UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Quantifying the variability in forest stream channel morphology Trainor, Kristie Marie 2001

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

Item Metadata

Download

Media
831-ubc_2001-0124.pdf [ 5.9MB ]
Metadata
JSON: 831-1.0089911.json
JSON-LD: 831-1.0089911-ld.json
RDF/XML (Pretty): 831-1.0089911-rdf.xml
RDF/JSON: 831-1.0089911-rdf.json
Turtle: 831-1.0089911-turtle.txt
N-Triples: 831-1.0089911-rdf-ntriples.txt
Original Record: 831-1.0089911-source.json
Full Text
831-1.0089911-fulltext.txt
Citation
831-1.0089911.ris

Full Text

QUANTIFYING THE VARIABILITY IN FOREST STREAM CHANNEL MORPHOLOGY by KRISTIE MARIE TRAINOR B.Sc, The University of British Columbia, 1995 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in THE F A C U L T Y OF G R A D U A T E STUDIES (Department of Geography) We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA January, 2001 © Kristie Marie Trainor, 2001 In presenting this thesis in partial fulfillment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of Geography The University of British Columbia Vancouver, Canada 31 January 2001 Abstract Within forests, large variability exists in stream channel morphology. Recognizing this variability is important when attempting to characterize, quantify and/or compare stream channels. This point becomes extremely significant when considering the idea of "restoring" streams, a concept which seems to imply an ideal or "target" state. The idea of target states is intricately connected to stream channel variability, and it is this variability which is explored in this project. The study design incorporates comparison of stream channels, the drainage basins of which have similar biophysical, morphometric, and hydroclimatic characteristics. These characteristics are all known to affect or exert considerable influence on the processes which occur in forest streams. Numerous contingencies may also affect channel morphology locally. The key research objective is to determine the range of variability that these streams (under similar basic governing conditions and theoretically similar channel morphologies) possess. Nine stream channel characteristics (channel unit frequency, channel unit length, pool spacing, depth variability, width variability, LWD jam spacing, LWD volume, relative roughness, and average bankfull width [used as a surrogate for scale]) are measured in 12 old growth reaches and 6 managed reaches in the Queen Charlotte Islands and Vancouver Island. The data are split into six groups: all old-growth, uncoupled old-growth (stream is buffered from hillslopes by floodplain), coupled old-i i growth (stream receives material directly from adjacent hillslopes), selected old-growth, managed, and 'old-growth vs. managed'. Within each group, the stream channel characteristics can be analyzed by calculating the dissimilarity (a form of Euclidean distance measure) for all possible reach pair combinations. Frequency distributions based on the resulting dissimilarity values are constructed for each group. These distributions express the range of variability present in the streams analyzed. The resulting range of dissimilarity values precludes the definition of a single, ideal target state. However, the dissimilarity method of comparing stream channel reaches does enable definition of ranges of desirable or undesirable states and quantification of impact. Dissimilarity values for the 'all old-growth' reach pair group ranged from 2.73 to 10.92. For this reach pair group high dissimilarity was judged to be greater than or equal to 8.56. This value does not by any means constitute a regional reference dissimilarity value, as the sample size is simply too small. Reach pairs exhibiting high dissimilarity values tend to have significant differences in several key stream channel characteristics. These key stream channel characteristics vary between reach pairs. Those reaches consistently appearing in reach pairs with high dissimilarity values are considered 'severely impacted' (within this system of comparison). i i i Table of Contents Abstract ii Table of Contents iv List of Tables vii List of Figures x Acknowledgements xii Dedication xiii C h a p t e r 1 I n t r o d u c t i o n 1 1.1 Research Objectives 3 1.2 Organization of the Study 5 C h a p t e r 2 D r a i n a g e B a s i n s 7 2.1 Introduction 7 2.2 Drainage Basin Classification 9 2.3 Assessment of Drainage Basin Classification 16 2.4 Drainage Sub-Basin Classification and Selection 23 2.4.1 Selection of Old-Growth Sub-Basins 24 2.4.2 Selection of Managed Sub-Basins 26 2.5 Summary 29 C h a p t e r 3 S t r e a m C h a n n e l C h a r a c t e r i s t i c s 31 3.1 Introduction 31 3.2 Selection of Stream Channel Characteristics 31 3.2.1 Channel Unit Frequency and Length 31 iv 3.2.2 Pool Spacing 35 3.2.3 Width and Depth Variability 36 3.2.4 Large Woody Debris (LWD) 36 3.2.5 Sediment (Relative Roughness) 37 3.2.6 Summary 38 3.3 Field Methodology 38 3.4 Stream Channel Sub-Reach Selection 42 3.4.1 Representative Reach Lengths 44 3.4.2 Stream Channel Sub-Reach Selection Procedure 47 3.4.3 Selection of Old-Growth Stream Channel Sub-Reaches 49 3.4.4 Selection of Managed Stream Channel Sub-Reaches 55 3.5 Summary 60 Chapter 4 A Method for Stream Channel Comparison 61 4.1 Introduction 61 4.2 Dissimilarity 62 4.3 Cheong's Dissimilarity Testing Procedure 63 4.4 Stream Channel Dissimilarity Testing Procedure 65 4.5 Summary 68 Chapter 5 Results and Discussion 69 5.1 Introduction 69 5.2 Old-Growth Stream Channels 69 5.2.1 All Reach Pair Combinations 72 5.2.2 Uncoupled Reach Pair Combinations 77 v 5.2.3 Coupled Reach Pair Combinations 79 5.2.4 Selected Reach Pair Combinations 81 5.2.5 Discussion 83 5.3 Managed Channels 89 5.3.1 Uncoupled Reach Pair Combinations 91 5.4 Old-Growth vs. Managed Stream Channels 93 5.4.1 Uncoupled Reach Pair Combinations 93 5.5 Discussion 94 Chapter 6 Conclusions 98 Bibliography 103 Appendix A Measurement Methodology 108 Appendix B Variance Plots for Old-Growth and Managed Reaches 110 vi List of Tables Table 1 Attributes of the Morphometric Features of a Drainage Basin 11 Table 2 Basin Characteristics Measured from 1:50 000 NTS Maps Used to Classify and Compare Individual Watersheds 13 Table 3 Watershed Classification Types 14 Table 4 Drainage Basin Similarity Assessment 15 Table 5 General Characteristics of Old-Growth QCI and VI Watersheds 20 Table 6 Drainage Basin/Stream Reach Similarity Assessment 23 Table 7 Additional Biogeophysical, Morphometric and Hydroclimatic Characteristics of Old-Growth QCI and VI Watersheds 24 Table 8 General Characteristics of Managed QCI Watersheds 27 Table 9 Final Selection of Old-Growth Sub-Basins for Project Database 30 Table 10 Final Selection of Managed Sub-Basins for Project Database 30 Table 11 Channel Unit Types and their