the surface in this reach is most likely from supply limited conditions in this reach, causing surface winnowing. At the fan, we find surface fining in 2007, followed by coarsening in 2008 (Fig. 4.16 inset B). The characteristically fine surface material in 2007 was likely a response to the increase in sediment supply. However, the bed material is becoming coarser in 2008, likely due to reduced sediment supply in 2008 and low flows. In this case, the 62 4.6. Sedimentologic Response Figure 4.16: Comparisons of the Wolman grain size distribution for select areas within the study reach 63 4.7. Discussion bed material seems to be reverting back to grain size distributions similar to the 2006 surface. The log jam-induced bar deposition is finer in 2007 than in 2008 (see inset C). Again, no data were taken in 2006, so the original bar grain size distribution is unknown. However, it is speculated, given the degree of aggradation in this reach, the surface in 2006 was probably coarser than in 2007 - similar to the fan reach (inset B). Finally, in the lower reach, we find that the surface in 2006 was coarser than in 2008, however no data was collected in 2007 (Fig. 4.16 inset D). The fining in 2008 may be in part due to the avulsion, which may reduce flow enough to cause deposition of finer grains along this bar. However, it is speculated that the fining in this reach is also influenced by LW. 4.7 Discussion Numerous studies have documented significant increases in hillslope erosion and runoff following fire. However, these changes have not been observed following the McLure fire. The morphologic changes in Fishtrap Creek are primarily a response to increased sediment supply from bank failures within the study reach. Peak flows occur earlier, but are not significantly larger than pre-fire flows. While suspended sediment concentrations are slightly higher, it is speculated that this is a response to localized bank failures rather than hillslope inputs. Surprisingly, however, given these unremarkable changes in watershed processes, the magnitude of channel morphologic response following the fire have been significant. The first documented change in channel morphology was the transition of the bed from that of a featureless bed morphology in 2004 and 2005 to a more laterally vari- able bed in 2006. According to Montgomery and Buffington (1997), the morphological classification of Fishtrap Creek prior to the 2006 freshet was typical of a plane bed morphology, in which the bed exhibited very little variation in depth across the active channel. The relatively small addition of sediment in 2006 from bank failures resulted in increased storage in bedforms, in which bars were built up and pools were scoured. As a result, Fishtrap Creek shifted to a more characteristically riffle-pool morphology, which is characterized by undulating beds with sequences of bars, pools and riffle reaches (Montgomery and Buffington, 1997). Evidence of the transition from plane bed to riffle- pool morphology can be seen in 2005 in some cross-sections, but is more obvious from the cross-sectional changes documented in 2006. The daily average peak flow in 2004 was only 4.56 m3/s., while the daily average peak flow in 2005 was the highest in the years following the fire, at 8.93 m3/s. During 2006, when most of the bed morpho- logic changes started to occur, the daily average peak flow was approximately equal to 64 4.7. Discussion bankfull discharge, 7.18 m3/s. The 2007 freshet daily average peak flow was a modest 6.16 m3/s. However, this freshet was responsible for the most remarkable morphologic channel re-configuration within the years following the fire. The most notable changes were observed in the upper reach and middle reach above jam 2. Bank failures and bed scour in the upper reach resulted in a large supply of material to the fan reach and jam 2. The vertical scour near XS 17 and 18 resulted in a significant decrease in bed slope and increases in bar amplitudes. Aggradation in XS 9, 10, 13 and 14 resulted in a significant reduction in slope and the formation of secondary channels during moderate to high flows. With the exception of a few cross-sections (e.g. XS 11, 12 and 17), bar amplitude changes in 2007 were less evident than in 2006, and often reduced due to deposition resulting in the smoothing of bedforms. The number of bars and their sizes increased significantly in 2007, while grains size distributions generally decreased in response to the increased sediment supply. The earlier onset of flows observed in 2004 to 2007 was not evident in 2008, likely a result of cooler weather. The degree of bank widening, aggradation and degradation within the channel was minor in 2008 relative to 2007. While there was evidence of bar building and minor bank widening, the most notable change in 2008 is the formation of a permanently occupied secondary channel within the middle reach. The reduction in sediment supply, and moderate average peak daily flows (6.83 m3/s) resulted in the redistribution and re-organizing the pre-existing deposits within the channel. As a result, increases in bar amplitudes in most cross-sections was observed. In many parts of the channel, bed surface grain size distributions are reverting back to pre-2007 characteristics, as the stream re-works and moves sediment around in an attempt to find a new equilibrium. The widespread occurrence of bank failure resulted in the recruitment of large quan- tities of LW, which has played an important role in the channel morphologic adjustment (discussed in depth in Chapter 5). The jam located at XS 9 provides an important mor- phologic divide within the study site. Upstream of cross-section 9, we see significant morphologic adjustments, while downstream in the lower reach, the changes have been minor in comparison. As a result, we do not see significant changes in bed elevation and/or bar sizes and numbers in any year following the fire. 65 Chapter 5 Instream Large Wood It is well understood that wood plays a critical role in the morphologic adjustment of river systems. Individual wood pieces and wood jams are integral to the form and function of small, intermediate and large alluvial rivers. Effects of stable pieces of wood on local channel hydraulics and sediment transport can influence the rate of bank erosion, create pools, or initiate sediment deposition and bar formation. At the reach scale, wood can influence the frequency of pools, channel roughness, stream competence, and overall channel reach morphology (Montgomery et al., 2003a). The function of LW within a channel is highly dependent on the size and frequency of LW pieces within the channel boundaries (Montgomery et al., 2003a). In jams, LW pieces that create a point of initiation for accumulation of smaller pieces, which can result in LW accumulation and jam building over time, are termed “key” pieces. A key piece can be defined the largest piece within a jam that acts as support for the main structure of the jam. Typically these pieces are oriented perpendicular to flow. LW pieces and key pieces in jams that are oriented perpendicular to channel flow result in the greatest influence on channel hydraulics (Gippel, 1995; Richmond and Fausch, 1995; Webb and Erskine, 2003), and can result in significant channel change (Montgomery et al., 2003a). This chapter will provide detailed information regarding the post-fire changes in wood loading and the current influence on Fishtrap Creek. Results will be compared to wood loads documented in disturbed and undisturbed basins. The recruitment rate, mobility and distribution of LW following the fire will be evaluated, and the influence of LW on stream hydraulics and channel morphology will be discussed. 5.1 Volumetric Estimation of LW and Error Length and diameter data were collected for all LW pieces within 10 m segments along the 400 m reach. These segments were then aggregated into larger 20 m segments since many LW pieces spanned more than 10 m of channel. LW pieces that crossed segment boundaries were divided equally between the two 20-m segments. Figure 5.1 presents the 2008 planimetric survey of the channel banks, location of LW (combined from 2007 to 2009), log jams and the location of each 20 m segment. 66 5.1. Volumetric Estimation of LW and Error Figure 5.1: Planimetric map of Fishtrap Creek study site in 2008 showing LW segment locations, LW and log steps from 2007-2009 and LW jam locations. Segments are identified by their distance upstream and associated segment number (e.g. 300 - 15). 67 5.1. Volumetric Estimation of LW and Error Each LW piece diameter and length was classed visually while in the field according to Table 5.1. The arithmetic mean value for each diameter and length class was used as a representative length and diameter value for each piece of wood identified in the field (Table 5.1). Table 5.1: Length and Diameter Classes and Representative Length and Diameter of Large Wood Class Length Class Representative Diameter Class Representative Range Length Range Diameter (m) Lc (m) (m) Dc (m) 1 2-4 3.0 0.10-0.20 0.15 2 4-8 6.0 0.20-0.40 0.30 3 8-16 12.0 0.40-0.80 0.60 4 16-32 24.0 0.80-1.60 1.20 Each piece of surveyed wood was assigned the converted value, and the volumes were calculated using a modified equation described by Gomi et al. (2001): VLW = π( Dc 2 )2LcP (5.1) In this case, VLW is volume in m3, Dc is the converted diameter in m, Lc is the converted length in m, and P is proportion of the tree within the bankfull width of the channel. Total LW volumes in jams were simply summations of each individual LW surveyed within the jam. LW Volume Error The LW in the channel was surveyed from end to end during 2006, 2007 and 2008 in order to get a length estimate for each LW piece. In ArcGIS 9.3, the two points associated with each end of the LW were joined with a line and the actual length (La) of the LW was recorded using the measure tool in ArcGIS. This allowed us to evaluate the accuracy of our class data for each piece of LW collected to the real lengths of the LW measured in ArcGIS. A total of 36 LW pieces (of all different size and orientation) along the study reach were measured and compared to the class data for this error analysis. All visual estimates of LW length and diameter were classed appropriately within these class ranges. The degree of error within these classes generally increased with increasing class range. On average, the error for class 1 and 2 was La= Lc ± 0.6 m, class 3 had an error 68 5.2. Longitudinal Influence of LW of La= Lc ± 1.9 m and class 4 had a total error of La= Lc ± 3 m. In this case, the proportional error (or δ L / Lc) for class 1 through 4 is approximately 1/5, 1/10, 1/6 and 1/8th respectively. In terms of volumetric LW errors, for a sample size of 36 LW pieces, the arithmetic mean underestimates the volume of LW by 0.14 m3, which is only a 1% error. 5.2 Longitudinal Influence of LW Increased bank erosion, in combination with the increased LW recruitment, has resulted in remarkable channel changes. The resulting influence on the active bed is evident in pool formation and frequency, localized scour and deposition, as well spatial variability in bed surface texture. Pool Formation and Frequency Figure 5.2 A shows the location of all the pools and the mechanisms driving pool scour along the longitudinal profile for the study site. The largest jams following the fire are located approximately 10 bankfull widths apart, while the newer and thus smaller jams (jam A and B) are only two bankfull widths apart (Fig. 5.2 A). There are approximately 1.25 jams per 100 m of stream. This value lies within the range of data values recorded by Wohl and Jaeger (2009). Pools forced by large woody debris account for 9 of the 11 large pools within the study site. Pool spacing is dependent on the distribution of wood and, as a result, is highly variable. On average, pools are spaced three bankfull widths apart, averaging 2.75 pools per 100 m. Montgomery et al. (1995) found that the spacing of pools in forested streams with large woody debris loading were highly variable. They found that increased wood loading resulted in more pools, averaging anywhere from 1 to 13 channel bankfull width apart. 69 5.2. L on gitu d in al In fl u en ce of LW Figure 5.2: A) Shows the longitudinal profile of 2008 bed surface and water surface elevation along the thalweg of the channel. Locations of pools created by log steps have been labelled as LS 1 - 3. Large, well formed jams are identified by hatched circles and labeled as JAM number 1-3. Smaller, newly formed jams are identified as JAM A-B and pool forming Large Wood pieces are labeled as LW 1- 3. B) Shows the total wood load within each segment (excluding 2009 wood load) and the proportion of wood within a particular orientation to flow (excluding jams) along the longitudinal profile. 70 5.2. Longitudinal Influence of LW The volume of wood per unit area of channel, referred to as wood load (m3/m2), is a useful quantitative measure for assessing the amount of LW within a channel across different systems (Bragg et al., 2000). Wood load is highest in segments 19, 10 and 6, which are associated with jams 3, B, and 1, respectively (see Fig. 5.2 B). The highest wood loads are not necessarily associated with the greatest variation in bed surface elevation. Section 19 has little variation in bed elevation, even though it is associated with the second highest wood load within the channel, whereas the area around segment 10 has some of the most variable bed topography. This may in part be due to the relative proportion of LW oriented normal or at a 45◦ angle to flow. Segment 11 exhibits a modest wood load; however, there is remarkable variation in bed topography in the form of aggradation upstream of and scour and pool formation downstream of jam 2. There are a number of log steps within jam 2 (LS 1 for example in Fig. 5.2 A) that have resulted in plunge pools and significant bank and bed scour within segment 11. Similarly, segment 4 has a relatively high wood load, primarily due to LW 1, which has resulted in evident variations in bed topography, most obviously in a large pool formation directly downstream of the LW. Jam 1 is associated with a very large pool directly beneath the jam structure. This may in part be due to the channel spanning, suspended nature of the LW in this jam. There has been localized scour beneath the jam, with little to no sediment deposition upstream or scour downstream (see Fig. 5.2). Very little bed topographic variation is seen in segments 7, 12, 16 and 17. These areas are also associated with comparably low values of wood loading. Almost half of the wood load in segments 7 and 16 are oriented parallel to flow7. Segment 17 has the overall lowest wood load in the channel. The fan section, segments 14 and 15, is characterized by bankfull widths ranging from 20 to 25 m, nearly double the average bankfull width of the study reach. As a result, the wood loads in this area are modest. Interestingly however, most of the wood found in these segments is situated normal to or at a 45◦ angle downstream or upstream to flow. As a result, there are notable variations in bed topography approximately 300 to 360 m upstream (refer to Fig. 5.2 A). Given the large bankfull widths, the resulting bankfull depths in the fan reach are remarkably shallow. As a result, this area is a deposition zone for water-bourne LW and fine wood (FW), most of which is transported through segment 17. The lower end of the study reach, where pool formation is prevalent but wood loading is low, snags and localized bank irregularities influence the channel topography rather than LW-induced pool formation. 7Pieces that are parallel to flow typically have reduced influence on localized erosion and deposition. 71 5.3. LW Recruitment and Movement 5.3 LW Recruitment and Movement Recruitment Type and Timing The size and rate of large wood supplied to streams is a function of the disturbance history, physiographic setting, size and type of river, as well as the underlying geology of a basin. The riparian vegetation along Fishtrap Creek is dominated by Lodgepole Pine, which have characteristically small diameters and heights compared to species on the Coast. The average diameter of the LW within the active channel is approximately 30 cm in diameter, or 60% of the average bankfull depth, while the average length is 8 m, or 66% of the average bankfull width. The total volumes of LW and type of recruitment over the 6 years of study at Fishtrap Creek are presented in Fig. 5.3. Figure 5.3: LW recruitment volumes from wind throw and bank failure for each year following the fire (excluding 2004) The total LW input from bank erosion and wind during the last five years comprised 41 and 9 pieces, respectively. With the exception of 2009, the LW recruited from bank erosion exceeded wind throw in all years. The 2007 freshet was responsible for 68% of the total LW recruitment over the last 5 years, of which over 90% was recruited via 72 5.3. LW Recruitment and Movement bank erosion. Minimal LW was recruited to the channel in 2005 and 2006 and 2008. In 2009, slightly more wood was recruited; however, 75% was recruited by wind. Wind throw does not play an important role in LW recruitment prior to 2009. In 2007 only three pieces, amounting to only 6% of the total volume of wood recruited following the fire, was recruited to the channel by wind throw. With the exception of 2009, each year the total volume of wood recruited to the channel due to bank erosion is higher relative to wind throw volumes. The total post-fire volume of LW recruited to the channel via bank erosion and wind throw is approximately 36 m3 and 4 m3 respectively. Volumes of wood recruitment from bank erosion exceeded the total input volume from wind by nearly 10 times. The total volume of wood recruited during 2007 was almost 21 m3, which exceeded the combined wind and bank erosion volumes from 2005, 2006, 2008 and 2009. Recruitment Rate and LW Movement As of August 2009, Fishtrap Creek had approximately 192 LW pieces within the bankfull channel width, of which approximately 50 (or 26%) were recruited in the years following fire. This amounts to an additional 40 m3 of LW to the channel. The average recruitment rate following the fire is approximately 8.3 pieces per year. The estimated recruitment rates for bank erosion and wind are approximately 6.8 pieces per year, or 5.9 m3/yr and 1.5 pieces per year, or 0.6 m3/yr respectively. The post-fire LW recruitment per unit length of stream over 6 years of survey data is approximately 9.9 m3/100 m. Bragg (2000) modeled the recruitment rate during and immediately after a fire, in which over 90 % of the total riparian trees were killed, to be 13.7 m3/100 m. This was followed by low recruitment loads (approximately 2.9 m3/100 m) 10 to 20 years after the fire, followed by a second peak of 15 m3/100 m 30 years following the fire. This model, however, does not take into consideration lag time from root decay. The first year after the fire at Fishtrap Creek, no wood was recruited. In 2005 and 2006, two and three years following the fire, only 1.0 m3/100 m of wood was recruited each year. These values are similar to Bragg’s (2000) model for LW recruitment in undisturbed channels. In 2007, the volume of wood recruited per 100 m of stream channel for Fishtrap increased significantly, reaching 5.5 m3/100 m — still under half of the recruitment volumes documented by Bragg (2000). If recruitment inputs continue to reflect inputs recorded in in 2008 or 2009, it can be expected that an average of 1.3 m3/100 m may enter the stream each year. There is a discrepancy in the predicted timing of wood recruitment from Bragg (2000) and the results