Associated Characteristics 34 Table 12 Selected Stream Channel Characteristics 38 Table 13 LWD Classification 40 Table 14 Classification of the Span of a LWD Jam 41 Table 15 Classification of LWD Jam Integrity 41 Table 16 Classification of LWD Jam Height 41 Table 17 Classification of LWD Jam Age 41 Table 18 Classification of LWD Jam Location 41 vii Table 19 Classification of LWD Jam Shape 42 Table 20 Old-Growth Stream Channel Reach Lengths 46 Table 21 Managed Stream Channel Reach Lengths 47 Table 22 Sample Spreadsheet Illustrating Method of Calculating Variances (Jason Lower) 48 Table 23 Old-Growth Sub-Reach Sections to be Analyzed 52 Table 24 Managed Sub-Reach Sections to be Analyzed 57 Table 25 Old-Growth Sub-Reach Selection 60 Table 26 Managed Sub-Reach Selection 60 Table 27 Stream Channel Characteristics - Units, Information Type 65 Table 28 Selected Stream Channel Characteristics 68 Table 29 Dissimilarity Matrix Table 71 Table 30 Dissimilarity Results - All Possible Reach Pair Combinations 72 Table 31 Dissimilarity Results - Uncoupled Reach Pair Combinations 77 Table 32 Dissimilarity Results - Coupled Reach Pair Combinations 80 Table 33 Dissimilarity Results - Selected Reach Pair Combinations 82 Table 34 Spearman Rank Correlation Test for Selected Old-Growth Reach Pair Combinations 85 Table 35 Spearman Rank Order Correlations 85 Table 36 Dissimilarity Matrix Table for Managed Sub-Reaches 90 Table 37 Dissimilarity Results - All Managed Reach Pair Combinations 91 Table 38 Dissimilarity Results - Selected Old-Growth vs. Managed Reach Pair Combinations 93 viii Table 39 Comparison of Mean Dissimilarity Values and Standard Deviations Between Different Reach Pair Combination Groups 94 Table 40 Dissimilarity Results - Reference Set (Selected, Uncoupled Old-Growth Reach Pairs) 96 Table 41 Dissimilarity Results - Reference Set with Mosquito Upper 96 ix List of Figures Figure 1 Location Map of Queen Charlotte Islands 17 Figure 2 Location Map of Vancouver Island 18 Figure 3 Downstream Organization of Stream Channel Units 33 Figure 4 Completed Sketch of Cross-Section at 560m - Carmanah Creek Survey 39 Figure 5 Scaled Diagram of Carmanah Creek (120m - 240m) 43 Figure 6 Mean Runoff Volume and Catchment Size 45 Figure 7 Illustration of the 2 Primary Sampling Lengths that Require Definition: Reach Length and Measurement Interval Length 45 Figure 8 Illustration of a Fitted 2 n d -Order Polynomial Regression Line (Estimated Thalweg Elevation) Superimposed on a Longitudinal Profile (Actual Thalweg Elevation) 48 Figure 9 Plot of Horizontal Distance vs. Variance of Depth Deviation - Jason Lower 49 Figure 10 Variance Plots: Entire Reach Length and Selected Sub-Reach Length for Inskip SB 50 Figure 11 Variance Plot for Government NB NF 52 Figure 12 Variance Plots for Government U M : Sections 800m - 1740m and 1000m- 1925m 53 Figure 13 Variance Plot: Selected Sub-Reach Length for Government U M 53 Figure 14 Variance Plots for Gregory U M : Sections 4664m - 5746m and 4886m -5866m 54 Figure 15 Variance Plot: Selected Sub-Reach Length for Gregory U M 54 Figure 16 Variance Plot for Jason Lower 55 Figure 17 Variance Plots for Riley Lower: Sections Om - 1275m, 250m - 1525m, 500m - 1775m, and 1525m - 2800m 58 Figure 18 Final Variance Plots for Riley Lower 1 (0m - 1525m), and Riley Lower 2 (1525m-2800m) 58 Figure 19 Variance Plots for Tarundl (0m - 825m, and 405m - 1410m) 59 Figure 20 Final Variance Plot for Peel (0m - 800m) 59 Figure 21 Frequency Distribution of Dissimilarity Values: (A) All Old-Growth Reach Pairs, (B) All Uncoupled Old-Growth Reach Pairs, (C) All Coupled Old-Growth Reach Pairs, and (D) All Selected Old-Growth Reach Pairs 86 Figure 22 Frequency Distribution of Dissimilarity Values for All Old-Growth Reach Pair Combinations (Lognormal Distribution Superimposed) 87 Figure 23 Frequency Distribution of Dissimilarity Values for Selected Old-Growth Reach Pair Combinations (Normal Distribution Superimposed) 88 Figure 24 Box and Whisker Plot for Groups A - C: A = Selected Old-Growth (uncoupled), B = Managed (uncoupled), and C = Selected Old-Growth (uncoupled) vs. Managed (uncoupled) 95 xi Acknowledgements First and foremost, I wish to sincerely thank my supervisor, Dr. Michael Church, for all his support, encouragement and sage advice, and Dan Hogan, for his guidance, helpful suggestions and discussions. I also wish to thank Steve Bird for all his help delving into the FFIP archives; Tony Cheong for his advice and comments; Dr. Olav Slaymaker for his encouragement; and Elaine Cho for her help with things administrative. This project was funded by Forest Renewal British Columbia, Department of Geography teaching assistantships, and a research assistantship from Dr. Church. A very special thanks to Craig Jones and Dave Campbell for their good humour and assistance in the field ("Great! Only 5 more kilometres to hike before we start working "). My dear departmental comrades Russ White and Dave Oldmeadow were also instrumental in keeping things afloat, both literally and figuratively. Finally, I wish to thank my friends and family, and in particular, Craig, whose love and support were a help beyond measure. xii for my dad, David Thomas Trainor Chapter 1: Introduction The physical processes that occur in forest streams are reflected in the channel morphology. These processes are subject to a variety of factors, the fundamental ones being the volume and time distribution of water supplied to the stream; the volume, character and timing of sediment conveyed to the channel; the terrain through which the stream flows; and the geologic history of the local landscape (Church, 1992). Other factors which can influence channel morphology include climate, the character of riparian vegetation, and land use (including direct modification of the channel) within a watershed (Church, 1992). Within forests, large variability exists in stream channel morphology (Montgomery and Buffington, 1997). This variability can occur between different portions of a single stream, between tributaries of a single system, or between different stream systems, particularly with respect to sediment supply and transport, channel geometry, and characteristics of structural features such as large woody debris (Wood-Smith and Buffington, 1996). Recognizing this variability is important when attempting to characterize, quantify and/or compare stream channels. This point becomes extremely significant when considering the idea of "restoring" streams, a concept which seems to imply an ideal or "target" state. Can a target state (or range of states) be defined, considering the abundant variability that exists in stream channel morphology? It can be argued that the idea of identifying disturbed stream channels is not properly founded without some measure of the variability that exists in undisturbed channels. l Such