presented from Fishtrap. It is evident that more field studies are required to 73 5.3. LW Recruitment and Movement better understand timing and quantities of LW recruitment rate following fire. However, in general, the recruitment rate in Fishtrap Creek is lower the first year following the fire than the modeled results predicted by Bragg (2000) and larger four years following the fire. Results documented by Minshall et al. (1997) found that wood contributed to the stream was highest three years after a fire of moderate intensity (in which at least 40% of the catchment was burnt, up to a maximum of 92 %) in second order streams. These results agree better with the LW recruitment seen at Fishtrap Creek compared to the model prediction by Bragg (2000). The differences in recruitment of LW could in part be due to the type and density of the riparian vegetation available to the stream, as well as the fire intensity within the riparian zone. The total wood load of pre-fire wood, post-fire wood and moved wood for each segment are presented in Fig. 5.4. Figure 5.4: The total volume of pre and post wildfire LW and LW that has moved within the channel since the fire, including 2005 - 2009 LW Post-fire wood loads are concentrated in the upper reach at jam 3 (between XS 17 - 19), near jam 2 and jam B and downstream of jam 1 (see Fig. 5.4). In fact, 90% of the wood load in jam 3 (segment 19) is post-fire LW recruited from bank failure during the 2007 freshet. Segments 4 and 5 are characterized by an an actively eroding cut bank, which as resulted in significant bank undercutting and LW recruitment. The middle of the reach, segments 12 to 14, and downstream of jam 2, A and B (segments 6 and 74 5.3. LW Recruitment and Movement 7), are generally characterized by stable, pre-fire wood; jam 2, in segment 11, is almost entirely comprised of pre-fire wood. This jam has become a sediment trap, which has resulted in significant burial of the key pieces within this jam. As a result, the pre-fire wood load is difficult to determine, and is most likely under-represented. The volume of new wood, found on the surface of the sediment, is only 1/8 of the total volume within the jam; the remaining 7/8 is old wood that was not entirely buried with sediment, mostly in the form of log steps (see Fig. 5.2 A) Given that the size of the LW recruited to the channel is small enough to be trans- ferred through a channel of this size, it is expected that LW would be mobile. However, it seems transport is limited by the morphologic complexity of the stream. The movement of individual pieces has not been measured directly. However, qualitative measurements based on repeated field visits provides information on readily observed LW movement through the study reach. In general, movement has been limited following the fire. The only case where movement of LW is evident is in the mid reach at segment 15 (on the fan) and in the lower reach at segment 8 (see Fig. 5.4), both incidences of movement were observed following the 2007 freshet and most likely associated with jam break-up (see section 5.5.2) Wood Loading The higher the wood load, the more it will affect the hydraulic roughness, sediment storage, localized erosion, and other geomorphic processes in the stream (Wohl and Jaeger, 2009). The total volume of wood within Fishtrap Creek, including jam pieces, was calculated and divided by the average bankfull width and study reach length using equation 5.2: WL = ￿n k=1(VLWSk) WbfLr (5.2) where n is the number of segments within the study reach, VLWS is the volume of large woody debris in m3 within each segment, and Lr is the total length of the study reach in m. With jams included, the average wood load for the entire study reach is 2.06 ×10−2 m3/m2; however, if jams are removed, the average wood load is 1.44 ×10−2 m3/m2. The wood load of the new wood that has entered the channel since the fire is approximately 0.78 ×10−2 m3/m2, while the old wood is 1.28 ×10−2 m3/m2. The new wood accounts for approximately 38% of the total wood load currently in the stream. As stream size increases, the ability for a stream to transport LW downstream also increases. Using data from Bridger-Teton National Forest, Bragg et al. (2000) found that the relationship between LW load and mean bankfull width can be evaluated as a 75 5.3. LW Recruitment and Movement negative exponential curve (equation 5.3). WL = 0.0561e−0.0843Wbf (5.3) WL is the wood load expressed as a volume per unit stream area (m3/m2), andWbf is the average bankfull width of the channel in m. For bankfull widths of 12 m, comparable to Fishtrap Creek, the typical volume per surface area predicted by this exponential function is 2.0 ×10−2 m3/m2. This model accurately predicts the wood load for Fishtrap Creek. Figure 5.5 compares the results using the Bragg et al. (2000) model to calculate wood loads within each segment at Fishtrap Creek compared to the observed wood loads calculated within each segment. Typically, the natural variability in wood load observed at Fishtrap Creek is higher. Figure 5.5: Wood load boxplot showing observed wood load for each segment within Fishtrap Creek and the predicted wood load from Bragg et al. (2000). The median value for wood load in Fishtrap is approximately 1.5×10−2 m3/m2, while Bragg’s model estimates a slightly higher median value for Fishtrap of 2.0 ×10−2 m3/m2. The observed wood load per unit area of stream in Fishtrap Creek has much greater spread than the model predicted for Fishtrap, suggesting that the reach scale variations in wood load as a function of channel area cannot be captured by Bragg’s model (which is 76 5.3. LW Recruitment and Movement based on average bankfull widths for multiple streams). The extreme values in Fishtrap Creek reach maximum wood loads of approximately 5 ×10−2 m3/m2 and minimum wood loads of 0.1 ×10−2 m3/m2, while Bragg’s model predicts extreme values of 2.7 ×10−2 m3/m2 and 0.5 ×10−2 m3/m2 for maximum and minimum loads, respectively. Fishtrap Creek is an unstable river that exhibits highly complex morphology. These complexities are likely responsible for the deviation between observed wood loads for a given Wbf in Fishtrap Creek versus Bragg’s model, which is based on more stable systems. The degree of movement of LW through the system in a typical intermediate stream is higher than that of the movement documented in FIshtrap Creek (Zelt and Wohl, 2004). LW gets caught up in the channel from reduced channel depth due to significant bank widening, which results in large wood loads in areas where bankfull widths have become significantly larger. Many researchers have found that wood load is negatively related to stream dimensions (Bilby and Ward, 1989; Robison and Beschta, 1990; Bilby and Ward, 1991; Richmond and Fausch, 1995), which agrees with Bragg’s exponential decay function. Rivers with bankfull widths ≥ 16 m typically have wood loads ≤ 1.0 ×10−2 m3/m2 according to Bragg’s model. Fishtrap has wood loads in segments 14 and 15 (where bankfull widths average 22 m) of more than double this (refer to Fig. 5.2). Typically, in stable rivers with larger widths, LW is easily moved through the channel, however, this is not the case in the fan reach. In addition, post-fire LW is not yet fully functioning in the stream, as many pieces are still propped on the banks and floodplains. As a result, wood loads in these areas are high, but bankfull widths are typically lower than what is seen in channels with similar wood loads. Segment 10 is characterized by the highest wood load within the channel (refer to Fig. 5.2); however, this segment has an average bankfull width of only 13 m. Bragg’s model predicts 1/3 the wood load in streams of this size, suggesting wood loads should be closer to 1.8 ×10−2 m3/m2. Similarly, Bragg’s model suggests smaller channels have the highest wood loads. However, this is not the case in some segments throughout Fishtrap Creek. Segments 1, 3, 16 and 17, which have the smallest bankfull widths within the channel (averaging about 9.5 m), have the lowest wood loads within the creek (refer to Fig. 5.2). Again, Bragg’s model is unable to account for this. In general, the model presented by Bragg et al. (2000) is not suitable for predicting small scale fluctuations in wood loads within a system. The variability found within a stream, especially one as unstable as Fishtrap, cannot be adequately measured by this model. However, taking the average bankfull width over the entire system, this model proves to be an effective way of estimating average wood loads. Table 5.2 presents wood loads from various disturbed river studies. On average, disturbed sites have less wood load than undisturbed sites. Richmond and Fausch 77 5.3. LW Recruitment and Movement Table 5.2: Wood Load Results From Studies on Similar Sized Streams in Disturbed and Undisturbed Watersheds, Including Fishtrap Creek. Stream Name and Location W̄bf Vegetation Disturbance Time Since Wood Load (m) Disturbance ×10−2 m3/m2 Florentina River, Italy 11.4 Dd D Current 0.2 Cordon River, Italy 5.4 Cf D Current 0.25 Code River, Italy 4.8 Dd D Current 0.75 Pettorina River, Italy 7.4 Dd D Current 0.1 Mack Ck, OR 9.1 Cf U n/a 8.12 Highland, U.K 4.9 Dd D current 0.88 N. St. Vrain Creek I, CO 7.9 Cf U n/a 1.23 N. St. Vrain Creek II, CO 6.7 Cf U n/a 0.24 N. St. Vrain Creek III, CO 14.0 Cf U n/a 0.29 N. St. Vrain Creek IV, CO 16.2 Cf U n/a 0.10 N. St. Vrain Creek V, CO 15.8 Dd/Cf U n/a 0.04 N. St. Vrain Creek VI, CO 11.1 Dd/Cf U n/a 0.0006 S. Fork Poudre River, CO 13.9 Dd/Cf U n/a 0.18 Baker Gulch, CO 7.0 Cf U n/a 1.06 Bowen Gulch, CO 8.0 Cf U n/a 0.95 Arapaho Creek, CO 10.2 Cf U n/a 1.71 S. Fork Gypsum Creek, WY 12.0 Dd/Cf U/L <10 ya 4.5 Ditch Creek, WY 11.2 Dd/Cf U/L <10 ya 0.40 Hoback River, WY 14.4 Cf U/L <10 ya 0.80 Mosquito Creek, WY 9.2 Cf U/L <10 ya 4.3 Fishtrap Creek, BC 12.0 Dd/Cf D 7 ya 2.06 **Results from Richmond and Fausch (1995), Gurnell et al. (2002), Comiti et al. (2006), Bragg et al. (2000) and Wohl and Jaeger (2009) Cf = Coniferous, Dd = Deciduous U = Undisturbed, L = Low levels of human disturbance, D = Disturbed 78 5.4. Hydraulic Influence of LW (1995) found that sites disturbed either naturally or anthropogenically during the 1900’s had significantly lower wood loads than in undisturbed sites. The total wood load at Fishtrap is one order of magnitude larger than all disturbed sites presented in Table 5.2. The wood loads in Fishtrap Creek are more comparable to the undisturbed streams. Fishtrap Creek exceeds the wood load for all but three undisturbed sites presented in Table 5.2. The old (or pre-fire) wood load for Fishtrap is estimated to be 1.28 ×10−2 m3/m2, which is still slightly higher than most reported wood loads for undisturbed streams. With jam wood removed, the total pre-fire wood load becomes 0.78 ×10−2 m3/m2. This result is more comparable to the results listed above. While Fishtrap Creek plots within the appropriate range of data values for debris loading verses drainage basin area presented by Gippel (1995), the average wood load for Fishtrap Creek, when jam wood is included, is on average considerably higher than the documented wood loads in other disturbed systems. However, there seems to be a very wide range of reported values for wood loads, ranging over 4 orders of magnitude, from 0.0006 ×10−2 m3/m2 to 8.12 ×10−2 m3/m2 for comparable sized streams to Fishtrap. The differences between studies may be due to the function of the size of stream, the type of wood being recruited, and other basin characteristics, as well as the definition of LW that is used for the study and whether or not LW jams are included in the study. 5.4 Hydraulic Influence of LW The orientation of LW in the stream determines the relative influence on flow. The hydraulic impact of LW can be measured by estimating blockage ratios for each piece of large wood. The blockage ratio is the a relationship between the LW projected area, which is a function the orientation of LW, and the flow area (Gippel, 1995; Gippel et al., 1996b,a). A blockage ratio was calculated for each individual piece within each 20 m segment. Equation 5.4 is used to calculated the LW projected area for each LW piece within the bankfull channel: A￿ = LcDcsin(σ)P (5.4) where A￿ is the projected area in m2, Lc and Dc are the converted length and diameter of LW in m, σ is the angle of LW orientation to flow and P is the proportion of wood in bankfull flow. The projected area is then divided by the area of the bankfull channel cross-sectional area (m2) to determine the blockage ratio (B). B = A￿ Af (5.5) 79 5.4. Hydraulic Influence of LW In this case, the flow area (Af ) is calculated using the average bankfull width and depth for the nearest cross-section within each 20 m segment (Lseg). Total blockage ratios for each segment were calculated by summing the A￿ for each piece of wood within the 20 m segment and dividing by the flow area multiplied by the segment length (Af · Lseg). Figure 5.6 shows the blockage ratio for new wood, old wood and the total combined ratio for all wood in the channel. Figure 5.6: A shows the blockage ratios for each segment excluding jams, divided into year of recruitment. B shows the total wood blockage for each 20-m segment and C and D show the post-fire and pre-fire wood blockages for each 20-m segment. The mean blockage ratio is approximately 0.27. Gippel et al. (1992) suggest that blockage ratios greater than 0.1 have significant impact on the water surface elevation (Webb and Erskine, 2003). Only six segments (segment 8 (jam A), 10 (jam B), 11 (jam 2), 14 and 15 (the fan)) are above the threshold of hydraulic significance, while the remaining 14 are below (Fig. 5.6). The blockage ratios of new wood are all below the threshold of hydraulic significance. The blockage ratio from the old wood contributes almost entirely to the hydraulic influence in segments 10, 14 and 15 (refer to Fig. 5.6). In this case, new wood contributes significantly to the overall hydraulic significance of wood in segments 8, 10 and 13. In areas where new wood loads are highest (segments 4, 80 5.4. Hydraulic Influence of LW 5, 18, 19 and 20), the blockage ratios (and by inference, the overall hydraulic significance of LW) are low (≤ 0.03). The low hydraulic significance of new wood in the channel is likely a result of its type of recruitment. Often LW pieces that are recruited by bank failure are quite long, becoming channel spanning pieces that do not interact with the channel. This is the case in segments 4 and 5. The one very large piece of wood recruited in the channel is propped up on the right bank, so only part of the wood is interacting with the flow. In addition, in areas in which large wood have been recruited and propped up on high bars, in the case of segments 18 and 19, the LW does not interact with flows below bankfull discharge. In this case, the widening of the channel in 2007 resulted in lowered water depths, which in turn reduces proportion of time the LW can interact with flows, especially when flows are below bankfull discharge. LW Induced Sedimentology The influence of LW on flow hydraulics results in the formation of bedform features such as pools and bars, as well as variations in the bed surface texture, or sedimentology. Typically grain sizes are coarser in the centre of the channel or the thalweg, where the flow velocities are highest (see Fig. 5.7), and finer along the channel margins where the velocities are lower. Fine deposits are also abundant where wood loads are high, typically around jams. The abandonment of the secondary channel on the right bank, which used to flow from segment 19 down to segment 15 (refer to Fig. 5.9), is a result of the widespread degradation that occurred between XS 17 and 19. The abandonment of this channel has resulted in a large stagnant pool where the secondary channel entered the main channel. This area is also characterized by large volumes of LW. As a result, the material in this area is very fine. Pools with characteristically fine material are also found near most avulsion entrances and exits (see Fig. 5.7). Segments 3 and 4 are characterized by significantly high wood loading (refer to Fig. 5.2) and highly variable, characteristically fine facies units (see Fig. 5.7). Only a small part of the LW influ- ences the channel bed, which has resulted in an overall lower estimation of hydraulic influence (see Fig. 5.6). However, branches have been retained (which is not included in the calculation of hydraulic roughness), which has had a significant influence on the deposition of fines on the right bank, upstream of this LW. In addition, the avulsion upstream between XS 3 and 4 is likely to have reduced the transport capacity of the stream from a reduction in flow, which may also causes more fines to be deposited in this area. The channel morphologic adjustment that occurred in 2007 left the channel in a state of disequilibrium. Typically, jams have a definite pattern of sediment grain size 81 5.4. Hydraulic Influence of LW Figure 5.7: Facies map of the 400 m study reach, categorized by D50. Bars and thalweg are also included on the map, along with the sorting coefficient of each facies, LW seg- ment locations, cross-section locations, avulsions and the location of the 2006 secondary channel, log jams and identifiable morphologic units (riffles and pools). 82 5.4. Hydraulic Influence of LW distribution upstream and downstream of the structure such that a finer than average surface is prevalent upstream and a coarser than average surface is prevalent down- stream. However, identifying a definite pattern of surface fining and/or coarsening at any of the major jams is complicated by the influence of newly formed secondary chan- nels, in combination with the massive and widespread deposition that has occurred, especially around jam 2 (bed material transport influence on surface textures will be discussed in Chapter 6). There are signs of surface coarsening downstream of jam 2; however, the proximity of jam B has resulted in variable and generally finer texture patches within the lower section of segment 11 and upper section of segment 10 (refer to Fig. 5.7). If we combine the effects of jam 2 and B, the material downstream in segment 9 is overall coarser than the material upstream of these jams in segment 12. A similar scenario is seen at jam A (segment 8), in which the bed material in half of segment 8 and all of 9 is finer than in segment 7. Patterns of coarsening downstream and fining upstream are not evident in jam 1 or 3. Jam 3 is the result of massive bank failures resulting in LW recruitment over a very short period of time. The jam is cur- rently sitting on a large mid-channel bar - which is likely still intact from the addition of LW which has stabilized this bar feature. As a result, the variable facies textures evident at XS 17 (refer to Fig. 5.7) are likely an artifact of the bank material which has not yet been re-worked by flows high enough to interact with this material. Figure 5.8 shows the relation between number of facies as a function of the number of LW pieces (including Jams) in each 20 m segment. The complexity of the bed surface, or the total number of facies per unit area, as a function of number of LW pieces shows a positive trend, with an R2 value of 0.78 (refer to Fig. 5.8). These results agree well with field studies performed by Buffington and Montgomery (1999b) in which they found the number of facies patches in a reach to be positively related to the frequency of wood obstructions. Jam 3 in segment 19 is the only area that does not follow this trend (see Fig. 5.8), reducing the power of this trend to an R2 value of 0.46. The hydraulic significance of the wood in segment 19 is low (refer to Fig. 5.6), which suggests that while the loading is high, the wood in this jam is not fully functioning in this segment. Segment 18 shows relatively high facies complexity relative to the number of LW pieces; however, as mentioned previously, this is a result of degradation that occurred in this area, and not a function of hydraulic influence of this jam. Segment 6 (jam 1) also exhibits high sedimentologic variability, with high wood load and low total blockage (refer to Fig. 5.6 and Fig. 5.2). The facies complexity in this segment is most likely a function of the relative age of this jam within the stream. On average, segments characterized by 80% of their LW by volume that is angled downstream and/or upstream and perpendicular to flow (segments 10, 14 and 15) have the highest blockage ratios and 83 5.4. Hydraulic Influence of LW Figure 5.8: Relationship between the number of facies units as a function of the LW numbers within each segment (excluding segment 1 as it was not part of the study reach facies mapping) correspond with a relatively high number of facies units per unit channel length (refer to Fig. 5.2, Fig. 5.6 and Fig. 5.8). Jam 2 (segment 11) and the fan (segments 14 and 15) are characterized by moderate wood loads; however, almost all of the LW is pre-fire wood angled perpendicular and/or upstream or downstream to flow, which results in high blockage ratios (≥ 0.2) (refer to Fig. 5.4 and Fig. 5.6). As a result, these areas are characterized by the highest number of facies per unit length stream within the channel (refer to Fig. 5.7). However, facies complexity in this reach is complicated by the avulsion that has been created upstream, which has resulted in a reduction of transport capacity and thus more sediment deposition. Likewise, segment 12 has characteristically high facies units per channel length with very small wood loads, which is likely a function of the right bank avulsion (refer to Fig. 5.7) and of the influence of jam 2 downstream. Segment 10 is the location of jam B, which is characterized by LW oriented perpendicular or downstream to flow (see Fig. 5.2). This jam has been building since the fire; however, over 50% of the LW is pre-fire wood that has larger hydraulic influence on flow. In addition, this jam is also the exit site for the upstream avulsion, which likely results in more variable bed surface facies distribution. The relatively high surface complexity compared to the overall number of LW found in segments 3, 7 and 84 5.5. Large Wood Debris Jams 9 are likely more influenced by these avulsions and less by the LW. Segments 20, 17, 16 and 2 all exhibit low sedimentological complexity, which relates well with their overall low wood loads and hydraulic blockage (refer to Fig. 5.6). 5.5 Large Wood Debris Jams Older jams, characterized primarily by pre-fire wood, have significant sedimentologic influence within Fishtrap Creek. Not only do jams provide important ecologic habitat, they are also an important physical component of channels in forest landscapes (Abbe and Montgomery, 2003). The sedimentologic response of these jams can significantly influence fish habitat and spawning grounds for numerous fish species. However, the overall function and influence of jams on stream hydraulics and channel morphology is a function of their size, orientation, structure, integrity and age. 5.5.1 Accumulation Rate of LW in Jams Abbe and Montgomery (2003) developed a classification scheme, in which 10 types of woody debris accumulations were identified based on the number, size and orientation of key pieces, structure and presence of racked and loose wood, mode and rate of recruit- ment, and their overall influence on channel morphology. Their findings are based on LW accumulations in the Queets River basin in the Olympic Mountains in NW Wash- ington. This classification scheme allows for a better understanding of the influence a jam may have on local morphology and also reach-length morphologies. There are three well developed jams within Fishtrap Creek (jam I.D 1, 2, and 3), and two smaller, newly created jams (jam I.D. A and B) (refer to Fig. 5.1) which have been classified here according to Abbe and Montgomery (2003). The average accumulation rate of LW within these jams is determined over the six years following the fire. Jams 1, 2, 3 and A each span one 20 m segment, while jam B only occupies half of segment 10 (approximately 10 m). The accumulation rate over the six years for each jam is given by the following equation: Ār = Vn Ajt (5.6) where Ār is the average accumulation rate in m3/m2/yr over five years, Vn is the volume of new wood in m3, Aj is the surface area on the channel which the jam occupies (m2) and t is the time since disturbance. The area was estimated using the nearest cross- section to the jam, and the average LW length within the jam. The area was also cross referenced in GIS using the measure tool based on visual inspection of the area with 85 5.5. Large Wood Debris Jams Table 5.3: Fishtrap Creek LW Old and New Jam Characteristics and Jam Types (From Abbe and Montgomery, 2003). Jam I.D and Jam type Volume No. of New LW Aj Ar location (m3) pieces (%) by volume (m2) (m3/m2/yr) 1 (Segment 6) Combination- 9.7 17 2 40 0.0008 Valley Jam 2 (Segment 11) Combination- 8.3 26 13 60 0.003 Flow Deflection 3 (Segment 19) Bank Input 13.2 29 91 100 0.02 In Situ A (Segment 8) Combination 4 6.5 79 40 0.013 Valley Jam B (Segment 10) Combination 8.9 6 37 40 0.014 Valley Jam the densest LW accumulation. Table 5.3 lists characteristics of the new and old jams within Fishtrap Creek, along with the accumulation rate and jam type based on Abbe and Montgomery (2003) LW accumulation classification system. According to Abbe and Montgomery (2003), combination jams have portions of wood that are in situ and portions of wood that are loose water-borne pieces. These jams can form an effective barrier, deflecting flows around the structure or completely impounding the channel. With the exception of jam 3, all jams can be categorized as combination jams. Jam 3, a bank input in situ jam, is typically made of wood that has not moved from the original position since recruitment into the channel. The LW in these jams lies only partially within the bankfull channel while the remainder lies on the banks and adjacent hillslopes (Abbe and Montgomery, 2003). As a result, the LW only acts as a partial flow obstruction, which results in local effects on channel morphology. All jams have new wood present, with jam 1 and jam 3 having the least and most new wood respectively. The accumulation rates of LW in jams 1 and 2 are 0.0008 and 0.003 m3/m2/yr respectively, while A and B have accumulation rates of 0.013 and 0.014 m3/m2/yr respectively. Over 79% of the LW in jam 3 is new since 2007, resulting in accumulation rates of 0.02 m3/m2/yr. Jam 3 has seen little to no movement of key pieces. Jams 1, A and B can be further classified as valley jams. According to Abbe and Montgomery (2003), LW within valley jams span the entire width of the channel, 86 5.5. Large Wood Debris Jams often constricting the majority of the cross-section. The key pieces in valley jams are typically inclined less than 30◦ to the bed and can result in flow diversion that can lead to further recruitment from localized lateral bank erosion (Abbe and Montgomery, 2003). Jam 2, on the other hand, can be classified as a flow deflecting jam since it does not span the entire width of the channel. These jams typically consist of racked and loose debris that can be delivered during high flows and often contribute to channel complexity (Abbe and Montgomery, 2003). 87 5.5. L arge W ood D eb ris Jam s Figure 5.9: Study site in 2006 and in 2007-2009 showing the location of all LW and jams within each segment 88 5.6. Discussion 5.5.2 Break-up and Movement of Jams The function of jams, both hydraulically and ecologically, depends primarily on their age, structure and integrity. Typically, the movement of LW pieces within the stream has been relatively insignificant over the past 5 years. The same can be said for pieces within jams. Figure 5.9 shows the distribution of LW along Fishtrap Creek study site for 2006 and 2007-2009. The location of jams 1 and 2 have not changed; however, there has been some movement within jam 1 (Fig. 5.9). The key pieces within the jam seem to be rotating, while the racked and loose pieces do seem to be moving downstream slightly. The area around jam 2 has experienced significant bank widening, but there has been little movement of key pieces within the jam. Most of the LW identified in 2006 has been covered with sediment; however, the new pieces that have been recruited since 2007 have accumulated on the left bank bar. This sediment is acting as a stabilizer in this jam, however the racked material that has accumulated against the large root wads are free to move downstream with the next large flood event. The jam previously identified as jam C in 2006 has broken up and the LW was deposited approximately 15 m downstream on the fan (refer to Fig. 5.9). Along with the addition of wood downstream, the break up of this jam resulted in a pulse of fine sediment that was transported downstream. The volume of material that was moved from upstream during the 2007 freshet was approximately 6 m3 (Fig. 5.4). There has been minor movement and shifting of wood elsewhere in the channel, including some floating pieces that moved from jam 2 to segment 8; however, there has been no further evidence of jam removal or breakup. 5.6 Discussion Timing of LW Recruitment From Bank Failures It is evident that LW plays a major role in the bed topography and surface textures found within Fishtrap Creek. However, this is compounded by the complex channel morphology, irregular banks and bars, and by the significant degree of channel recon- figuration that occurred during 2007. Given that this stream is currently in transition, reach scale conclusions on the influence of LW are difficult to make; however, local ef- fects of LW on sedimentologic and topographic variations are discernible. LW and jams in Fishtrap Creek are responsible for increased surface texture variability in, around and under jams, as well as upstream and downstream of individual logs. The number of wood-induced pools is high within this reach compared to natural, undisturbed streams and disturbed streams. The pattern of fining upstream and coarsening downstream of LW jams are not evident in all jams within the study reach; however, these patterns 89 5.6. Discussion of fining and coarsening are expected to become more defined as jams characterized primarily by post-fire wood become more hydraulically influential on the stream bed. In addition, the channel will need to establish a new state of equilibrium in response to the creation of these newly formed, and also permanent, secondary channels. Recruitment of LW in Fishtrap Creek has been dominated by bank erosion, primarily in 2007. Volumes of LW recruited from wind throw have been significantly less compared to the number of bank failures the first 6 years following the fire. With regard to intense, stand replacing wildfire, it can be expected that standing dead trees in the riparian zone will follow a decay function similar to the root decay and vegetation re-growth model presented by Benda and Dunne (1997). Their model combined the rate of decay of roots, with the re-growth of vegetation to see the relationship of relative root strength for a Douglas Fir forest in the Oregon Coast Range. The negative exponential curve of root strength over time is calculated using the following equation: D = exp(−ktn) (5.7) where D is the dimensionless relative reinforcement during decay, k (0.5 yr−1) and n (= 0.73) are constants, and t is time since disturbance in years. The rate of increase in root strength from vegetation re-growth follows a sigmoid curve and is expressed by the following equation: R = [(a+ b)exp(−ft)]−1 + c (5.8) R in this case is the dimensionless relative reinforcement due to regrowth; while a, b, c and f are empirically derived constants with values 0.95, 19.05, -0.05 and 0.25 yr−1 respectively. Summing the root decay and vegetation re-growth functions, we can find the relative root reinforcement for any year following the fire. Figure 5.10 shows the negative exponential decay function for root decay, the sigmoid function for root re-growth and the net bank strength from the combination of the two functions. This model predicts rapid reduction in root strength with time following widespread death of vegetation. This negative exponential curve suggests that the relative bank strength will be half of it’s original value two years after a fire and virtually non-existent after 30 years. However, vegetation re-growth (shown as root re-growth curve) gradually increases following the fire. As new vegetation emerges, the relative bank strength slowly increases, however at a slower rate than root decay. The resulting relation, shown as net bank strength in Fig. 5.10, shows a bank strength minimum 4 to 5 years following fire. This lowered bank strength resides for 3 to 4 years until vegetation re-growth has a more pronounced effect on the bank strength. The Benda and Dunne (1997) model of root decay and vegetation re-growth agrees 90 5.6. Discussion Figure 5.10: Rate of root decay and vegetation re-growth based on Benda and Dunne (1997) well with the occurrence of bank failures (resulting in tree recruitment) documented at Fishtrap Creek. The minimum relative bank strength for Fishtrap was likely in 2007, four years following the fire, when approximately 31 of the 50 LW pieces recruited following the fire were recruited via bank failures. Prior to the 2007 freshet, only minor channel change and wood recruitment was documented (Phillips, 2007). Since 2007, the LW input has been reduced significantly. While Benda and Dunne (1997) suggested the lowered bank strength should reside for a few years, we do not see evidence of this at Fishtrap Creek. In fact, we see far less LW recruitment in 2008 and 2009, even though daily average flows in 2008 were almost identical to 2007 (daily average flows in 2009 are unknown; however, field visits during high flow suggest flows were no higher than 2007 or 2008). This limited recruitment of LW may be in part due to the reduced number of trees that could be potentially recruited, in combination with the prevalence of small shrubs and tree re-growth in the under-story which may have increased bank strength (re-growth was facilitated by hydro-seeding initially following the fire). Recruitment Rate, Movement and Loading of LW Compared to Other Studies The average size of LW supplied to Fishtrap Creek is 8 m long and 0.3 m in diameter; as a result, the potential for movement of large wood debris in Fishtrap Creek is limited 91 5.6. Discussion not by the size of wood, but by the complex morphology and relatively high wood loads. Once in motion, a piece of wood will not have long to travel before something impedes its path. The diameter of log to depth of water ratio, Dlog/dw, for Fishtrap Creek is 1.67. According to Braudrick et al. (1997), a log will continue to float until the depth drops below half of the log diameter. The aggradation seen within the study reach since the fire, most prevalent in the fan during the 2007 freshet, has resulted in shallower bankfull depths. The flow depths on the fan are lower than the average diameter of wood within the channel, in some cases, as shallow as 0.1 m. As a result, LW transported through the channel is deposited within this area, and the total volume of wood within this reach is high compared to other segments. Compared to studies documenting relatively high mobility of in-stream wood, Fishtrap has had very little wood movement within the channel since the fire. Movement that has occurred, occurred due to extreme morphologic adjustment, not because of reduced integrity from fire. On average, material moved approximately 15 m downstream. Sub-reach Organization, Deposition and Hydraulic Influence of LW The disorganized deposition and accumulation of wood within the fan, in combination with very shallow bankfull depths and propensity for wood to accumulate perpendic- ular to flow, has resulted in significant hydraulic roughness elements from relatively high blockage ratios. In general, however, new wood recruited following the fire is less functional in the channel. Older wood has more impact on the stream bed and on flows. New wood from bank failures is often channel spanning and does not influence the bed or flow as much as older pieces that have had time to break down and become part of the bed. The only incidence of new wood resulting in high blockage ratios occurs when accumulation of these pieces forms small jams (jams A and B in section 8 and 10). In the middle reaches, where older, more stable wood with higher blockage ratios has been accumulating, finer surface textures and highly variable surface grain size distributions have resulted. The same scenario is occurring in some segments within the lower reaches where wood loading is relatively high compared to bankfull width. On average, segments characterized primarily by pre-fire wood that have more than 50 % of their LW by volume angled upstream/downstream or perpendicular to flow, have the highest blockage ratios and correspond with the highest number of facies units per unit length of channel. The only place where high wood loading is not reflected in surface texture variability is in segment 19 by jam 3. This is likely because the jam is perched on a mid channel bar, in which there has been no deposition and thus no mechanism for sedimentologic diversity. In addition, flows since the fire have not been large enough to interact with the sediment stored within the jam, so there is very little interaction of 92 5.6. Discussion the LW with the bed facies. The surface textures around jam 2 are highly variable due to the complexity of this jam and surrounding morphology. In 2007, this jam was almost completely covered with sediment. Perpendicular, low lying logs, and the presence of log steps within jam 2 have resulted in the formation of plunge pools and hydraulic jumps. As a result, the material within segment 11 and, to some degree, segment 10, is characteristically fine. Compounding this effect, a newly formed jam only 8 m downstream further complicates the patterns of surface textures within these segments. Material that may move over and/or around jam 2 is impeded by this new channel spanning jam. As a result, we see significant deposition of fines in this region, and migration of the left bank bar into the area directly upstream of this jam. Downstream of jam 2 and B, we see typical surface coarsening and reduced variability of surface textures. The integrity and function of jams are of critical importance when evaluating the overall morphologic and surficial textural response. The minor surficial response seen at jam 3 suggests this jam is not functioning within the channel - other than to stabilize the bar in which it is sitting. It is expected that this jam will become a functional jam when the flows are high enough to shift the wood and move it downstream, in which case, the bar deposits below the jam will likely be entrained and transported downstream. Jam 2 is a fully functioning jam. While the accumulation rate is reduced by the massive amount of material within the matrix of the key pieces, it is expected to remain a functional jam for some time. Jam 1 is functioning within the channel, however the evidence of movement within the structure suggests the integrity of this jam will be reduce and become less functional in the near future. Patterns of surface fining from avulsions, in which flow and thus transport capacity is reduced within the channel, are not obviously distinguishable from the surface response from LW forcing at this time. This may be due to the complexity of the channel morphology, in combination with the state of disequilibrium of this creek. As a result, LW has been the primary factor for localized changes in bed textures. As this stream reaches a new equilibrium, in which these avulsions may become permanent secondary channels; the reach scale surface textures within the creek may shift and equilibrate as a result. 