measures are not currently available. Forest Renewal British Columbia (FRBC) has invested more than $300 million dollars in watershed restoration projects (ASPECT editorial, August 2000). The success of this initiative has been seriously questioned, as the project goal (to return watersheds to conditions more similar to those found in undisturbed watersheds) is not clearly defined or quantified (ibid.). The notion of an ideal or "target" state is intricately connected to the variability of stream channel morphology, and it is this variability which will be formally explored in this thesis. Attempts to compare stream channels commonly focus on (1) channel unit1 characteristics (the proportions, spacing, slope and shape of channel units) (Keller and Melhorn, 1978; Grant et al., 1990; Montgomery et al., 1995); and (2) changes in large woody debris (LWD) (Keller and Tally, 1979; Hogan, 1986, 1989; Andrus et al., 1988; Ralph et al., 1994; Wood-Smith and Buffington, 1996). Several key weaknesses stand out in the current literature. Although basin morphometry (defined here as quantitative measurement and generalization of the land surface geometry) is cited as a critical factor contributing to channel form (Hogan, 1986), attempts to systematically account for it have been inadequate. Basin morphometry will be addressed in this thesis by use of Cheong's (1996) watershed classification program and other common morphometric indices. While the use of channel units in morphological studies is clearly profitable, the lack of clarity in defining them is a limitation. Further exploration of other stream channel descriptors such as width and depth variability is required, and will be addressed in this thesis. As stated previously, large variability exists in all the properties just listed. 1 Channel units consist o f various types o f pools and shallows that are the basic morphological components o f a reach (Hogan and Church, 1989). They are important descriptors o f aquatic habitat and are often used as indicators o f a stream's response to land-use changes. 2 This variability is not obviously correlated or structured from property to property, nor has it been shown to be systematically associated with disturbance in any simply predictive sense. Some measure of this variability is essential for comparing stream channels or identifying disturbed ones. This project represents a first attempt at quantifying the variability inherent in stream channel morphology. 1.1 Research Objectives The key research objective is to determine the range of variability that streams under similar basic governing conditions, and hence with theoretically similar channel morphologies, possess. Time has been substituted by space in order to study both logged (managed) and unlogged (old-growth) stream channels. This involves a multi-step approach: 1. Compilation of a project database. 2. Determination of which properties best characterize stream channels. 3. Development and evaluation of a method of quantifying the variability in stream channels (based on the properties selected in Objective 2). 4. Quantification of the variability inherent in old-growth and in managed stream channels. 5. Assessment of the variability exhibited by old-growth and by managed stream channels. 6. Comparison of the variability in stream channel morphology between old-growth and managed streams. "Old-growth streams" are defined here as those located in forested drainage basins which have not experienced a major human disturbance (e.g., logging). "Managed 3 streams" are defined here as those located in forested drainage basins which have experienced spatially extensive logging. For inclusion in the project database the stream channels had to have similar governing conditions. Geology, hydrological zone, biogeoclimatic zone and basin type were all controlled as they are known to exert considerable influence over the processes which occur in forest streams. This requirement created significant limitations on the size of the project database. Of the remaining intact, old-growth drainage basins in British Columbia, most are found in remote locations that are not accessible given the practical constraints of the project. Of those that exist, it is difficult to find a substantial grouping with suitably similar basic governing conditions. The bulk of the project database is composed of a pre-existing database containing detailed information on four old-growth streams and four managed streams located on the Queen Charlotte Islands. This database was created for the Canada/British Columbia Fish-Forestry Interaction Program (FFIP). A method of comparing stream channel reaches was adapted from a drainage basin testing procedure developed by Cheong (1992). This modified testing procedure involves calculating the dissimilarity between two reaches based on key stream channel characteristics. If this dissimilarity is calculated for a large enough data set (e.g., > 30 reach pair combinations), a frequency distribution of the dissimilarity values can be constructed. This distribution expresses the range of variability present in the streams analyzed. The project design involves compilation and subsequent comparison of both old-growth and managed stream channels whose drainage basins have similar biophysical 4 characteristics (e.g., climate, geology, vegetation and morphometry). The goal is to determine the range of variability that streams, under similar basic governing conditions, hence with theoretically similar channel morphologies, possess. Without some measure of the variability that does exist it is questionable whether the concept of comparing stream channels or identifying disturbed ones is properly founded. As such, the results of this project are expected to benefit other studies and government programs related to stream channel impact assessment and watershed restoration. 1.2 Organization of the Study Chapter 2 presents background information concerning drainage basin description and classification. A comprehensive procedure for analyzing drainage basin similarity is reviewed and adapted for the selection of suitable drainage basins. Detailed descriptions of the selected drainage basins are also presented and assessed. Chapter 3 includes a critical discussion regarding key issues related to stream channel characterization. The rationale for choosing specific stream channel characteristics for use in this project is explained. Limitations of data collection procedures are assessed and the methodology utilized to obtain the project data is described. Using the database made available through FFIP, the question of representative reach lengths will also be explored and assessed. Based on that analysis stream channel sub-reaches are selected for use in this project. 