93 Chapter 6 Bed Material Sediment Transport 6.1 Introduction Using measurements of morphologic change in a stream, it is also possible to esti- mate event-scale bed material sediment transport rates (Neill, 1987). Researchers have been able to estimate sediment transport rates using repeated cross-sections (Goff and Ashmore, 1994; Martin and Church, 1995; Paige and Hickin, 2000), analysis of aerial photographs (Ham and Church, 2000), and digital elevation models (Stojic et al., 1998; Eaton and Lapointe, 2001). This approach — often referred to as “the morphologic method” — can be used to reconstruct event-scale transport rates or the average rates over several decades (e.g., McLean and Church, 1999). One way to apply the mor- phologic method is by assuming that a typical path length (Ls) can be applied to the total volume of erosion (Ve) or deposition (Vd) measured within a reach (c.f., Eaton and Lapointe, 2001): this will generally lead to an underestimation of actual sediment trans- port rates because it does not account for scour and compensating fill of the channel bed at the same location between surveys, and there are relatively large uncertainties associated with the morphologic method (see Ashmore and Church, 1998, for a review of sources and magnitudes of error). Using the observed volumes of net erosion/deposition at Fishtrap Creek, event-scale bed material transport rates and patterns will be evaluated. Estimates of bed material transport rates for Fishtrap Creek using the cross-sectional data have previously been presented by Eaton et al. (in press). Since the accuracy of estimates of the topographic changes depends greatly on the spacing and location of the cross-sections within the study reach, the cross-section method can result in significant bias. This error propa- gates into the evaluation of sediment transport. These estimates will be used as a point of comparison for estimates based on DEM analysis that uses all of the available topo- graphic data to generate bed material transport estimates. The DEM method estimates the total volumes of erosion by calculating volumetric changes using DEM’s for the bed surface before and after each freshet. 94 6.2. Volumetric Estimation of Erosion and Deposition 6.2 Volumetric Estimation of Erosion and Deposition Cross-Section Method Two methods were used to determine the volume of erosion and deposition within the study site for the 2007 and 2008 freshet. The cross-sectional method used year-to-year topographic changes for each cross-section, in which areas of erosion and deposition for each transect were calculated using a cell size of 0.1 m. Total volumetric changes, in m3/m, were determined by summing the areas of net deposition and erosion over this linear grid. Vertical changes of less than 0.10 m were filtered out so that the reported changes are minimum estimates of the actual net erosion or deposition. These values were then applied to a segment that extended halfway to the cross-sections immedi- ately upstream and downstream to get an estimate of the total volume of erosion and deposition for this segment of channel. DEM Method The DEM method used a simple model to subtract surface elevations before and after the 2007 and 2008 freshet. These changes in elevation were measured on a 0.10 m × 0.10 m grid in order to find the total volumes of erosion and deposition within a 10 m DEM segment that brackets each cross-section (see Fig. 6.1). In the middle and lower reach, (XS 1 -12 and XS A - H), where cross-sections are spaced at 10 m increments, the DEM method corresponded well with the distance estimated by the cross-section method (see Fig. 6.1). This allowed for direct comparison between the two methods. In the upper reach (XS 13 - 19), cross-sectional spacing is 20 m. As a result, additional 10 m DEM segments were delineated between each of the cross-sectional transects. A detailed topographic survey of the channel bed was performed in 2008. Topog- raphy was surveyed with a point-to-point spacing of about 1 m between the channel banks, but not upon the floodplain surface. A DEM of the 2008 topography by gener- ating a TIN surface8 from the topographic survey data using the ArcGIS 3D analyst in ArcGIS 9.3. In 2006 and 2007, fewer topographic points were collected (with a mean point-to-point spacing of about 1.6 m) for the purpose of documenting the morphologic boundaries (i.e. bars, banks, and thalweg locations) and the cross-sections, not generat- ing an accurate topographic DEM. In order to avoid the error associated with generating DEMs of the 2006 and 2007 topography from a relatively sparse data field, we used the 2008 topography DEM to estimate the net change that occurred at each topographic 8A TIN surface is typically used for high-precision modeling of smaller areas, such as in engineering applications, so linear interpolation using a TIN was the method chosen for this data set. 95 6.2. Volumetric Estimation of Erosion and Deposition Figure 6.1: Locations of sub-reaches A, B, C and D used for evaluation of sediment transport rates for 2007 and 2008 using the XS and DEM method. Locations of 10 m DEM segments correspond with XS sections, except for in the sub-reach D where additional 10 m segments are added between XS locations. 96 6.3. Evaluation of Error Using the DEM Method survey point in 2006 and 2007. DEMs were then fit to the point file containing the net topographic change points, from which the volumes of net erosion and deposition of the channel bed were estimated. Since there is relatively sparse topographic data on the floodplain surface, it is not possible to accurately estimate volumes of net erosion due to bank retreat using the DEM analysis. Bank-related erosion was estimated by calculating the areal extent of bank erosion from the planimetric surveys of the bank location in 2006, 2007 and 2008 and calculating a representative bank height from the survey cross-sections that pass through area of bank erosion. In the rare case when both bank recession and bar building along the banks were evident in one 10 m segment, a combined average was used to estimate of the volume of erosion and/or deposition in the segment. 6.3 Evaluation of Error Using the DEM Method Three different methods of surface interpolation were performed to determine the ac- curacy of each surface interpolation method for the Fishtrap Creek data set: i) linear interpolation, ii) nearest neighbour interpolation and ii) biharmonic spline interpola- tion. Linear interpolation creates a surface from X, Y, Z points by triangulation (TIN networks) which are then turned into a surface (this method was used for the analysis for this chapter). The nearest neighbour is a non-adaptive algorithm (treats all pixels equally) that simply selects the value of the nearest point, and does not consider the values of other neighbouring points at all. This method is the simplest and requires the least amount of data processing time and can often result in jagged boundaries and poor interpretation of complex surfaces. Biharmonic spline is the most complex interpolation method which retains the most image information after an interpolation. The bihar- monic operator is used for minimum curvature interpolations of irregularly spaced data points, which in three dimensions, corresponds to multi-quadric interpolation (Sandwell and Texas Univ., 1987). Images of the surface topography generated by each method, for elevation differences and the 2008 topography, are presented in Fig. 6.2. Upon inspection, the linear and biharmonic spline surface interpolation methods appear to show believable and com- parable results in both cases (surface of differences and 2008 topography), while the nearest neighbour method is the most different, and provides the least accurate sur- face topography (see Fig. 6.2). Table 6.1 presents statistics describing the differences between the interpolation methods. These statistics are approximate measures of how precisely the topographic surface is defined by the data. The differences between in- terpolation methods are generally normally distributed. There is no detectable mean 97 6.3. Evaluation of Error Using the DEM Method Table 6.1: Error Analysis for Three Different Surface Interpolation Methods: Linear, Biharmonic Spline and Nearest Neighbour Surface fit to net Surface fit to change data, 2007-2008 topographic data, 2008 (Lin. - B.H.S.) (Lin. - N.N.) (Lin. - B.H.S.) (Lin. - N.N.) Mean diff. (m) -0.002 0.001 0.005 0.000 S.D. of diff. (m) 0.079 0.110 0.038 0.076 Diff. Quantiles (m) Q5 -0.111 -0.149 -0.051 -0.124 Q10 -0.078 -0.103 -0.033 -0.088 Q25 -0.033 -0.043 -0.011 -0.039 Q50 -0.001 0.000 -0.002 0.000 Q75 0.029 0.042 0.020 0.041 Q90 0.071 0.105 0.046 0.090 Q95 0.109 0.156 0.072 0.126 Lin. = Linear fit, B.H.S. = Biharmonic Spline fit, N.N. = Nearest Neighbour fit bias introduced as a result of the choice of interpolation method, since both the mean difference and the median (Q50) of all the difference distributions are effectively zero. However, there are interpolation-related local differences. For the net change surface, the linear and biharmonic spline methods differ by less than ± 11 cm for 90% of the grid cells in the DEM (based on the Q5 and the Q95 of the difference distribution), while the linear and nearest neighbour methods differ by less than ± 15 cm for 90% of the cells. The 2008 topographic surface is based on data with a higher spatial density. For this surface, the linear and biharmonic spline methods produce surfaces that are very similar with differences that are less than ± 6 cm for 90% of the cells. The statistics for the differences between the linear and nearest neighbour surfaces for the 2008 topogra- phy are similar to those for the net change surface. Based on the visual inspection of the topographic surfaces, the nearest neighbour can be excluded as a possible method of interpolation (refer to Fig. 6.2). Not only does the surface look incorrect, but the spread of the data is consistently higher for both surface interpolations. In this case, either linear or biharmonic spline interpolation seems to provide accurate results for this surface. While one model cannot be proven to be better, considering the complicated 98 6.3. Evaluation of Error Using the DEM Method algorithms and data processing time required for the biharmonic spline method, the linear method is preferred in this case9. 9Error analyses of different surface interpolation methods were also performed using the data points from 2007. The topography points had the same distribution as the difference between 2007-2008, so either surface would be adequate for estimation of volumes of erosion and deposition. 99 6.3. E valu ation of E rror U sin g th e D E M M eth od Linear Interpolation 6.9498 6.95 6.9502 6.9504 6.9506 x 105 5.6674 5.6674 5.6674 5.6674 5.6675 5.6675 5.6675 5.6675 5.6675 5.6675 5.6675 x 106 Nearest Neighbour Interpolation 6.9498 6.95 6.9502 6.9504 6.9506 x 105 5.6674 5.6674 5.6674 5.6674 5.6675 5.6675 5.6675 5.6675 5.6675 5.6675 5.6675 x 106 Biharmonic Spline Interpolation 6.9498 6.95 6.9502 6.9504 6.9506 x 105 5.6674 5.6674 5.6674 5.6674 5.6675 5.6675 5.6675 5.6675 5.6675 5.6675 5.6675 x 106 Student Version of MATLAB Easting (UTM)Easting (UTM) Easting (UTM) No rth in g (U TM ) Surface of Differences between 2007 - 2008 Surface Topography for 2008 Linear Interpolation 6.9498 6.95 6.9502 6.9504 6.9506 5.6674 5.6674 5.6674 5.6674 5.6675 5.6675 5.6675 5.6675 5.6675 5.6675 5.6675 x 106 Nearest Neighbour Interpolation 6.9498 6.95 6.9502 6.9504 6.9506 x 105 5.6674 5.6674 5.6674 5.6674 5.6675 5.6675 5.6675 5.6675 5.6675 5.6675 5.6675 x 106 Biharmonic Spline Interpolation 6.9498 6.95 6.9502 6.9504 6.9506 x 105 5.6674 5.6674 5.6674 5.6674 5.6675 5.6675 5.6675 5.6675 5.6675 5.6675 5.6675 x 106 660 660.5 661 661.5 662 662.5 663 663.5 664 í í í í í 0 0.1 0.2 0.3 0.4 Meters MASL Easting (UTM) Easting (UTM) Easting (UTM) No rth in g (U TM ) x 105 Figure 6.2: Surface interpolation using three methods: Linear, biharmonic spline and nearest neighbour. All three interpolation methods used to generate surfaces for the differences in elevation between 2007 and 2008 and for the 2008 topography. 100 6.4. Volumetric Net Change 6.4 Volumetric Net Change Three different quantities were calculated for each segment of the channel: (1) the net deposition occurring during each freshet, (2) the net erosion occurring during the freshet, and (3) the net change in sediment storage, given by the sum of the net deposition and the net erosion. These values were combined to estimate the the total quantities for each freshet, and subtotals for four sub-reaches which represent the part of the reach downstream of the four tracer launch lines (see Fig. 6.1). For the 2007 freshet, the XS and DEM methods produce similar estimates of net deposition (1264 m3 vs 1227 m3, respectively) and net erosion (1258 m3 vs 1250 m3, respectively) for the entire reach (see Table 6.2). However, there are larger discrepancies between the two methods for the sub-reaches, particularly for the net deposition estimated in sub-reaches C and D, where the discrepancy between the two methods exceeds 100 m3. For the 2008 freshet, the two methods predict significantly different volumes of net deposition for the entire reach (745 m3 for the XS method versus 485 m3 for the DEM method), with the largest discrepancies occurring in sub-reaches B, C and D. In contrast, net erosion estimates are similar in 2008 for both methods, except for sub-reach C, where there is a moderate discrepancy between the XS and DEM methods (86 m3 and 137 m3, respectively). The data have also been analyzed on a section by section basis. The net changes (i.e. net deposition - net erosion) for the 2007 and 2008 freshets, estimated using both methods, are plotted in Fig. 6.3. The magnitude of net change for each zone differs significantly between the two methods. The most prevalent difference is found between XS 13 and 14 and at XS 9. The additional measurement between XS 13 and 14 estimated by the DEM method resulted in net aggradation two times that captured by the XS method using only XS 13 or XS 14. In XS 9, the opposite is true. In this case, the XS method overestimates the total aggradation by a factor of two compared to the DEM method. Downstream of jam 2, the XS overestimates net change in sub-reach C and A, while under-estimating net change in sub-reach B (see Fig. 6.3). In 2008, we see modest deviations in the magnitude of the net change within each zone between the two methods. The largest difference is found at XS 18, between XS 18 and 19 and at XS 8 and 9. The XS method underestimates the aggradation around XS 18, resulting in a maximum underestimate of approximately 5 m3/m2. The XS method underestimates the total degradation downstream of the jam, while overestimating the total aggradation within the jam. Downstream of jam 2, the XS method overestimates total aggradation in all cross-sections but XS A, B and C, with the differences in E, F and G being the highest. A plot of the change in elevation, net change per bankfull width in m3/m2/m, for 101 6.4. Volumetric Net Change Table 6.2: Estimated Net Volumes of Erosion and Deposition for Sub-Reach A-D Using the XS and DEM Method for the 2007 and 2008 Freshets. Net Deposition (m3) 2007 Freshet 2008 Freshet Sub-reach XS Analysis DEM Analysis XS Analysis DEM Analysis A 46.4 35.6 75.8 24.3 B 51.9 48.8 121.1 53.3 C 370.4 217.8 146.2 68.0 D 796.0 924.6 402.0 340.0 Total 1264.7 1226.7 745.0 485.7 Net Erosion (m3) 2007 Freshet 2008 Freshet Sub-reach XS Analysis DEM Analysis XS Analysis DEM Analysis A 39.5 85.5 107.7 73.8 B 99.6 70.3 48.7 63.7 C 101.2 122.2 86.0 136.7 D 1018.1 972.4 219.6 229.0 Total 1258.3 1250.4 462.0 503.2 Table 6.3: Estimated Net Change in Sediment Storage for Sub-Reach A-D Using the XS and DEM Method for the 2007 and 2008 Freshets. 2007 Freshet 2008 Freshet XS Analysis DEM Analysis XS Analysis DEM Analysis sub-reach Lr (m) Vd − Ve (m3) Vd − Ve (m3) Vd − Ve (m3) Vd − Ve (m3) A 54.9 7.0 -50.0 -31.9 -49.5 B 83.2 -47.7 -21.6 72.4 -10.4 C 79.9 269.1 95.6 60.2 -68.6 D 187.2 -222.1 -47.7 182.4 111.0 102 6.4. Volumetric Net Change Figure 6.3: Estimates of net change for the study reach using the XS and DEM analysis methods for the 2007 and 2008 freshets. Zones of light grey show areas where the cross- section method overestimated net change, while dark grey show areas of underestimated net change. Sub-reach extents are shown with arrows and labelled as A-D between the two plots along with the XS locations listed above. 103 6.4. Volumetric Net Change Figure 6.4: Relationship between the net change per bankfull width in m3/m/m for the DEM method versus the XS method. 30% error bars are included, with extreme outliers identified and labelled. 104 6.4. Volumetric Net Change each cross-section and associated DEM segment (Fig. 6.4) shows the magnitude of the difference between these two methods identified in Fig. 6.3. According to Fig. 6.4, the areas with the largest discrepancies between the two methods for the 2007 event are at cross-sections 14 and 9. The DEM method estimates nearly two times the amount of aggradation estimated by the XS section method in XS 14. This suggests that the aggradation between the cross-sections is being significantly underestimated, and that the aggradation is in fact much greater than what is being captured by XS 14. The aggradation at jam 2 (XS 9) was overestimated by the XS method by nearly three times in 2007 (see Fig. 6.4). It is likely this difference is a combination of two factors: i) the widespread bank recession from localized flow deflection around jam 2, which was not adequately captured by XS 9 (see Fig. 6.5), and ii) XS 9 crosses directly over a localized aggradation point which is not indicative of the aggradation occurring over the entire 10 m segment (especially downstream of XS 9). A similar scenario is evident in 2008. In this case, however, the DEM method suggests XS 8 and 9 are characterized primarily by erosion, while the XS method suggests aggradation is greater (see Fig. 6.4). In 2008 there is more localized bank retreat which was not captured by XS 9, in addition to pool deepening and bed scour upstream and downstream of XS 9 (see Fig. 6.5). Erosion in segment 8 is localized around a small jam (jam B) which has resulted in two newly formed pools along the right bank. Neither of these pools is captured by XS 8, thus again resulting in an underestimate of total erosion for this segment. In general, segments 7, 8 and 9 are characterized primarily by erosion which was not accurately captured by the XS method (refer to Fig. 6.3). The most contrasting results are found in sub-reach A during 2007, where the XS method suggests XS A, C, and D are characterized by net deposition, while the DEM method suggests erosion is more prevalent (see Fig. 6.4). In this case, it is likely the XS method underestimates the degree of bank scour occurring within sub-reach A during 2007, resulting in a lower estimate of total erosion within this reach. In 2008, the discrepancy between the XS and DEMmethod is again due to the underestimate of bank scour, in addition to the overestimate of aggradation. The recruitment of a piece of LW that spans XS D, E and F has induced aggradation within these cross-sections, resulting in a large bias of the total aggradation within these areas. Figure 6.4 also suggests the XS method overestimates the degradation occurring in XS 18 while underestimating the degradation in XS 17 in 2008. Vertical and horizontal scour evident in this area during the 2007 event resulted in massive bank failures and tree recruitment (refer to Chapter 4). As a result, the banks in this area are very unstable. It is likely that slumping into the channel from these unstable banks, in addition to small localized bar building near XS 18 was under represented by the XS method. On the other hand, localized 105 6.4. Volumetric Net Change Figure 6.5: Elevation and bank changes in the middle reach (sub-reach C) showing the change in elevation from 2006 to 2007 (in 0.5 m contours) and from 2007 to 2008 (in 0.2 m contours). The locations of jam 2 and jam B are identified and labelled with arrows, in which the size of the arrow indicates the relative size of the jam, along with the cross-section locations within each 10 m DEM segment. Bank changes are shown in black and were assigned a single value for the vertical change within the bank change for the area within the DEM segment. 