5 A method of quantifying the variability inherent in stream channels is developed and evaluated in Chapter 4 . The concept of dissimilarity is introduced and a comprehensive dissimilarity testing procedure is presented. The summary stream channel data required for the dissimilarity analysis are presented as a matrix table in Chapter 5. The results obtained from quantifying the variability in stream channel morphometry are discussed along with the issue of what constitutes critical dissimilarity. Sub-groups of the selected basins (based on characteristics such as channel configuration) are contrasted and compared. Managed stream channels are also introduced to the data set and analyzed. Chapter 6 presents the conclusions and recommendations for further research. 6 Chapter 2; Drainage Basins 2.1 Introduction Drainage basins have been recognized as fundamental process-response elements of the landscape since the beginning of the 19th century (Chorley et al., 1984). Playfair (1802) described the adjustment of a system of valleys contributing to the main trunk stream, and Gilbert (1877, cited in Rodda, 1976) referred to the dynamic equilibrium affecting all drainage lines and their flanking slopes. R. E. Horton (1945) extended the work of W. M . Davis (1899), describing the morphometry of drainage basins and rationalizing their features on the basis of hydrological process (Cheong, 1992). Horton's central concept of stream ordering provided 'the touchstone by which drainage net characteristics could be related to each other and to hydrologic and erosional processes' (Bowden and Wallis, 1964). This analysis established a basis for quantitative description and comparison of drainage basins. The work of Horton, further developed by A. N. Strahler (1964), established the erosional drainage basin as a fundamental geomorphic unit because it appeared to be: 1. a limited, convenient, usually clearly defined and unambiguous topographic unit; and 2. a physical process-response system open to material and energy transfer systems. (Chorley, et al., 1984) From as early as the beginning of the twentieth century, most basin studies have been comparative in nature. These comparative basin studies have incorporated criteria regarding location, geological structure, basin area and vegetation in order to gauge the 7 similarity between experimental basins. Comparisons made under this approach have remained largely qualitative and subjective. Presumably this is due to the perceived complexity of the comparison criteria. The first basin studies of modern times were conducted in Switzerland during the 1890's (Engler, 1919, cited in Rodda, 1976). Basin studies commenced in other parts of the world at the beginning of the twentieth century (see review in Rodda, 1976). While many studies have been published on experimental basins, many criticisms have been voiced over their validity. These criticisms and concerns relate to how well drainage basins are characterized. For example, Riekerk (1989) analyzed the effects of silvicultural practices on hydrology in three basins, one of which was a control. Although climate and vegetation were "consistent" over the three basins, soil structure and characteristics were not. Cronan et al., (1990) compared the aluminum biogeochemistry of two watersheds in the eastern United States. Although the two watersheds were of similar size, there were differences in climate, soil type and composition, and vegetation cover. No standard method of comparing two drainage basins has been widely accepted. As Hewlett (1971) stated, "the theory of the paired catchment experiment is basically simple, but has been widely questioned because no thorough treatise of the method has ever been published". Various procedures have been used to determine similarity, but no single method has become dominant, and rarely are these procedures explained in published documents (Cheong, 1992). The description of basins usually involves three types of parameters: morphometric, biogeophysical, and historical (Cheong, 1992). Historical information (e.g., fire history, mass wasting events) is not widely used. Descriptions of 8 drainage basins have generally been limited to biogeophysical parameters (e.g., geology, soils, vegetation, and climate - which are the dominant controls of many processes within the basin). The issue of basin morphometry is often not adequately addressed. For example, Riekerk (1989) made no mention of watershed morphometry, although it influences the amount and timing of runoff. Many of the problems with experimental basins relate to the uncertainty of their representativeness and the subsequent difficulty of transferring results from one landscape to another. Basin comparison techniques are generally quite subjective, and the procedures are usually based upon simple dissimilarity statistics, the researcher's judgement and time/cost considerations. Cheong (1992) argues that, while most "similar" basins may be identified using these techniques, the degree of similarity may not always be adequate for a particular study or even sufficiently objective to permit assured judgements. 2.2 Drainage Basin Classification If drainage basins and stream channel morphology are to be related to the geologic, climatic, and hydrologic character of a basin, then it becomes necessary to describe features quantitatively in order to formally investigate relations (Chorley et al., 1984). It was Horton's work (1945) which transformed the study of drainage basins and channel networks from a purely qualitative and deductive study to a rigorous quantitative science (Strahler, 1964). The science of morphometry is concerned with the quantitative measurement and generalization of the land surface geometry. Despite the large number of indices proposed by various workers, it has been argued that relatively few fundamental aspects of basin form are actually measured by the 9 available indices (Gardiner and Park, 1978). Table 1 summarizes the most frequently cited morphometric variables. In a review and evaluation of morphometric parameters, Mark (1975) ascertained that all important terrain information was contained within these variables. The majority of the parameters listed in Table 1 are used either in a descriptive sense or in a physiographic classification. Only the hypsometric integral, / , has been related to geomorphic processes (Mark, 1975). In 1952 Strahler proposed that the value of the hypsometric integral reflects the "stage" of landscape development. Where resistant bedrock maintains a portion of the summit plane during considerable erosion of the rest of the basin, / may reach low values. In uniformly erodible material the continued erosion of the basin high point may stabilize / in a middle range of values between roughly 0.4 and 0.6 (Chorley et al., 1984). Therefore using I as a measure of erosion is limited to situations where the elevation of the original summit plane can be estimated. More current research has analyzed relations among various morphometric variables (de Villiers, 1986; Tarboton et al., 1989) and classified similar landscape units based upon a detailed analysis of basin morphometry (Ebisemiju, 1986). Although these studies have quantified landscape comparison procedures more rigorously, greater emphasis still needs to be placed on the range of variability in landscape characteristics and measures of what constitutes acceptable 'similarity'. Hogan (1985) found that classical selection criteria such as biogeophysical characteristics and basin morphometry are inadequate for hydrologic purposes. Other researchers have concluded that a greater emphasis must be placed on a more detailed study of landscape morphometry when determining terrain similarity (Melton, 1957; Zavoianu, 1985). 10 Table 1 Attributes of the Morphometric Features of a Drainage Basin Important basin length measurements are: The length o f a stream segment o f a given order. L c The total length o f the channel system within a basin. L B The overall max imum basin length measured from the mouth. L E The length o f overland flow (map distance from a point on a divide orthogonally to the adjacent stream channel). X c The belt o f no sheet erosion. P The perimeter o f the drainage basin. Num >er measures: N„ The number o f streams o f a given order. Under the Strahler system a stream o f a given order (u+1) is initiated at the junction o f two streams o f the next lower order u. R b The bifurcation ratio, (Rb=N u /N u + 1 ) . It is a dimensionless number varying between approximately 3.0 and 5.0 for networks formed in homogenous rocks; it is fairly insensitive to al l but the most important structural controls. Meaningful areal variables used to define basin morphometry are: A u The area o f a drainage basin o f a given order. A The total area o f a given drainage basin. D The drainage density, equal to L c / A (initially defined by Horton in 1945). It is found to be closely related to mean stream discharge, mean annual precipitation and sediment y ie ld. D s The source density, which is very closely related to drainage density. It is the number o f stream sources per unit area. Both D and D s are very sensitive to possible map-to-map inconsistencies in the portrayal o f the drainage net. Dp The peak density, equal to the number o f closed hil ltop contours per unit area. F The stream frequency, wh ich expresses the number o f stream segments o f al l orders per unit area (=2N U /A) . Rc The circularity ratio (=A/A C , where A c = the area o f a circle having the same perimeter P as the basin). Re The elongation ratio (=d A /L B , where d A = the diameter of a circle o f area A ) . The relief of a basin may be described by: H Relief, which expresses the elevation difference between high and low points. This measure was first introduced by Partsch (1911) and was first used in the Engl ish language in 1935 (Mark, 1975). H a The available relief. This concept was introduced by G lock in 1932 and was rephrased by Johnson in 1933 (Mark, 1975). It is the vertical distance from the former position o f an upland surface down to the position o f adjacent graded streams. / The hypsometric integral. It expresses the unconsumed volume o f a drainage basin as a percentage o f that delimited by the summit plane, base plane and perimeter, and is the percentage area under the dimensionless curve relating h/H and a/A. H d Drainage relief, defined as the vertical distance between adjacent divides and streams. Strahler stated that local relief, H , was a measure o f the vertical distance from stream to adjacent divide, but this is only true i f the sample areas upon which local rel ief is based are o f an appropriate size (Mark, 1975). And inally, significant gradient measures include: The maximum slope of the ground surface at a given point. m^ax The maximum angle o f a given valley-side slope profi le. Sc The slope o f a reach o f a stream channel at a point or averaged over a reach. H / L B A dimensionless " re l ie f ratio", actually a generalized gradient measure. From this and closely related to mean slope is the ruggedness number (=H*D). HHA A dimensionless " re l ie f ratio", actually a generalized gradient measure. H/P A dimensionless "re l ief ratio", actually a generalized gradient measure. Taken from Mark , (1975) and Chorley et al. , (1984). 11 Ebisemiju (1986) stated that the findings of applied morphometric intercorrelation studies can be valid only if the observations are from a homogeneous environment. He argued that drainage basins with a wide range of environmental conditions should not be lumped together in studies aimed at highlighting the interdependence of morphometric attributes. His analysis suggests that different environments produce some variations in the interdependence of drainage basin morphometric properties because the runoff and erosion mechanics governing stream channel initiation, growth and integration and the development of drainage basin systems vary from one environment to another. When observations from such different environments are lumped together, each region loses its individuality, the differences between regions are masked, and significantly confounded interaction patterns may emerge. Classification based on similarity of basin morphometry is an approach used by some researchers (Lewis, 1969; de Villiers, 1986; Cheong, 1992). This approach incorporates both the geometric (e.g., area, stream length) and the topologic (e.g., stream order) characteristics of the watershed into a statistical index of similarity. Many of these studies use principal component analysis and/or correlation analysis to determine similar regions (Cheong, 1996). The work done by Cheong (1992) likely represents the first comprehensive, clearly defined quantitative procedure for classifying basins based on basin morphometry as well as biogeophysical parameters. Regional studies within the Queen Charlotte Islands were conducted to demonstrate this procedure. This work was extended in Cheong's 1996 12 study, in which an abridged, more user-friendly 3-level assessment procedure for determining drainage basin similarity is presented. Level 1 similarity is based on a general analysis comparing morphometry and large scale biophysical characteristics. In order for two basins to be similar at level 1, they must be in the same hydrological zone, have similar general geology, biogeoclimatic characteristics and watershed type. Watershed type is determined through a computer program which analyzes specific morphometric parameters (Table 2) that are highly correlated to many hydrological and geomorphological processes. Dissimilarity between basins is calculated using Euclidean distance standardized