106 6.5. Bed Material Transport Rates aggradation from the large jam in segment 18 has resulted in some accumulation of sediment near XS 17, again producing a biased estimate of aggradation for this area. 6.5 Bed Material Transport Rates The variability of bed material transport within the study reach is estimated using the estimated topographic variation from both the cross-sectional and DEM analysis in combination with an estimate of particle path length. Typical particle path lengths can be inferred from river morphology in which the average travel distance for a significant transporting event is equal to bar-to-bar spacing (Church, 1992). In order to evaluate the characteristic path length for bed material in Fishtrap Creek, magnetically tagged tracers were deployed in Fishtrap Creek before the 2006 and 2007 freshets. The mean step lengths for tracer launch line A - D during 2006 was 80 m, 73 m, 144 m, and 115 m respectively (refer to Fig. 3.1 in Chapter 4 for locations of launch lines). In 2007, the step lengths increased slightly, with distances of 138 m, 159 m, 143 m, and 130 m in launch line A through D respectively. Tracer stones were not deployed in 2008, so an equation based on excess unit stream-power (equation 6.1) was used to estimate path for predicting the step length for 2008 using the daily average stream flow data, as described by Eaton et al. (2008): Lp = n￿ i=1 0.045 ￿ γ(Qi −QcritS Wbf ￿1.62 (6.1) where Lp is the total path length for a given event, Qi is the ith of n daily discharges that exceed the critical discharge for particle movement (Qcrit), γ is the unit weight of water, S is the average energy gradient for the reach, and Wbf is the average bankfull width. Calibration of Eq. 6.1 is presented in detail in Eaton et al. (in press). Using this method, the average step length of 115 m was estimated for 2008. In this case, we apply the morphological method by assuming that a typical path length (Lp) can be applied to the total volume of erosion (Ve) or deposition (Vd) mea- sured within each cross-section and 10 m DEM segment. Estimates of the bed material transport rate variability at each 10 m DEM segment and cross-sectional transect can be made using yearly recorded topographic changes with the following equation: Qebm = Ae · Ls (6.2) Qdbm = Ad · Ls 107 6.5. Bed Material Transport Rates Figure 6.6: Estimates of bed material transport rate using cross-section and DEM analysis at the cross-section and 10 m scale along the study reach for both for the 2007 and 2008 freshet. Sub-reach extents are shown with arrows and labelled as A-D between the two plots along with the XS locations listed above. Areas shaded light grey represent transport rates overestimated by the XS method, while darker grey areas represent transport rates underestimated by the XS method. Grey and black dotted lines show actual transport rates estimated by the XS and DEM method for 2007 and 2008. 108 6.5. Bed Material Transport Rates where Qebm and Q d bm are the estimated bed material transport rate of erosion and de- position in m3/event, based on the cross-sectional area of erosion or deposition for a given cross-section/10 m DEM segment (Ae or Ad in m2), and Lp is the path length for the sub-reach in which each cross-section or 10 m segment is found. Given the negative bias associated with the morphologic method, the “real” transport rate (Qbm) is taken to be the larger of the values from Qebm and Q d bm. Local sediment inputs from bank erosion have resulted in highly variable bedload transport rates with progression downstream (Fig. 6.6). The bed material transport rate estimated by the DEM and XS methods are on the same order of magnitude for 2007 and 2008 (3000 m3/yr compared to 2000 m3/yr and 600 m3/yr compared to 450 m3/yr respectively). In 2008, the DEM method estimates a drop by nearly half downstream of jam 2, while the XS method estimates an almost equal transport rates upstream and downstream of this jam. This suggests that the DEM method still estimates a relatively large deviation in sediment transport upstream and downstream of jam 2, in which case this jam is still acting as a morphologic barrier for sediment trapping. However, higher transport rates immediately downstream of jam 2 from the erosion captured in XS 7 and 8 from the DEM method, resulted in higher transport rates downstream of jam 2 (refer to Fig. 6.5). It is possible to define the actual transport rate (shown as dotted lines in Fig. 6.6) for Fishtrap Creek using the information provided in Fig. 6.3. Areas associated with the highest net change, either deposition or erosion, are zones in which actual sediment transport rates can be estimated. When the cross-sectional area is multiplied by a path length (see Eq. 6.2), the event transport rate can be determined and dominant maximum and minimums shown in Fig. 6.3 can be related to the actual transport rates observed in Fig. 6.6. Dominant cycles of bed sediment transport rates are evident in 2007 and 2008 (shown as E, D and T in Fig. 6.6). In 2007, XS 17 and 18 in sub-reach D are the primary erosional sinks (E1) within the reach. This sediment is transported through a relatively stable transport zone (T1), in which there was little change in bed topography, and thus very low sediment transport rates. According to the XS method, the first zone of deposition is found in the lower end of sub-reach D (D1), which is followed by a second transport zone (T2), and finally another deposition zone (D2). However, this cycle is less evident in the DEM method, in which the we seem to see higher transport rates in T2 and lower transport rates in D2 compared to the XS method. The size of the second transport zone seems to have reduced according to the DEM method, in which D1, T2 and D2 can be considered one large depositional sink D1(DEM) (see Fig. 6.6). Immediately downstream of the jam in 2007, the XS method shows very stable reaches, resulting in the final transport zone (T3). The higher bed 109 6.6. Surface Response and Morphologic Adjustment material rates estimated by the DEM method immediately downstream of jam 2 suggest this area is in fact an erosional zone (E2(DEM)), which is followed by a transport zone further downstream in sub-reach B (T3(DEM)) and another erosional zone in sub-reach A (E3(DEM)). In 2008 the transport cycles estimated by the XS method upstream of jam 2 are similar to 2008. The transport cycles in sub-reach D according to the DEM method, labelled E1(DEM), T1(DEM) and D1(DEM) in Fig. 6.6, correspond well to the cycles identified in 2007. Again, the transport zone (T2) identified by the XS method in 2007 and 2008 is not evident in the DEM method in 2008. Rather, this entire area seems to be undergoing relatively consistent deposition. The cycles identified by the 2008 DEM not only relate better to the cycles seen in the 2007 DEM method compared to that of the XS method, but they also show a more definable pattern over the whole study reach. Similar to 2007, the XS method found the majority of sub-reach C and part of sub-reach B (immediately below the jam) to be stable transport zones (T3). This is followed by a large depositional sink in sub-reach A (D3). However, the cycles identified by the DEM method are very different. An erosional sink (E2(DEM)) - similar in location to 2007, but much larger - is located immediately downstream of jam 2. A transport zone (T3) found in sub-reach B precedes a final erosional zone (E3(DEM)) found in sub-reach A. 6.6 Surface Response and Morphologic Adjustment As long as bedload transport is variable in the downstream direction, local changes in channel sediment storage must occur (Jackson and Beschta, 1982). The widespread occurrence of bank failures and sediment input from within the channel network has resulted in highly variable sediment storage patterns within the active bed. While many of the localized, small-scale variability in bed surface facies can be explained by in-stream LW (see chapter 5), there are distinct patterns of sediment deposition and sorting patterns which cannot. In the case of Fishtrap Creek, there are two primary sources which have resulted in distinguishable patters of sediment deposition. The first source, from the breakdown of jam C (refer to 5.9 in Chapter 5), resulted in deposition of fine material around jam 2. The second, much larger, source from XS 17 and 18 resulted in three phases of deposition approximately 50 to 150 m downstream. Longitudinal Patterns of Sediment Deposition The mean grain size, sorting and relative skewness of alluvial deposits are dependent on the sediment grain size distribution of its source and the sedimentary processes of erosion, transport and deposition (McLaren, 1981). The statistics of the source material 110 6.6. Surface Response and Morphologic Adjustment are unknown; however, since it has been shown that the material transported within Fishtrap Creek during the 2007 and 2008 freshet originated from within the channel, we can assume the material scoured from sub-reach D reflects the grain size distribution of the sub-surface material near XS 18 (see Fig. 6.7). In this case, the sorting coefficient for Figure 6.7: Upper reach sub-surface grain size distribution showing D16, D50 and D84. Sample collected between XS 17 and 18 the sub-surface material is 4.5. If source sediment undergoes erosion, and the resultant sediment in transport is deposited completely, the deposit must be finer, better sorted and more negatively skewed than its source (McLaren, 1981). With the exception of localized fine deposits around LW and from the backwater trap on the right-bank between XS 13 and 14, the majority of the bed deposits are better sorted than the source material (refer to Fig. 5.7). While the quantity of material entrained and transported during the 2008 freshet was far less than in 2007 (1250 m3 in 2007 compared to 500 m3 in 2008), much of the eroded material originated from bank erosion. As such, it is anticipated the grain size distribution in 2008 is similar to the source material in 2007. There were two primary sources of material in 2007. Prior to the erosion that occurred near XS 18, and subsequent deposition on the fan (XS 13 and 14) and in jam 2 (XS 9), it is believed that the break-down of jam C resulted in the first pulse of material downstream. The material entrained from this source was finer than the typical bedload within the stream, since it originated behind the jam. This material was transported down the study reach to jam 2 and deposited. Most of the fine deposits on the left bank and within the jam are remnants of this material (refer to Fig. 5.7). The fine material in 111 6.6. Surface Response and Morphologic Adjustment the pools and along the thalweg along the right bank are likely from the erosional zone near XS 17 and 18, since they are replenished and re-worked frequently. The coarse, well sorted material (D50 ranging between 64 and 90 mm) found near the left bank and slightly upstream of jam 2 is likely material that existed in this jam prior to the massive amounts of deposition in 2007 and 2008. The second, larger source from XS 17 and 18, resulted in three distinct sediment distribution zones in 2007 and to a lesser degree in 2008. Deposition of material finer than the typical bedload material, with a range of D50 between 11 to 23 mm, is evident in the narrow reach downstream of XS 17 and in the uppermost segment of the fan (refer to Fig. 5.7). This represents the first zone of deposition, in which sediment is not only finer than the bedload, but also slightly better sorted. These deposits are characterized by sand and fine gravels which overlay the coarser bed material, similar to that of the surrounding material (with a D50 range between 45 and 64 mm). Lisle and Hilton (1999) found that fines often deposit separately, so often these fine grains become separate from the residual bed material. Deposition of this fine material from the bedload material occurs between 50 and 75 m downstream, or half of the typical path length for 2007. The second zone of deposition is found in the middle of the fan between XS 12 and 13 (refer to Fig. 5.7). This is a very large deposit of coarse material, with D50 ranging from 64 to 90 mm, and with a sorting coefficient of 1.88. This material is much coarser than the typical bedload material, and has been deposited at a distance approximately equal to the path length for 2007 and 2008. Given the new geometry of the fan reach, in which the w:d ratio is high relative to the other reaches in the study site, the reduction in depth and thus transport capacity over this fan area has resulted in armored surface in which fines are being winnowed away - especially in 2008 (see Fig. 4.16B in Chapter 4). The third zone is the final deposition ground for the remaining bedload material transferred from the upper reach. The distribution of this deposit is typical of the bedload material with far better sorting (approximately 2.0). Material deposition and facies complexity on the fan and on jam 2 are complicated by a number of issues: i) the two primary source zones have different grain size distri- butions, ii) the quantity of LW interacting with the channel on the fan and in jam 2, and iii) from the reduction in slope from continuous aggradation which has resulted in the formation of secondary channels reduced the transport capacity. Only fine material carried in suspension over jam 2 is delivered downstream. Along with finding distinct patterns of sediment deposition from these large source areas, there is also evidence of increased concentrations of fine material in pools. This is likely the result of two issues: i) increased sediment supply in 2007 and 2008 and ii) low flows incapable of destroying armor layers, which results in the transport fines 112 6.7. Discussion into and out of pools. Lisle et al. (1993) found that high concentrations of fine material on stream bed result from increased supply, which can result in more fines available to pools. The increase in sediment supply to the river in 2007 and 2008 from bank failures has resulted in significant deposition of fine material in 10 of the 12 pools identified within Fishtrap Creek (refer to Fig. 5.7). When the supply of fine bed material exceeds the storage capacity of framework voids, excess fine material forms surficial patches, which can be voluminous in pools during low flow (Lisle and Hilton, 1999). On average, pools have a D50 1/5th the size of the D50 in the riffles with an average D50 of 12 mm in pools compared to an average D50 of 58 mm on riffles. Lisle and Hilton (1999) found fine material in pools is replaced frequently throughout the year. Given the abundance of fine material available for transport, both within the pools and on the bed, it is expected this is also the case in Fishtrap Creek. In 2007 and 2008, daily average peak flows reached 6.61 m3/s and 6.83 m3/s respectively, slightly below bankfull discharge. According to Jackson and Beschta (1982), bedload transport occurs in two relatively distinct phases: i) phase I, in which the source material is the fine sediment found in pools in which transport occurs over stable gravel-surfaced riffles on the rising limb of the hydrograph and ii) phase II, which occurs in flows greater than or equal to bankfull, resulting in the entrainment of riffle armor and deposition at the downstream end of riffles. Given the abundance of fine material on the bed (refer to Fig. 5.7) and since bankfull discharge has not been reached or exceeded since 2006, it is likely that the riffle armor in Fishtrap Creek has not been significantly disrupted. Rather, winnowing of fines from these zones (resulting in greater armoring ratios and larger grain sizes), and the abundance of fine materials within the pools, can also be attributed to the occurrence of phase I bedload transport. 6.7 Discussion There are inherent problems and challenges that can arise from intensive small space and time scale case studies when evaluating channel morphologic change and sediment trans- port, especially when expanding and or explaining these processes over larger scales. Lane and Richards (1997) suggest different scales of form and process within river chan- nels are not causally independent of each other. They also suggest that a process has a characteristic time scale which is much shorter than the time scale of interest; it may be ‘relaxed’, but not ignored. Intensive case studies, such as this one, are becoming more prevalent in the literature as a method of understanding channel configuration. The internal distributions of erosion and deposition, as determined by channel morphology, account for the dependance of sediment transport rates on channel configuration (Lane 113 6.7. Discussion and Richards, 1997). When understanding the morphodynamics of an unstable system following a disturbances (such as Fishtrap Creek), event-scale methodology is adequate for understanding form and process. Given the high data resolution of the DEMmethod, the first part of the discussion will focus on the results documented by this method. The second part of the discussion will look at how the XS and DEM method differ, and how the contrasting results may influence morphodynamic evaluation at multiple scales. 