using standard deviations (Cheong, 1992, 1996). Table 2 Basin Characteristics Measured from 1:50 000 NTS Maps Used to Classify and Compare Individual Watersheds Watershed character is t ics: Def in i t ions: A rea (km 2 ) The drainage area o f the watershed. Perimeter (km) The length o f the perimeter o f the watershed. Mean basin elevation (m) The mean elevation o f the drainage basin. Ice (km 2 ) The total ice covered area within the basin. Lake (km 2 ) The total lake covered area within the basin. Va l ley flat extent (km 2 ) Total area with a gradient less than 7% and adjacent to the drainage network. Mean basin gradient (m/m) The mean gradient o f the drainage basin. Re l ie f The difference between the highest and lowest points in the basin. Elevation Rel ie f Ratio The proportion o f highland and lowland with respect to mean elevation. K w Shape Factor (km / k m ) The circularity o f the basin in relation to basin area and basin length. Steepland area (km 2 ) The total area with a gradient o f greater than 60%. Channe l character is t ics: Def in i t ions: Drainage density (km/ km 2 ) The total channel length divided by the drainage area. Max . channel elevation (m) The highest point on the channel M i d channel elevation (m) The elevation o f the point Vi the total distance up the mainstream M i n . channel elevation (m) The lowest point on the channel Ma in channel gradient Change in elevation over the main channel divided by channel length. Total channel length (km) The sum o f the length o f al l channels in the drainage basin. M a i n channel length (km) The length o f the longest channel from its point of origin to the outlet. Magnitude The total number o f first order streams. Taken from Cheong, (1996) and B i rd , (1998). 13 Using data from 506 randomly chosen watersheds in British Columbia, Cheong (1996) found that four main variables (ice cover, lake area, valley flat and steepland area) could be used to classify watersheds into 11 main groups (Table 3). Level 2 similarity requires the following characteristics to be similar: hydrological zone, geology, soils, biogeoclimatic/vegetation characteristics, climate (e.g., precipitation intensity, total precipitation, snowfall) and basin type. Geology, soils, and biogeoclimatic zone are based on information presented in 1:50 000 maps. Table 3 Watershed Classification Types T y p e l Relatively low proportions o f ice and val ley flat, high proportions o f lake and steepland Type 2 High proportions o f ice and lake, low proportions o f steepland Type 3 L o w proportions o f ice and very high proportions o f valley flat Type 4 L o w proportions o f ice, lake, and steepland and high proportions o f valley flat Type 5 Quite high proportions o f ice and relatively large amounts o f steepland Type 6 High proportions of ice, low proportions o f steepland and lakes Type 7 L o w proportions o f ice and lake, large extents o f steepland and low proportions o f valley flat Type 8 L o w proportions o f ice and lake cover, and relatively large extents o f steepland and valley flat Type 9 High proportions o f ice and steepland Type 10 L o w proportions o f ice, lake, steepland and valley flat Type 11 L o w amounts o f ice, valley flat and steepland, high proportions o f lakes Taken from Cheong, (1996). Level 3 assessment is appropriate for very detailed studies of watersheds. It is based upon detailed bedrock geology, soils/surficial geology, biogeoclimatic zones/vegetation analysis, climate (e.g., total precipitation and precipitation intensity, snowfall, temperature) and hydrology (e.g., hydrographic analysis, flood frequency analysis). Geology, soils, and biogeoclimatic zone are based on information presented in 1:20 000 maps. Level 3 similarity assessment is further based upon all morphometric parameters being within standard deviation limits and limited total dissimilarity. Table 4 summarizes the 3 levels of drainage basin similarity assessment. 14 Table 4 Drainage Basin Similarity Assessment Leve l 1 Leve l 2 Leve l 3 Geology Simi lar geology based on 1:250 000 or 1:50 000 maps Simi lar geology based on 1:50 000 or smaller scale maps Detailed bedrock geology analysis based on 1:20 000 maps or better So i ls /Sur f ic ia l Geology Simi lar soils and terrain based on 1:50 000 maps Simi lar soils, terrain, surf icial geology based on detailed 1:20 000 or better scale maps, mass wasting history Hyd ro logy Simi lar hydrological zone Simi lar hydrological zones, similar hydrograph characteristics Hydrograph analysis, f lood frequency analysis C l ima te Simi lar precipitation intensity, total precipitation, snowfal l Simi lar precipitation intensity, total precipitation, snowfal l , frost free days Vegetat ion Simi lar biogeoclimatic zone 1:2 000 000 mapping Simi lar biogeoclimatic zone from detailed mapping, 1:250 000 or 1:50 000 mapping Similari ty from biogeclimatic mapping (1:20 000) or vegetative surveys S im i la r i t y Simi lar basin type Simi lar basin type and dissimilarity within group ranges Simi lar basin type and dissimilarity within group ranges and below crit ical similarity Taken from Cheong, (1996). In summary, Cheong's (1996) classification procedure combines two elements: the comparison of filter parameters (nominal level data, e.g., geology) and the assessment of morphometric similarity. The procedure for assessing morphometric similarity varies according to the level involved: A level 1 comparison is based on the similarity of basin type (i.e., two basins are considered similar at level 1 if they are the same type). Two basins are considered similar at level 2 if they are the same type and their parameter values are within a set range. At level 3 basin type must be similar and morphometric parameters are limited to having a total dissimilarity of less than 2.5 (based on standardized dissimilarity distribution characteristics). As well as classifying basin data by type, Cheong's (1996) computer program can compare two watersheds, or compare and cluster the data for up to 15 drainage basins. 