6.7.1 Topographic Change, Transport Rate and Facies Distribution The morphologic response of Fishtrap Creek following the fire is very significant in 2007, and to a lesser degree, in 2008. There are identifiable zones of degradation and aggradation in 2007 which tend to be significantly greater upstream of jam 2. Smaller magnitude zones in the same areas were also identified in 2008. Total reach deposition and erosional volumes are almost balanced within the study reach in 2007 and 2008 (refer to Table 6.2). However, in 2008 there are far greater volumes of erosion upstream of jam 2 and greater volumes of erosion downstream of jam 2. This observation suggests that material is entering the stream from outside of the study reach and being deposited in the fan and in jam 2. The increase in erosion downstream of jam 2, and very little deposition within this reach (only at LW forced deposits at XS E, F and G), suggests material being transported beyond the study reach. The DEM method improves our ability to confidently conclude endogenous processes are occurring, since longitudinal differences in bank erosion and subsequent deposition can be quantified far better than with the XS method. The transport rates in 2007 and 2008 show a cyclical pattern of erosion, transport and deposition zones along the longitudinal profile of the river. Peaks occur at approx- imately the same location during the 2007 and 2008 freshet, with smaller magnitudes in 2008. The reach average transport rate in Fishtrap Creek is complicated by the mor- phologic barrier at jam 2. This results in two different transport regimes upstream and downstream of this barrier, especially in 2007. The estimated error in the transport zones, in which net change is approximately zero, can be estimated based on the differ- ence between the maximum peak transport (D1(DEM) in Fig. 6.6) and T1, the lowest transport rate. In this case, the estimated error is a factor of 10 in 2007 and in 2008. The bed surface facies exhibit obvious depositional characteristics which are related to their source material. While the characteristic facies cannot be explained entirely by sediment transport patterns, there are distinct patterns which can be attributed to these patterns. The first source of material, originating from the break up of jam C in 2007, resulted in the deposition of fine sediment on jam 2. The largest source of material, orig- inating from the degradation zone at XS 17 and 18, resulted in fine material deposition 114 6.7. Discussion between 50 and 75 m downstream, coarser material slightly further downstream and med-coarse sediment deposition furthest downstream, at approximately 175 m down- stream on jam 2. The substantial increase in sediment supply, originating primarily from these two sources within the stream, have resulted in the widespread deposit of fine materials - especially in pools. In the absence of flows greater than bankfull, the coarse riffle layers have been left intact, while fines have been winnowed and transported and deposited downstream in these pools. 6.7.2 Comparing DEM and XS Method The highest net changes documented were found in areas between cross-section segments (between XS 13 and XS 14) where cross-section locations were spaced 20 m apart. The XS method often captured localized aggradation from pieces of LW (XS E, F and G) or on jams (XS 9 and XS 17) resulting in overestimates of aggradation within these sections. On the other hand, bank failures and bed scour around log jams were typically underestimated, especially around jam 2 and B (refer to Fig. 6.4). As a result, the magnitude of net changes within the cross-sections were very different, and often opposite, to what was estimated by the DEM method. Downstream of jam 2, the DEM method captures important topographic changes that are often missed or over represented by the XS method. This is evident in sub-reaches A, B and C during 2007 and 2008. The major discrepancy in this lower reach is the inability for the XS method to capture localized bank erosion from LW inputs, which resulted in large underestimates of erosion volumes. Similar results were also documented by Fuller et al. (2003), who found that cross-sectional evaluation of sediment volumes derived underestimate the magnitude of volumetric changes that occur within the reach. When comparing transport rates, important topographic changes were often under or overestimated by the XS method, which resulted in variable longitudinal sediment transport cycles compared to the DEM method (see Fig. 6.6). There is inherently far greater error associated with the XS method, due to inaccurate estimation of between cross-section topography and changes in bank locations. As a result, measuring the important fluctuations in net change and transport rates are more accurately captured with more inclusive data techniques, such as provided by the DEM method. The key control upon the quality of cross-sectional data is the uncertainty in mean bed level change that arises from the density and location of cross-sections (Lane et al., 2003). It becomes obvious when identifying the morphodynamics of an unstable river similar to Fishtrap Creek, the DEM method is a far better method for estimating chan- nel changes. The greatest error in the estimate of bedload transport rates existed in areas where cross-sectional spacing was too large or in areas where localized aggrada- 115 6.7. Discussion tion and or degradation produced results not typical of the entire segment. The largest improvement that can be made to reduce this error with the XS method in Fishtrap Creek is to reduce the cross-sectional spacing, especially in the upper reach where XS spacing is ≥20 m. Lane et al. (2003) found that error generally increases as the cross- section spacing is increased. Their results were similar to Fishtrap Creek, in which they found typical cross-section spacing used for long-term river monitoring, a distance ap- proximately equal to one bankfull width, results in significant differences detected in the pattern of bed level change to that estimated from DEM’s. Lane et al. (2003) also found that systematic error and sensitivity to choice of cross-section location both tended to zero at cross-section spacings less than 1/8th of the channel bankfull width. Using this logic for Fishtrap Creek, the estimated spacing between cross-sections required to gather accurate changes in bed changes would be approximately 1.5 m. To make a representa- tive DEM, in which points are gathered every 1.0 to 1.5 m apart, requires 4 to 5 days of data collection (based on a 450 m study reach). This is approximately double the time it takes to survey 27 cross-sections; however, this does not include the initial set-up and survey that are required to install cross-sectional transects. The significant increase in data resolution resulted in far better representation of the channel morphodynamics - warranting the extra time of field data collection required to generate a surface model. 116 Chapter 7 Conclusions 7.1 Short Term Response to Wildfire at Fishtrap Creek Three years of reach scale morphologic adjustment, LW recruitment rates and loads, sediment transport rates, and longitudinal patterns of grain size distributions in an intermediate stream disturbed by wildfire have been evaluated. Observations suggest that the first evidence of channel change occurred in 2006, when the influx of sediment from small bank failures resulted in the creation of pronounced bars and riffles. As a result, the channel shifted from a characteristically plane-bed morphology in 2005, to a more characteristically riffle-pool morphology in 2006. The flows in 2005 were the highest since the fire; however, channel changes were generally modest. Suspended sediment monitoring since the fire suggest concentrations were not significantly elevated compared to background levels, nor were these changes accompanied by a substantial increases in bed material transport rates (Eaton et al., in press). Daily average flows in 2007 reached a modest 6.61 m3/s, the second lowest flow on record since the fire. However, the 2007 freshet marked the year of greatest morpho- dynamic adjustment. LW jam breakup initiated massive bank failures and vertical bed scour in the upper reaches, resulting in significant sediment supply downstream. De- position of this material resulted in wide-scale aggradation which resulted in prevalent changes in channel change, including exceedingly high width to depth rations (most prevalent in the fan reach), reductions in bed slope, and the formation of two large avulsions. The majority of the sediment originated from bank failures near XS 17 -19, as well as from minor bank failures throughout the study reach. Widespread bank fail- ures since the fire are likely due to decay of roots resulting in reduced bank strength. The quantity of LW recruited via bank erosion during 2007 suggests the bank strength reached a minimum four years following the fire. These bank failures resulted in reach- average transport rates almost seven times greater than the background transport rates. The volumes of sediment eroded and deposited within the study site balanced in 2007, suggesting channel changes were the result of endogenous sediment supply, in which hillslope processes did not contribute. These results are not typical of most studies that documented increases in sediment and flow from hillslope processes following wildfire. 117 7.1. Short Term Response to Wildfire at Fishtrap Creek The deposition zones in sub-reaches D and C did see a reduction in channel bedforms and organization compared to 2006. In addition, cross-sections that saw very little net volumetric change (XS 11 and 12) or significant degradation (XS 17) also saw significant increases in bar amplitudes, while most of the remaining cross-sections saw reductions. In this case, the inundation of sediment (in particular, fine gravels and sands) resulted in the smoothing of the bedforms created in 2006. Bank failures in 2007 were responsible for nearly half of the total wood volume recruited following the fire, in which 2007 saw a five fold increase in the volume of LW recruited compared to 2005 and 2006. The occurrence of wind throw in 2007 resulted in the addition of three pieces of wood to the channel, which accounted for only 0.2 m3 of the total volume of wood recruited this year. An evaluation of the contribution of new wood and old wood on bed facies distribution shows that most of the new wood does not significantly influence flow hydraulics, and thus is not yet functioning on the channel bed. In Fishtrap Creek, pre-fire wood is the most hydraulically significant within the channel, and accounts for nearly all deep pools within the channel and is responsible for highly complex facies distribution. In addition, the facies complexity is directly related to the total number of LW pieces within each LW segment. In this case, the most variable bed surfaces are found in areas where the number of old wood pieces is highest. In 2008, five years following the fire, the morphologic adjustment relaxed signif- icantly. While flows were almost identical to 2007 (approximately 6.8 m3/sec), the occurrence of bank failures and LW recruitment was significantly reduced. Localized bank erosion supplied less than half the volume of sediment to the channel compared to 2007. In addition, the reduction in bank failures resulted in a four-fold decrease in the total volume of wood recruited in 2007. Sediment transport rates were still ele- vated compared to background levels; however, they were an order of magnitude less than 2007. Lower flows, incapable of destroying armoured riffle surfaces, has resulted in continuous winnowing of fines from riffles in 2008. In 2009, volumes of recruitment were similar to those observed in 2005 and 2006. However, the mode of recruitment changed from primarily bank erosion to wind throw. The reduction in LW recruitment in 2008 and change in mode of recruitment in 2009 is likely a result of two factors: i) the regrowth of the understory bushes and shrubs causing an increase in bank strength, and ii) the riparian stand dynamics, in which the number of snags near the river bank available for recruitment is limited. 118 7.2. Influence of LW and Sediment Supply on Aquatic Habitat 7.2 Influence of LW and Sediment Supply on Aquatic Habitat How Much LW and When? The occurrence of pools and bar building and the creation of islands, and thus habitat formation, depends on the size and timing of LW recruited to the channel and the sediment transport through the reach. While the size of the wood entering Fishtrap Creek is average, and according to Bragg et al. (2000), the recruitment rate following the fire was not exceedingly high compared to model results, the wood load volumes are generally higher than other studies documenting disturbance. This is likely due to the degree of channel change that occurred in 2007, which resulted in highly complex morphology. As a result, the total wood load is likely higher because the stream inhibits movement of wood out of the system - so attrition rates are typically lower than studies involving fire. Given the significant reduction of LW recruitment to the channel in 2008, and the shift in recruitment modes to wind throw in 2009, it is expected that continued LW recruitment to the stream will be similar to or less than the volumes of wood recruited in 2008 and 2009. If wind does continue to become the dominant mode of recruitment to the channel (similar to what was observed in 2009), it is anticipated that an additional 1 to 2 m3/m of LW may enter the channel each year. If, however, this system does respond similar to the results of Bragg et al. (2000), we might expect another increase in LW recruitment 20 to 25 years from now. Movement of wood will likely occur only when flows are large enough to submerge the wood, in which case flows greater than bankfull will be required to dislodge and move LW within the current channel boundaries. LW and the Creation of Aquatic Habitat Intermediate gravel-bed channels combine the features of braided and irregularly sin- uous, single-thread channels, and are often characterized by backflow channels which provide important habitat for numerous fish populations (Church, 2002). Fishtrap Creek has high volumes of functional wood that is oriented perpendicular to the channel - re- sulting in increased hydraulic blockage and bed interaction. LW of this orientation not only provides important habitat for aquatic organisms, but has the greatest influence on channel morphology in intermediate sized streams (Chen et al., 2008). Almost 80% of the pools are a result of pre-fire LW and all of these pools are deeper than pools created by fluvial mechanisms. These deep pools have been found to provide an integral physi- cal component for the life stages of many aquatic organisms, such as providing refugia 119 7.2. Influence of LW and Sediment Supply on Aquatic Habitat and rearing habitat to anadromous fish (Abbe and Montgomery, 1996). LW and indi- vidual LW jams can be remarkably stable, in which distinct alluvial topography that is a result of these features can persist for at least as long as the structure remains stable; in addition, they can providing long-term bank protection that creates local refugia for ecosystems that are characterized by rapid channel migration and frequent disturbance (Abbe and Montgomery, 1996). The current jams in Fishtrap Creek are in two distinct functional phases - which is primarily related to their age and structure. Old jams, such as jam 1 and jam 2, are fully functioning in the channel - in which we see local- ized pools, bank widening and highly variable grain size distributions. These jams were created before the fire, and while some new material has been recruited to these jams, most of the material is old wood. Both of these jams are stable, and thus are expected to function as critical habitat as long as they remain intact. Newly formed jams, not yet functioning fully in the channel, are expected to become more hydraulically significant (thus creating more aquatic habitat zones) when more LW is added and/or decay of these key pieces begins. Gravel Facies and Fine Sediment Supply Grain sizes in Fishtrap Creek are directly related to the number of LW is found in a reach. In addition, areas of similar deposits can be related to their source material, while avulsions and secondary channels can increase the complexity of the longitudinal facies distributions. The grain size distribution of stream bed gravels can limit the success of spawning by salmonids. Material that is too coarse impeded the ability for the spawning fish to move, while excessive supply of fine sediment (≤ 10 mm in size) can clog spawning gravels (Kondolf, 2000). Following the channel reconfiguration in 2007, bars became significantly finer, which could have adverse effects on fish habitat and spawning grounds. However, in the absence of significant increases in sediment or increased flows in 2008, winnowing of fines during low flows caused most of these bars and riffles to become coarser. Most of the LW-forced pools in Fishtrap Creek have median grain sizes less than 10 mm (refer to Fig. 5.7). The abundance of fine material in these pools is a result of increased sediment supply, in addition to continuous redistribution of fine sediment from riffles into pools during sub-bankfull flows. It is likely that the surface facies distribution in 2007 would have been significantly finer in many parts of the channel compared to 2008 - most especially in the fan and near jam 2. The 2008 grain size distributions on bars and riffles are beginning to resembled grain size distributions in 2006. The timing of sediment input is very important when evaluating the influence on 120 7.3. Management Implications and Recommendations aquatic life, most especially salmonid spawning. The morphologic adjustments in Fish- trap Creek occurred over a short period of time during peak flows. Unlike channel response to anthropogenic disturbance, which results in increased sediment supply that may occur during low flows in the channel, material was supplied to Fishtrap Creek during high flows. In addition, contrary to many anthropogenic disturbances (most es- pecially urban construction, including communities and roads), the large scale response in Fishtrap Creek only occurred during one freshet. While changes were noted in 2006, and changes are still ongoing in 2008, they are minor in comparison to the channel reconfiguration that occurred in 2007. In fact, the slight increase in sediment during 2006, which was distributed throughout the channel, resulted in more complex channel topography - which would likely provide greater and more abundant habitat areas for aquatic organisms. In 2007, however, massive degradation of the bed and banks (up to nearly 2 m in depth), would have resulted in the complete evacuation of eggs during the spring freshet. Subsequently, the deposition of this material downstream would have resulted in complete burial of these eggs - which again would be extremely detrimental to the survival of these fish populations. 