15 2.3 Assessment of Drainage Basin Classification Five intact, old-growth watersheds and four managed watersheds were initially selected for this project. They represent the largest surveyed grouping of small to intermediate-sized streams (upper width limit in the range of 20-30 m) located in the same biogeoclimatic zone. All four of the managed watersheds (Riley Creek, Mosquito Creek, Peel Creek and Tarundl Creek) and four of the old-growth watersheds (Jason Creek, Inskip Creek, Gregory Creek and Government Creek) are found on the Queen Charlotte Islands (see Figure 1). These streams have previously been surveyed by Provincial and Federal Government agencies as part of the FFIP. The remaining old-growth watershed (Carmanah Creek) is located on Vancouver Island (see Figure 2). Carmanah Creek was surveyed by myself and two assistants in August 1999. The old-growth drainage basins are used here to test the efficacy of Cheong's level 1 assessment procedure (1996). Thirteen stream reaches were delimited from the 5 old-growth watersheds. These reaches were initially classified into 2 general categories by field workers: (1) uncoupled and (2) coupled. Coupled streams receive material directly from the adjacent hillslopes by creep and episodic mass-movement processes (Church, 1983; Rice, 1990). Uncoupled streams are generally flanked by floodplains and receive sediment inputs entirely by fluvial means (downstream sediment transport). The distinction between coupled and uncoupled channels can be related to position within a drainage basin: The headwaters area of a basin (where there is often strong coupling between the hillslopes and the channel) supplies sediment to zones of transport and deposition downstream (Schumm, 16 18 1977). Thus the distinction between coupled and uncoupled reaches identifies significant differences in key processes which may affect or alter stream channel morphology. Key biogeophysical and morphometric characteristics of the 13 stream reaches and their associated sub-basins are compared in Cheong's level 1 similarity assessment procedure. Table 5 summarizes the general characteristics of the old-growth watersheds. All areas in question are located in the Coastal Western Hemlock biogeoclimatic zone and fall under the same major subdivision in the Western System subcategory of the Outer Mountain Area physiographic region (Holland, 1965). Two different types of basin classification are included in the last two columns of Table 5. The Cheong classification was based on the level 1 assessment procedure outlined in Table 4. Uncoupled reaches and their associated sub-basins tend to be Type 8, while the coupled reaches tend to be Type 7. This is expected, as Type 8 basins are described as having large extents of valley flat, while Type 7 basins are characterized by low proportions of valley flat. The Hogan et al. classification was derived from Cheong's (1996) multivariate procedure which calculates a dissimilarity matrix among sub-basins that can then be analyzed by cluster analysis. The cluster analysis delineated 5 watershed types (Types A through E - see notes under Table 5). The coupled reaches contain mainly Types C through E (with some exceptions) while the uncoupled reaches tend to be Types A through C. Again, this is expected, as channel response to landslides and forestry activities is expected to increase accordingly moving from Type A to Type C (Hogan etal., 1998). However, discrepancies were also apparent when looking at Table 5. Gregory Upper 19 •V ii JS ii « a « I—( u a p f l H-< o o o on ii « U JS U "« u a a « — H 00-2 O " ^ CO e <*3 8 ' » 43 J O U o a t - ~ o * . w o o o o v o r - r o o o o o o o o o r—; •—; ON r«-j t—; d H oo d so » n H CN"— •—' ro >—' •—' • — i ro < co u u oo oo. t- oo oo v© Tfr oo — so O — — ~* O o o o o o m >n m oo o\ CN 00 CN —' 0O ro --^  CN CN ro u * t/3 &| .2 -S CQ "o .Br k "5 I o o a o OH 00 l - ~ r ~ o c N C N r o r o ( " > t N ( N n v i i c i i f ) ( S t N O CN O M O N as o o W 00 u u o u u a u u o o o o o o o o o o o o o o o o r o r o r o r o r o r o r o r o A A A A A A A A oi oi oi oi oi oi n n o o o o o o O h M ' ! O O > r-^  ^ r r 2 t"" o \ O 0 \ 1*^  o CN cn co ^ ON _H V© m o CN r - o CN ro so <o O U u a U o o o o _ o o o o 2 SO SO SO SO ~ ro co ro ro JZ? A A A A m m X U X X oi oi oi oi u> U O U U g r - ro ro : C "33 •§ c b co oo J= IS U S3 O o •a "E. 3 o o c D 60 O X E o a .S c IS H Pi u PH ^ J3 & i-i _o -a DH o o CD 60 O 5 c •co u is a oo c o J3 U i CD •3 CO 00 T3 CO § T3 O. CL> 43 oo c 1 CO = -a I o. CO CD o CN J3 00 43 -2 CO co > CO CQ CO CD S3 to 53 CO > £ CO CO CQ t « 2 <; -> w CO U 13 +H 00 o Main has been classified as a coupled reach by field surveyors and a Type 7 watershed, which is consistent with field observations. However, it has also received a Type A classification, which supposedly represents an uncoupled, buffered basin. In addition, Jason Upper has been classified as an uncoupled reach by field surveyors. However, using Cheong's classification program Jason Upper is classified as a Type 7 watershed, commonly associated with coupled reaches/sub-basins. Similarly, Government NB is a coupled reach that has been classified as a Type 8 (i.e. uncoupled) watershed. These inconsistencies highlight several points. First, the sensitivity of these classification methods needs to be assessed. Considering that the largest sub-basin in the study group is only 35.2 km 2, it is questionable whether a classification method based on information gleaned from 1:50 000 topographic maps is appropriate. This could possibly explain the lack of definition between coupled and uncoupled reaches. Apart from stream channel gradient, the split between coupled reaches and uncoupled reaches is not reflected in the data. Values for steepland area, valley flat extent and mean basin gradient do not vary significantly between coupled reaches and uncoupled reaches. Considering that the Hogan et al. classification is derived from the Cheong classification, it is also odd that there are discrepancies between the two methods. It is worth noting that there was limited confidence in the discriminatory power of the cluster analysis expressed by the authors themselves (D. Hogan, 2000, personal commun.). It is also important to point out that no formal study has been undertaken to test the efficacy of the Cheong program. 21 From these initial results it becomes clear that the Cheong (1996) level 1 assessment procedure cannot be solely used in this project to determine suitably similar sub-basins. The main problem is that the map criteria are not sensitive enough for the small basins and sub-basins initially selected for this project. It is not clear that 1:50 000 scale base maps adequately resolve steepland area and valley flat area properties. More importantly, field assessments of "coupled" or "uncoupled" channel configuration may be reach specific, and are certainly not resolved from maps, or systematically correlated with valley flat area. Furthermore, assuming the Cheong program adequately accounts for basin shape, it is still possible that other factors may have a more significant impact on basin processes and, ultimately, the stream channel morphology (which is the main focus in this project). For example, two basins with similar morphometric properties could have significantly different surficial geology. This could affect slope stability, vegetation, hydrological processes and stream channel morphology. A compromise between pure morphometric comparison and formal drainage basin classification (as attempted by Cheong) appears to be the best approach for selecting suitable sub-basins. Suitability of stream reaches and their associated sub-basins was based on a 2- stage procedure: 1. Comparison of filter parameters: This is analogous to Cheong's level 2 assessment of qualitative biophysical parameters (see Table 4). The subjective quality of the filter approach is advantageous due to the complexity of the available information: It allows emphasis to be 22 placed on those elements that contribute to fluvial variability, which is most important for this particular study. 