7.3 Management Implications and Recommendations Understanding channel morphologic response following wildfire, including the distribu- tion and complexity of the bed facies following significant channel modification, can provide critical insight into the success and/or failure of fish populations in similar sys- tems disturbed by fire. However, all streams respond differently to disturbance, and not all disturbances are the same. The appropriate data collection and analysis methods depend on the questions being asked — whether it is for construction purposes, post dis- turbance (including human and natural disturbance) evaluation of channel morphology and/or management of previously disturbed streams, habitat survival and functioning, and/or channel restoration. While Fishtrap Creek is not one of the primary fish bearing streams in British Columbia, there are numerous intermediate gravel-bed rivers in the interior and coast similar to Fishtrap Creek that are. Facies mapping proves to be an important tool for the evaluation of short term response and change of grain-size distri- butions over time. While a facies map of this size and detail may not be warranted for every study, the map provided by this research provides critical information on the in- fluence of increased sediment supply and hydraulic roughness elements (primarily LW) in this system. Evaluation of grain sizes and sorting is imperative for understanding fish spawning and the survival of other biologic life, as many species prefer very particular sediment substrates. In addition, facies that differ in size and sorting can provide insight 121 7.3. Management Implications and Recommendations into the mechanisms and sources of sediment, as well as bed load transport rates and the potential influence of these surface textures (and potential roughness elements) on sediment mobility. While an evaluation of the influence on fish populations was not performed, it is believed that a disturbance and response similar to that seen at Fishtrap Creek would likely have adverse side-effects on aquatic life in the short term (3 to 5 years following fire), but more positive influence in the longer term (≥ 6 years following wildfire). The delayed response reported at Fishtrap Creek suggests that the influence of the fire on aquatic life may not occur immediately following the fire. The increased recruitment of LW provides ample habitat space and pools formation for aquatic life, including numer- ous bars and riffles for fish populations. The time it will take for this stream to become stable again is unknown; however, sediment transport rates and LW recruitment in 2008 have relaxed significantly compared to 2007 and it is expected that the current state of the channel is adequate for aquatic life and fish spawning. Madej (2001) hypothesized that the spatial scales of channel structure and organization, as well as the time required to develop channel organization, differ in different parts of the channel network, based on channel gradient and occurrence of organizing flows. There are a number of variables that can interact and influence the current trajectory of channel adjustment and form within Fishtrap Creek. However, it is anticipated that a system similar to Fishtrap Creek may see improvement of the habitat conditions and spawning grounds within 1 to 2 years after major channel adjustment. In addition, the increased supply of LW to the channel may facilitate habitat creation — resulting in a richer, more ecologically diverse system in the future. Evaluation of two different approaches to the morphologic method, cross-sectional analysis versus surface DEM data, shows the major shortcomings of using the cross- section method. Lane et al. (1994) suggest that much more detailed information on river bed topography is needed to accommodate the increasing recognition of spatially distributed form-process feedback in fluvial environments and to provide a more rigorous linkage between channel topography and sediment transport processes. The DEM-based approach to monitoring the three-dimensional nature of river channel form has a major advantage over the cross-section approach in that data collection is topographically de- fined, resulting in better surface representation (Lane, 1998). Cross-section data miss important details between survey sites, and introduces significant bias in the total net changes seen within segments more than one bankfull width in length. Even in ar- eas where the cross-section method evaluated net change within one bankfull width, the estimates were still significantly different than the DEM method. Discrepancies in the results between the two methods are a result of: i) highly variable bank widening 122 7.3. Management Implications and Recommendations throughout the study reach that is not accurately captured by the cross-section method and ii) localized areas of sedimentation forced by LW produces biased estimates of aggra- dation and degradation throughout the segment when using the cross-section method. In light of these results, the DEM approach to morphologic change and transport rates for Fishtrap Creek is a better method for this system. More precise estimate of mor- phodynamic behaviour can be obtained by setting cross-sectional spacing to no more than approximately 1/8· Wbf . The field time, and thus expense, required to gather topographic data for a detailed surface DEM is greater than the time required to sur- vey cross-section spaced at approximately one bankfull width apart. However, in order to obtain results similar to the DEM method from cross-sectional data, 6 to 8 times more cross-sectional transects must be added. In this case, depending on the size of the stream, the time and cost required design and set up the number of cross-sections required may warrant the collection of DEM points. Fishtrap Creek provided a unique opportunity to evaluate channel response following wildfire. The absence of increased hillslope erosion and/or runoff allowed us to evaluate stream processes and response with the reduction of two external, very typically influ- ential variables. However, caution must be exercised when evaluating channel response and applying it over larger time and space scales. The unique case study provided here exemplifies the importance of appropriate field techniques and data resolution. When attempting to monitor post-fire channel changes, short study reaches could lead a re- searcher to incorrectly attribute internally generated sediment pulses (such as described herein) to inferred increases in sediment supply from the watershed upstream. Study reaches should be several times longer than the characteristic bed material transport length, Ls (in our stream Ls ≈ 7 to 14 times Wbf ). In addition, for responses similar to Fishtrap Creek, it is essential that post-fire monitoring occur for at least five years following fire. 123 Bibliography Abbe, T., Montgomery, D., 1996. Large woody debris jams, channel hydraulics and habitat formation in large rivers. Regulated Rivers Research & Management 12 (23), 201–221. Abbe, T. B., Montgomery, D. R., 2003. Patterns and processes of wood debris accumu- lation in the Queets River basin, Washington. Geomorphology 51 (1-3), 81 – 107. Abernethy, B., Rutherfurd, I., 2000. The effect of riparian tree roots on the mass- stability of riverbanks. Earth Surface Processes and Landforms 25 (9). Adams, J., Beschta, R., 1980. Gravel bed composition in Oregon coastal streams. Cana- dian Journal of Fisheries and Aquatic Sciences 37 (10), 1514–1521. Anderson, R., Bledsoe, B., Hession, W., 2004. Width of streams and rivers in response to vegetation, bank material, and other factors. Journal of the American Water Re- sources Association 40 (5), 1159–1172. Ashmore, P., Church, M., 1998. Sediment transport and river morphology: a paradigm for study. Gravel-bed Rivers in the Environment, 115–148. Ashmore, P., Church, M., 2001. The impact of climate change on rivers and river pro- cesses in Canada. Geological Survey of Canada. Baker, V. R., 1977. Stream-channel response to floods, with examples from central Texas. Geological Society of America Bulletin 88, 1057–1071. Barro, S. C., Wohlgemuth, P. M., Campbell, A. G., 1988. Post-fire interactions between riparian vegetation and channel morphology and the implications for stream channel rehabilitation choices. In: California Riparian Systems Conference. Vol. Gen. Tech, Rep. PSW-110. USDA Forest Service, Davis, CA, 51–53. Barry, J., Buffington, J., King, J., 2004. A general power equation for predicting bed load transport rates in gravel bed rivers. Water Resour. Res 40 (10). Bathurst, J., Graf, W., Cao, H., 1987. Bed load discharge equation for steep mountain rivers. Sediment Transfer in Gravel-Bed Rivers. John Wiley & Sons New York. 1987. 124 Bibliography Beche, L., Stephens, S., Resh, V., 2005. Effects of prescribed fire on a Sierra Nevada (California, USA) stream and its riparian zone. Forest Ecology and Management 218, 37–59. Benda, L., Dunne, T., 1997. Stochastic forcing of sediment supply to channel networks from landsliding and debris flow. Water Resources Research 33 (12), 2849–2863. Benda, L., Miller, D., Bigelow, P., Andras, K., 2003. Effects of post-wildfire erosion on channel environments, boise river, idaho. Forest Ecology and Management 178 (1-2), 105 – 119. Benavides-Solorio, J., MacDonald, L. H., 2001. Post-fire runoff and erosion from simu- lated rainfall on small plots, Colorado Front Range. Hydrological Processes 15, 2931– 2952. Bilby, R., Ward, J., 1989. Changes in characteristics and function of woody debris with increasing size of streams in western Washington. Transactions of the American Fisheries Society 118 (4), 368–378. Bilby, R., Ward, J., 1991. Characteristics and function of large woody debris in streams draining old-growth, clear-cut, and 2nd-growth forests in southwestern Washington. Canadian Journal of Fisheries and Aquatic Sciences 48 (12), 2499–2508. Bragg, D., 2000. Simulating catastrophic and individualistic large woody debris recruit- ment for a small riparian system. Ecology 81 (5), 1383–1394. Bragg, D., Kershner, J., Roberts, D., 2000. Modeling large woody debris recruitment for small streams of the central rocky mountains. General Technical Report RMRS-GTR- 55, United States Department of Agriculture, Fort Collins, COL U.S. Department of Agriculture, Forest Service, Rocky Mountain Research Station 36 1–34. Braudrick, C., Grant, G., Ishikawa, Y., Ikeda, H., 1997. Dynamics of wood transport in streams: A flume experiment. Earth Surface Processes and Landforms 22 (7), 669–683. Braudrick, C., Dietrich, W., Leverich, G., Sklar, L., 2009. Experimental evidence for the conditions necessary to sustain meandering in coarse-bedded rivers. Proceedings of the National Academy of Sciences 106 (40), 16936–16941. Brewer, P. A., Lewin, J., 1998. Planform cyclicity in unstable reach: Complex fluvial response to environmental change. Earth Surface Processes and Landforms 23 (11), 989–1008. 125 Bibliography Brooks, A. P., Brierley, G. J., Millar, R. G., 2003. The long-term control of vegeta- tion and woody debris on channel and flood-plain evolution: insights from a paired catchment study in southeastern Australia. Geomorphology 51 (1-3), 7–29. Brown, A. G., Cooper, L., Salisbury, C. R., Smith, D. N., 2001. Late Holocene chan- nel changes of the Middle Trent: channel response to a thousand-year flood record. Geomorphology 39, 69–82. Brummer, C. J., Abbe, T. B., Sampson, J. R., Montgomery, D. R., 2006. Influence of vertical channel change associated with wood accumulations on delineating channel migration zones, Washington, USA. Geomorphology 80 (3-4), 295–309. Buffington, J., Montgomery, D., 1999a. A procedure for classifying textural facies in gravel-bed rivers. Water Resources Research 35 (6), 1903–1914. Buffington, J., Montgomery, D., 1999b. Effects of hydraulic roughness on surface tex- tures of gravel-bed rivers. Water Resources Research 35 (11), 3507–3521. Buffington, J., Montgomery, D., 1999c. Effects of sediment supply on surface textures of gravel-bed rivers. Water Resources Research 35 (11), 3523–3530. Bunte, K., Abt, S., 2001. Sampling surface and subsurface particle-size distributions in wadable gravel-and cobble-bed streams for analyses in sediment transport, hydraulics, and streambed monitoring: RMRS-GTR-74. US Department of Agriculture, Forest Service, Rocky Mountain Research Station. Bunte, K., Potyondy, J., Ryan, S., 2004. Measurement of coarse gravel and cobble transport using portable bedload traps. Journal of Hydraulic Engineering 130, 879. Campbell, R., Smith, D., Arsenault, A., 2005. Dendroentomological and forest man- agement implications in the Interior Douglas-fir zone of British Columbia, Canada. Dendrochronologia 22 (3), 135–140. Cannon, S. H., Reneau, S. L., 2000. Conditions for generation of fire-related debris flows, Capulin Canyon, New Mexico. Earth Surface Processes and Landforms 25, 1103–1121. Carson, M. A., Griffiths, G. A., 1989. Gravel transport in the braided Waimakariri River: Mechanisms, measurements and predictions. Journal of Hydrology 109 (3-4), 201 – 220. Cerda, A., Doerr, S. H., 2008. The effect of ash and needle cover on surface runoff and erosion in the immediate post-fire period. Catena 74 (3, Sp. Iss. SI), 256–263. 126 Bibliography Cerda, A., Lasanta, T., 2005. Long-term erosional responses after fire in the central Spanish Pyrenees 1. water and sediment yield. Catena 60, 59–80. Chanasyk, D., Whitson, I., Mapfumo, E., Burke, J., Prepas, E., 2003. The impacts of forest harvest and wildfire on soils and hydrology in temperate forests: A baseline to develop hypotheses for the Boreal Plain. Journal of Environmental Engineering and Science 2 (Suppl. 1), S51–S62. Chen, X., Wei, X., Scherer, R., Hogan, D., 2008. Effects of large woody debris on surface structure and aquatic habitat in forested streams, southern interior British Columbia, Canada. River Research and Applications 24 (6), 862–875. Chew, L. C., Ashmore, P. E., 2001. Channel adjustment and a test of rational regime theory in a proglacial braided stream. Geomorphology 37 (1-2), 43–63. Church, M., 1992. Channel morphology and typology. The rivers handbook 1, 126–143. Church, M., 1995. Geomorphic response to river flow regulation: case studies and time- scales. Regulated Rivers Research and Management 11, 3–22. Church, M., 2002. Geomorphic thresholds in riverine landscapes. Freshwater Biology 47 (4), 541–557. Church, M., 2006. Bed material transport and the morphology of alluvial river channels. Annual Review of Earth and Planetary Sciences 34, pp. 325–354. Church, M., Hassan, M., 2002. Mobility of bed material in Harris Creek. Water Re- sources Research 38 (11). Church, M., McLean, D., Wolcott, J., 1987. River bed gravels: sampling and analysis. Sediment Transport in Gravel-Bed Rivers. John Wiley and Sons New York. 1987. p 43-88, 17 fig, 3 tab, 50 ref. Comiti, F., Andreoli, A., Lenzi, M., Mao, L., 2006. Spatial density and characteristics of woody debris in five mountain rivers of the dolomites (Italian Alps). Geomorphology 78 (1-2), 44 – 63, mountain Rivers Part I: Watershed Scale Processes and Channel Morphology. Coulthard, T. J., Kirkby, M. J., Macklin, M. G., 2000. Modelling geomorphic response to environmental change in an upland catchment. Hydrological Processes 14 (11-12), 2031–2045. 127 Bibliography Covert, A., Jordan, P., Curran, M., June 2009. Effects of burn severity on soil char- acteristics and erosion response. FORREX, Kelowna, BC, wildfire and watershed hydrology: workshop proceedings Edition. Curran, J., Wilcock, P., 2005. Effect of sand supply on transport rates in a gravel-bed channel. Journal of Hydraulic Engineering-ASCE 131 (11), 961–967. Dade, W., 2000. Grain size, sediment transport and alluvial channel pattern. Geomor- phology 35 (1-2), 119–126. D’Agostino, V., Lenzi, M., 1999. Bedload transport in the instrumented catchment of the Rio Cordon Part II: Analysis of the bedload rate. Catena 36 (3), 191–204. Desilets, S. L. E., Nijssen, B., Ekwurzel, B., Ferre, T. P. A., 2006. Post-wildfire changes in suspended sediment rating curves, Sabino Canyon, Arizona. Hydrological Processes 21 (11), 1413–1423. Desloges, J. R., Church, M., 1992. Geomorphic implications of glacier outburst flooding: Noeich River valley, British Columbia. Canadian Journal of Earth Science 29, 551– 564. Dietrich, W., Kirchner, J., Ikeda, H., Iseya, F., 1989. Sediment supply and the develop- ment of the coarse surface-layer in gravel bedded rivers. Nature 340 (6230), 215–217. Doerr, S., Shakesby, R., Blake, W., Chafer, C., Humphreys, G., Wallbrink, P., 2006. Ef- fects of differing wildfire severities on soil wettability and implications for hydrological response. Journal of Hydrology 319 (1-4), 295–311. Eaton, B. C., 2008. Potential channel changes following wildfire: applying the UBC Regime Model to Fishtrap Creek, BC. Streamline 12 (1), 28–33. Eaton, B., Andrews, C., Giles, T., Phillips, J., in Press. Wildfire, morphologic change and bed material transport at Fishtrap Creek, British Columbia. Geomorphology. Eaton, B., Church, M., 2004. A graded stream response relation for bed load-dominated streams. Journal of Geophysical Research-Earth Surface 109 (F03011) 1–18. Eaton, B., Giles, T., 2009. Assessing the effect of vegetation-related bank strength on channel morphology and stability in gravel-bed streams using numerical models. Earth Surface Processes and Landforms 34 (5), 712–724. Eaton, B., Hassan, M., Phillips, J. C., 2008. A method for using magnetic tracer stones to monitor changes in stream channel dynamics. Streamline 12 (1), 22–27. 128 Bibliography Eaton, B., Lapointe, M., 2001. Effects of large floods on sediment transport and reach morphology in the cobble-bed Sainte Marguerite River. Geomorphology 40 (3-4), 291– 309. Eaton, B., Moore, R., Giles, T., 2009. Forest fire, banks strength, and channel instabil- ity: the ”unusual” response of Fishtrap Creek, British Columbia, (In Press). Eaton, B. C., Moore, R. D., Giles, T. R., in press. Forest fire, bank strength and channel instability: the “unusual” response of Fishtrap Creek, British Columbia. Earth Surface Processes and Landforms. Eaton, B. C., 2006. Bank stability analysis for regime models of vegetated gravel bed rivers. Earth Surface Processes and Landforms 31 (11), 1438–1444. Eaton, B. C., Church, M., Davies, T. R. H., 2006. A conceptual model for meander initi- ation in bedload-dominated streams. Earth Surface Processes and Landforms 31 (7), 875–891. Environment Canada, 2009. National climate data and information archive: Canadian climate normals, 1971-2008. URL http://climate.weatheroffice.ec.gc.ca/ Flannigan, M., Logan, K., Amiro, B., Skinner, W., Stocks, B., 2005. Future area burned in Canada. Climatic Change 72 (1-2), 1–16. Fuller, I., Large, A., Charlton, M., Heritage, G., Milan, D., 2003. Reach-scale sediment transfers: an evaluation of two morphological budgeting approaches. Earth Surface Processes and Landforms 28 (8), 889–903. Gaeuman, D., Schmidt, J. C., Wilcock, P. R., 2005. Complex channel responses to changes in stream flow and sediment supply on the Lower Duchesne River, Utah. Geomorphology 64 (3-4), 185–206. Galay, V., 1983. Causes of river bed degradation. Water Resources Research 19 (5) 1057–1090. Gardner, J. S., 1977. Some geomorphic effects of a catastrophic flood on the Grand River, Ontario. Canadian Journal of Earth Science 14, 2294–2300. Gippel, C., 1995. Environmental hydraulics of large woody debris in streams and rivers. Journal of Environmental Engineering-ASCE 121 (5), 388–395. 129 Bibliography Gippel, C., Finlayson, B., ONeill, I., 1996a. Distribution and hydraulic significance of large woody debris in a lowland Australian river. Hydrobiologia 318 (3), 179–194. Gippel, C., O’Neill, I., Finlayson, B., 1992. The hydraulic basis of snag management. Land and Water Resources Research and Development Canberra, Australia. 91 (2), 610–61. Gippel, C., ONeill, I., Finlayson, B., Schnatz, I., 1996b. Hydraulic guidelines for the re-introduction and management of large woody debris in lowland rivers. Regulated Rivers-Research & Management 12 (2-3), 223–236. Goff, J., Ashmore, P., 1994. Gravel transport and morphological change in braided Sunwapta River, Alberta, Canada. Earth Surface Processes and Landforms 19 (3), 195–212. Gomez, B., Church, M., 1989. An assessment of bed load sediment transport formulae for gravel bed rivers. Water Resources Research 25 (6). Gomi, T., Sidle, R., Bryant, M., Woodsmith, R., 2001. The characteristics of woody debris and sediment distribution in headwater streams, southeastern Alaska. Cana- dian Journal of Forest Research-Revue Canadienne de Recherche Forestiere 31 (8), 1386–1399. Graham, D., Reid, I., Rice, S., 2005. Automated sizing of coarse-grained sediments: image-processing procedures. Mathematical Geology 37 (1), 1–28. Gurnell, A., Piegay, H., Swanson, F., Gregory, S., 2002. Large wood and fluvial pro- cesses. Freshwater Biology 47 (4), 601–619, 1st International Symposium on Riverine Landscapes, Ascona, Switzerland, Mar 25-30, 2001. Ham, D., Church, M., 2000. Bed-material transport estimated from channel morphody- namics: Chilliwack River, British Columbia. Earth Surface Processes and Landforms 25 (10), 1123–1142. Haschenburger, J., Wilcock, P., 2003. Partial transport in a natural gravel bed channel. Water Resources Research 39 (1), 1020. Hassan, M., Church, M., 2000. Experiments on surface structure and partial sediment transport on a gravel bed. Water Resources Research 36 (7), 1885–1895. Hassan, M., Egozi, R., Parker, G., 2006. Experiments on the effect of hydrograph char- acteristics on vertical grain sorting in gravel bed rivers. Water Resources Research 42 (9), W09408. 130 Bibliography Hickin, E., 1984. Vegetation and river channel dynamics. Canadian Geographer 28 (2), 111–126. Hoffman, D., Gabet, E., 2007. Effects of sediment pulses on channel morphology in a gravel-bed river. Bulletin of the Geological Society of America 119 (1-2), 116–125. Hogan, D., 1989. Channel response to mass wasting in the Queen Charlotte Islands, British Columbia: Temporal and spatial changes in stream morphology. In: Pro- ceedings of Watersheds’ 89, A conference on the stewardship of soil, air and water resources, Juneau, Alaska, March 21-23, 1989, United States Department. pp. 125– 144. Inbar, M., Tamir, M., Wittenberg, L., 1998. Runoff and erosion processes after a forest fire in Mount Carmel, a mediterranean area. Geomorphology 24, 17–33. Jackson, M., Roering, J. J., 2009. Post-fire geomorphic response in steep, forested land- scapes: Oregon coast range, USA. Quaternary Science Reviews 28 (11-12), 1131 – 1146, natural Hazards, Extreme Events and Mountain Topography. Jackson, W., Beschta, R., 1982. A model of two-phase bedload transport in an Oregon Coast Range stream. Earth Surface Processes and Landforms 7 (6), 517–527. Johnson, P., Simon, A., 1997. Effect of channel adjustment processes on reliability of bridge foundations. Journal of Hydraulic Engineering-ASCE 123 (7), 648–651. Jones, L. S., Harper, J. T., 1998. Channel avulsions and related processes, and large- scale sedimentation patterns since 1875, Rio Grande, San Luis valley, Colorado. Ge- ological Society of America Bulletin 110 (4), 411–421. Jordan, P., Covert, A., Curran, M., June 2009. Post-wildfire mass movement and ero- sion events in British Columbia, 2003-2008. FORREX, Kelowna, BC, wildfire and watershed hydrology: workshop proceedings Edition. Keller, E., Swanson, F., 1979. Effects of large organic material on channel form and fluvial processes. Earth Surface Processes and Landforms 4 (4), 361–380. Keller, E., Valentine, D., Gibbs, D., 1997. Hydrological response of small watersheds following the Southern California Painted Cave Fire of June 1990. Hydrological Pro- cesses 11 (4), 401–414. Kiley, D., Schneider, R., 2005. Riparian roots through time, space and disturbance. Plant and Soil 269 (1), 259–272. 131 Bibliography Kondolf, G., 2000. Assessing salmonid spawning gravel quality. Transactions of the American Fisheries Society 129 (1), 262–281. Lane, E., 1957. A study of the shape of channels formed by natural streams flowing in erodible material. MRD Sediment series 9. Lane, S., 1998. The use of digital terrain modelling in the understanding of dynamic river channel systems. Landform monitoring, modelling and analysis, 311–342. Lane, S., Chandler, J., Richards, K., 1994. Developments in monitoring and modelling small-scale river bed topography. Earth Surface Processes and Landforms 19 (4), 349–368. Lane, S., Richards, K., 1997. Linking river channel form and process: time, space and causality revisited. Earth Surface Processes and Landforms 22 (3), 249–260. Lane, S., Westaway, R., Hicks, D., 2003. Estimation of erosion and deposition volumes in a large, gravel-bed, braided river using synoptic remote sensing. Earth Surface Processes and Landforms 28 (3), 249–271. Larsen, I. J., MacDonald, L. H., Brown, E., Rough, D., Welsh, M. J., Pietraszek, J. H., Libohova, Z., de Dios Benavides-Solorio, J., Schaffrath, K., 2009. Causes of post-fire runoff and erosion: Water repellency, cover, or soil sealing? Soil Science Society of America Journal 73 (4), 1393–1407. Legleiter, C. J., Lawrence, R. L., Fonstad, M. A., Marcus, W. A., Aspinall, R., 2003. Fluvial response a decade after wildfire in the northern yellowstone ecosystem: a spatially explicit analysis. Geomorphology 54 (3-4), 119 – 136. Leopold, L., Maddock, T., 1953. The hydraulic geometry of stream channels and some physiographic implications, Professional Paper 252. US Geological Survey, Washing- ton, DC. Lindsay, J., Ashmore, P., 2002. The effects of survey frequency on estimates of scour and fill in a braided river model. Earth Surface Processes and Landforms 27 (1). Liebault, F., Piegay, H., 2001. Assessment of channel changes due to long-term bedload supply decrease, Roubion River, France. Geomorphology 36, 167–186. Lisle, T. E., 1982. Effects of aggradation and degradation on riffle-pool morphology in natural gravel channels, northwestern California. Water Resources Research 18 (6), 1643–1651. 132 Bibliography Lisle, T., 1989. Sediment transport and resulting deposition in spawning gravels, north coastal California. Water Resources Research 25 (6), 1303–1319. Lisle, T., Hilton, S., 1992. The volume of fine sediment in pools - an index of sediment supply in gravel-bed streams. Water Resources Bulletin 28 (2), 371–383. Lisle, T., Hilton, S., 1999. Fine bed material in pools of natural gravel bed channels. Water Resources Research 35 (4), 1291–1304. Lisle, T., Iseya, F., Ikeda, H., 1993. Response of a channel with alternate bars to a decrease in supply of mixed-size bed load: A flume experiment. Water Resources Research 29 (11), 3623–3629. MacDonald, L., Huffman, E., 2004. Post-fire soil water repellency persistence and soil moisture thresholds. Soil Science Society of America Journal 68 (5), 1729–1734. Madej, M., 2001. Development of channel organization and roughness following sediment pulses in single-thread, gravel bed rivers. Water Resources Research 37 (8), 2259–2272. Madej, M., Ozaki, V., 1996. Channel response to sediment wave propagation and move- ment, Redwood Creek, California, USA. Earth Surface Processes and Landforms 21 (10), 911–927. Makin, J., 1948. Concept of the graded river. Geol. Soc. Amer. Bull 59, 463–512. Manga, M., Kirchner, J. W., 2000. Stress partitioning in streams by large woody debris. Water Resources Research 36, 2373–2380. Marcus, W. A., Marston, R. A., Colvard, C. R., Gray, R. D., 2002. Mapping the spatial and temporal distributions of woody debris in streams of the greater yellowstone ecosystem, USA. Geomorphology 44 (3-4), 323 – 335. Martin, Y., Church, M., 1995. Bed-material transport estimated from channel surveys: Vedder River, British Columbia. Earth Surface Processes and Landforms 20 (4), 347– 361. Meyer, G. A., Wells, W. G., 1997. Fire-related sedimentation events on alluvial fans, Yellowstone National Park, USA. Journal of Sedimentary Research 67, 776–791. McLaren, P., 1981. An interpretation of trends in grain size measures. Journal of Sedi- mentary Research 51 (2), 611–624. 133 Bibliography McLean, D. G., Church, M., 1999. Sediment transport along lower Fraser River – 2. estimates based on the long-term gravel budget. Water Resources Research 35 (8), 2549–2559. McNeil, W., Ahnell, W., 1964. Success of pink salmon spawning relative to size of spawning bed materials. US Fish and Wildlife Service. Meyer, G., Wells, S., 1997. Fire-related sedimentation events on alluvial fans, yellow- stone national park, u.s.a. Journal of Sedimentary Research 67 (5), pp. 776–791. Meyer-Peter, E., Muller, R., 1948. Formulas for bed-load transport. International Asso- ciation of Hydraulic Research. In: Proceedings of Second Meeting, Stockholm. Vol. 3. pp. 39–64. Millar, R., 2000. Influence of bank vegetation on alluvial channel patterns. Water Re- sources Research 36 (4). Millar, R., 2005. Theoretical regime equations for mobile gravel-bed rivers with stable banks. Geomorphology 64 (3-4), 207–220. Miller, D., Benda, L., 2000. Effects of punctuated sediment supply on valley-floor land- forms and sediment transport. Geological Society of America Bulletin 112 (12), 1814– 1824. Ministry of Forest and Range B.C, 2009. Ministry of forest and range b.c. wildfire management branch: Summary of previous fire seasons. URL http://bcwildfire.ca/History/SummaryArchive.htm#2003 Minshall, G., Robinson, C., Lawrence, D., 1997. Postfire responses of lotic ecosystems in Yellowstone National Park, USA. Canadian Journal of Fisheries and Aquatic Sciences 54 (11), 2509–2525. MoFR, 2009. Biogeoclimatic zones of british columbia. URL http://www.for.gov.bc.ca/HRE/becweb/resources/maps/map download.html Montgomery, D., Buffington, J., 1997. Channel-reach morphology in mountain drainage basins. Bulletin of the Geological Society of America 109 (5), 596–611. Montgomery, D., Buffington, J., Peterson, N., Schuett-Hames, D., Quinn, T., 1996. Stream-bed scour, egg burial depths, and the influence of salmonid spawning on bed surface mobility and embryo survival. Canadian Journal of Fisheries and Aquatic Sciences 53 (5), 1061–1070. 134 Bibliography Montgomery, D., Buffington, J., Smith, R., Schmidt, K., Pess, G., 1995. Pool spacing in forest channels. Water Resources Research 31 (4), 1097–1105. Montgomery, D., Collins, B., Buffington, J., Abbe, T., 2003a. Geomorphic effects of wood in rivers. In: Gregory, SV and Boyer, KL and Gurnell, AM (Ed.), Ecology and management of wood in world rivers. Vol. 37 of American Fisheries Society Sympo- sium. pp. 21–47, International Conference on Wood in World Rivers, Corvallis, OR, OCT 23-27, 2000. Montgomery, D., Massong, T., Hawley, S., 2003b. Influence of debris flows and log jams on the location of pools and alluvial channel reaches, Oregon Coast Range. Geological Society of America Bulletin 115 (1), 78–88. Moody, J., Martin, D., 2001. Initial hydrologic and geomorphic response following a wildfire in the Colorado Front Range. Earth Surface Processes and Landforms 26 (10), 1049–1070. Moody, J. A., Martin, D. A., Cannon, S. H., 2008. Post-wildfire erosion response in two geologic terrains in the western USA. Geomorphology 95 (3), 103–118. Moody, J. A., Martin, D. A., 2009. Synthesis of sediment yields after wildland fire in different rainfall regimes in the western United States. International Journal of Wildland Fire 18 (1), 96–115. Moore, R.D., Eaton, B., June 2009. Streamflow response to wildfire and salvage harvest- ing at Fishtrap Creek. FORREX, Kelowna, BC, wildfire and watershed hydrology: workshop proceedings Edition. Murray, A., Paola, C., 2003. Modelling the effect of vegetation on channel pattern in bedload rivers. Earth Surface Processes and Landforms 28 (2), 131–143. Naiman and, R., Décamps, H., 1997. The ecology of interfaces: riparian zones. Annual Review of Ecology and Systematics 28 (1), 621–658. Neill, C., 1987. Sediment balance considerations linking long-term transport and channel processes. Sediment Transport in Gravel-Bed Rivers. John Wiley and Sons New York. 1987. p 225-249, 9 fig, 25 ref. Nelson, P. A., Venditti, J. G., Dietrich, W. E., Kirchner, J. W., Ikeda, H., Iseya, F., Sklar, L. S., 2009. Response of bed surface patchiness to reductions in sediment supply. Journal of Geophysical Research-Earth Surface 114. 135 Bibliography Nitschke, C. R., Innes, J. L., 2008. Climatic change and fire potential in south-central British Columbia, Canada. Global Change Biology 14 (4), 841–855. Owens, P. N., Blake, W. H., Petticrew, E. L., 2006. Changes in sediment sources follow- ing wildfire in mountainous terrain: a paired-catchment approach, British Columbia, Canada. Water, Air, and Soil Pollution: Focus 6, 637–645. Paige, A., Hickin, E., 2000. Annual bed-elevation regime in the alluvial channel of Squamish River, southwestern British Columbia, Canada. Earth Surface Processes and Landforms 25 (9), 991–1009. Perucca, E., Camporeale, C., Ridolfi, L., 2007. Significance of the riparian vegetation dynamics on meandering river morphodynamics. Water Resources Research 43 (3), W03430. Petticrew, E., Owens, P., Giles, T., 2006. Wildfire effects on the quantity and composi- tion of suspended and gravel-stored sediments. Water, Air, and Soil Pollution Focus (2006) (6), pp.647–656. Phillips, J. C., 2007. Post-Fire dynamics of a gravel bed stream: Fishtrap Creek, British Columbia. Master’s thesis, University of British Columbia. Pierson, F., Robichaud, P., Spaeth, K., 2001. Spatial and temporal effects of wildfire on the hydrology of a steep rangeland watershed. Hydrological Processes 15 (15), 2905–2916. Pitlick, J., 1993. Response and recovery of a subalpine stream following a catastrophic flood. Geological Society of America Bulletin 105 (5), 657–670. Pyrce, R., Ashmore, P., 2003. Particle path length distributions in meandering gravel- bed streams: results from physical models. Earth Surface Processes and Landforms 28 (9), 951–966. Ralph, S., Poole, G., Conquest, L., Naiman, R., 1994. Stream channel morphology and woody debris in logged and unlogged basins of western Washington. Canadian Journal of Fisheries and Aquatic Sciences 51 (1), 37–51. Rice, S., Church, M., 1998. Grain size along two gravel-bed rivers: Statistical variation, spatial pattern and sedimentary links. Earth Surface Processes and Landforms 23 (4), 345–363. 136 Bibliography 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 Sci- ences 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. Canadian Journal of Fisheries and Aquatic Sciences 58 (2), 282–292. Ryder, J., Fulton, R., Clague, J., 1991. The Cordilleran ice-sheet and the glacial geo- morphology of southern and central British-Columbia. Geographie Physique et qua- ternaire 45 (3), 365–377. Sandwell, D., Texas Univ., A., 1987. Biharmonic spline interpolation of GEOS-3 and SEASAT altimeter data. Geophysical Research Letters 14 (2), 139–142. Scherer, R. A., 2008. The effects of wildfire disturbance and streamside clearcut har- vesting on instream wood and small stream geomorphology in south-central British Columbia. PhD Thesis, University of British Columbia. Scott, D., 1997. The contrasting effects of wildfire and clearfelling on the hydrology of a small catchment. Hydrological Processes 11 (6), 543–555. Scrivener, J., Brownlee, M., 1981. A preliminary analysis of Carnation Creek gravel quality data, 1973-1980. Salmon spawning gravel: a renewable resource in the Pacific Northwest, 195–226. Sedimetrics Digital Gravelometer, 2009. Sedimetrics digital gravelometer documentation (version 1.0). URL http://www.sedimetrics.com/documentation/data collection.html Shakesby, R., Chafer, C., Doerr, S., Blake, W., Wallbrink, P., Humphreys, G., Harring- ton, B., 2003. Fire severity, water repellency characteristics and hydrogeomorpholog- ical changes following the Christmas 2001 Sydney forest fires. Australian Geographer 34 (2), 147–175. Shakesby, R., Doerr, S., 2006. Wildfire as a hydrological and geomorphological agent. Earth-Science Reviews 74 (3-4), 269 – 307. 137 Bibliography Silins, U., Bladon, K., Stone, M., Boon, S., Emelko, M., Williams, CHS and Wagner, M., June 2009. Effects of wildfire and salvage logging on hydrologic and sediment regimes in Alberta’s Southern Rocky mountains. FORREX, Kelowna, BC, wildfire and watershed hydrology: workshop proceedings Edition. Simon, A., Rinaldi, M., 2006. Disturbance, stream incision, and channel evolution: The roles of excess transport capacity and boundary materials in controlling channel response. Geomorphology 79 (3-4, Sp. Iss. SI), 361–383. Sloan, J., Miller, J. R., Lancaster, N., 2001. Response and recovery of the Eel River, California, and its tributaries to floods in 1955, 1964 and 1997. Geomorphology 36, 129–154. Spigel, K. M., Robichaud, P. R., 2007. First-year post-fire erosion rates in Bitterroot National Forest, Montana. Hydrological Processes 21 (8), 998–1005. Stojic, M., Chandler, J., Ashmore, P., Luce, J., 1998. The assessment of sediment transport rates by automated digital photogrammetry. Photogrammetric Engineering and Remote Sensing 64 (5), 387–395. Surian, N., Rinaldi, M., 2003. Morphological response to river engineering and manage- ment in alluvial channels in Italy. Geomorphology 50 (4), 307–326. Tal, M., Paola, C., 2007. Dynamic single-thread channels maintained by the interaction of flow and vegetation. Geology 35 (4), 347. Taylor, S., Flannigan, M., Moore, R., van der Kamp, D., Meyn, A., Regnierre, J., Amant, R. S., June 2009. Wildfire risk in British Columbia: A global context for re- gional change. FORREX, Kelowna, BC, wildfire and watershed hydrology: workshop proceedings. Tribe, S., 2005. Eocene paleo-physiography and drainage directions, southern Interior Plateau, British Columbia. Canadian Journal of Earth Sciences 42 (2), 215–230. Walling, D., Fang, D., 2003. Recent trends in the suspended sediment loads of the world’s rivers. Global and Planetary Change 39 (1-2), 111–126. Warrick, J. A., Rubin, D. M., 2007. Suspended-sediment rating curve response to urban- ization and wildfire, Santa Ana River, California. Journal of Geophysical Research- Earth Surface 112 (F2). 138 Bibliography Wathen, S. J., Hoey, T. B., 1998. Morphological controls on the downstream passage of a sediment wave in a gravel-bed stream. Earth Surface Processes and Landforms 23, 715–730. Webb, A. A., Erskine, W. D., 2003. Distribution, recruitment, and geomorphic signifi- cance of large woody debris in an alluvial forest stream: Tonghi Creek, southeastern Australia. Geomorphology 51 (1-3), 109 – 126. Wells, W. G., 1987. The effects of fire on the generation of debris flows in southern California. In: Costa, J., Wieczorek, G. (Eds.), Debris Flows/Avalanches: Process, Recognition, and Mitigation. Vol. VII. Geological Society of America Review in En- gineering Geology, Boulder, CO, pp. 105–114. Whiting, P. J., King, J. G., 2003. Surface particle sizes on armoured gravel streambeds: Effects of supply and hydraulics. Earth Surface Processes and Landforms 28, (13), 1459–1471. Wilcock, P., McArdell, B., 1993. Surface-based fractional transport rates: Mobilization thresholds and partial transport of a sand-gravel sediment. Water Resources Research 29 (4). Wilcock, P., McArdell, B., 1997. Partial transport of a sand/gravel sediment. Water Resources Research 33 (1). Winkler, R. D., Spittlehouse, D. L., Golding, D.L., 2005. Measured differences in snow accumulation and melt among clearcut, juvenile, and mature forests in British Columbia. Hydrological Processes 19, 51-62. Winkler, R., June 2009. Snow accumulation and ablation post-wildfire on the Thomp- son Plateau. FORREX, Kelowna, BC, wildfire and watershed hydrology: workshop proceedings Edition. 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. Wolman, M., 1954. A method of sampling coarse river-bed material. Transactions of the American Geophysical Union 35 (6), 951–956. Wondzell, S., King, J., 2003. Postfire erosional processes in the Pacific Northwest and Rocky Mountain regions. Forest Ecology and Management 178 (1-2), 75–87. 139 Bibliography Woods, S. W., Balfour, V. N., 2008. The effect of ash on runoff and erosion after a severe forest wildfire, Montana, USA. International Journal of Wildland Fire 17 (5), 535–548. Zelt, R. B., Wohl, E. E., 2004. Channel and woody debris characteristics in adjacent burned and unburned watersheds a decade after wildfire, park county, wyoming. Ge- omorphology 57 (3-4), 217 – 233. 140 Appendices 141 Appendix A: List of Incorporated Work Eaton, B.C., Andrews, C.A.E., Giles, J.R., Phillips, J.C., (in press). Wildfire, morpho- logic change and bed material transport at Fishtrap Creek, British Columbia. Geomor- phology. • Text and figures from the paper have been incorporated into Chapter 4, Sections 4.1 to 4.3. • Data from this paper are used in Chapter 6 for the analysis of net change and bed material transport rates and patterns . Eaton et al. (in press) was written by B. Eaton and edited by all other authors. The work is based on data collection and analysis conducted by B. Eaton, C. Andrews and J. Phillips. 142

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.24.1-0069407/manifest

Comment

Related Items