2. Comparison of morphometric and hydroclimatic parameters: This is partly analogous to Cheong's level 1 assessment of basin similarity (see Table 4). While the Cheong basin classification scheme does not appear to effectively characterize the small sub-basins initially selected for this project, some general quantitative comparisons can be made and will assist in the selection of suitable stream reaches and their associated sub-basins. Table 6 summarizes the assessment procedure used in this project. Table 6 Drainage Basin/Stream Reach Similarity Assessment Stage 1 Biogeophysica l F i l t e r Parameters : Stage 2 Morphome t r i c /Hyd roc l ima t i c Parameters : • Simi lar geology based on 1:125 000 maps • Simi lar soils and terrain based on 1:50 000 maps • Same physiographic zone • Simi lar total precipitation • Same biogeoclimatic zone from 1:50 000 mapping • Same hydrological zone (Church, 1997) • Simi lar mean basin elevation, ice cover, lake area, val ley flat area, shape factor and steepland area (here defined as >60% slopes) • Simi lar drainage density, channel gradient and stream magnitude (here defined as total number o f 1 s t order channels) • Simi lar hydrograph characteristics 2.4 Drainage Sub-Basin Classification and Selection The 2-stage procedure developed in Section 2.3 is used here to classify and select suitably similar sub-basins. As stated previously, five old growth watersheds and four managed watersheds were initially selected for this project. From these 9 watersheds 13 old-growth sub-basins and 6 managed sub-basins were delimited. 23 2.4.1 Selection of Old-Growth Sub-Basins Table 7 summarizes the additional old-growth reach characteristics (not found in Table 5) required for the 2-stage procedure. Physiographic zone, biogeoclimatic zone, hydrological zone, total precipitation and soils meet the Stage 1 assessment criteria. Table 7 Additional Biogeophysical, Morphometric and Hydroclimatic Characteristics of Old-Growth QCI and VI Watersheds Channel Configuration Sub-basin Coupled Govt. N B Govt. N B N F Govt . N B E F Inskip M a i n Inskip N B Inskip S B Gregory N B Gregory Upper M a i n Uncoupled Govt. M a i n Govt. Upper M a i n Jason Upper Jason Lower Carmanah Upper [Taken in part from Hogan et al, 1998] Soils P = Podzol ic soils (mainly ferro-humic podzols). Soils Mean basin Ice Lake Shape Drainage Stream elevation Cover Cover Factor Density Magnitude (m) (km 2 ) (km 2 ) (km/km 2 ) P 207 0 0 3.43 0.924 2 P 257 0 0 1.45 0.923 1 P 213 0 0 4.19 0.999 1 P 454 0 0.3 0.82 0.481 2 P 467 0 0 1.12 0.474 1 P 484 0 0.3 0.83 0.408 1 P 342 0 0 1.22 0.397 1 P 324 0 0 1.80 0.443 5 P 183 0 0 1.72 0.837 6 P 168 0 0 2.14 0.896 3 P 373 0 0 0.88 0.346 1 P 347 0 0 1.19 0.358 1 P 403 0 0.02 9.3 1.56 42 Podzolic soils are found along the southwest coast of Graham Island and over most of Moresby Island (Valentine et al., 1978). They are moist to wet over most of the year and rarely freeze to any significant depth. Soil texture tends to be medium to coarse and leaching is intense (ibid.). No information is available on the spatial variation of the soils at the sub-basin scale. The geology of the Queen Charlotte Islands is complex. Geochronological schemes, based on time of rock formation, are not nearly as important as general rock type and structure, which directly relates to the erosional processes in a 24 basin. The Karmutsen formation, composed primarily of 'hard' volcanics which are not prone to rapid erosion, dominates in the majority of the sub-basins (Sutherland Brown, 1968; Banner et al., 1983). While the bedrock geology in the Carmanah sub-basin is not part of the Karmutsen formation, it is also primarily volcanic and not extremely prone to erosion. However, in Gregory NB and Gregory Upper Main the Yakoun formation (composed primarily of 'soft' volcanics which are prone to erosion) dominates. In Stage 2 the mean basin elevation, valley flat area and hydrograph characteristics are all reasonably similar. With the exception of Inskip Main and Inskip SB, ice cover and lake area are effectively zero for all the sub-basins. Furthermore, the shape factor values all fall within one order of magnitude of each other. However, some discrepancies are apparent. The proportion of steepland area varies from 0% to as high as 84%. While the majority of reaches have very low stream magnitudes, Carmanah Upper stands out with a stream magnitude of 42. This could be due in part to the fact that Carmanah Upper has a contributing area of 35.2 km , the largest in the database. However, Gregory Upper Main has a comparable contributing area (31.1 km ) yet it's magnitude is only 5. This discrepancy could be related to the use of different source base maps. With the exception of Carmanah Upper, all drainage density values fall within one standard deviation of the mean. There are no significant differences in stream channel characteristics between coupled and uncoupled reaches, with the exception of channel gradient. This distinction is not surprising as channel configuration can be related to position within a drainage basin, which can be related to gradient: In general there is a systematic decline in gradient 25 moving downstream from the headwaters region (where there is often strong coupling) to the floodplain. Inskip Main, Inskip SB, Carmanah Upper, Gregory NB and Gregory Upper Main do not meet the assessment criteria summarized in Table 6. While the bedrock geology in all the sub-basins is primarily volcanic in origin, Gregory NB and Gregory Upper Main are questionable in terms of their geologic structure. It is debatable whether this difference in rock strength is significant enough to expel Gregory NB and Gregory Upper Main from the database. Carmanah's extreme stream magnitude value and the lake cover values for Inskip Main and Inskip SB are also questionable. However, as the majority of morphometric parameters for Carmanah, Inskip Main and Inskip SB do conform it is difficult to justify removing them. This is particularly true when bearing in mind the already small number of sub-basins in the database. A moderately-sized database is necessary in order to construct dissimilarity distributions in the final analysis. This requirement outweighs the observed differences in stream channel characteristics and therefore no old-growth sub-basins have been discarded at this stage. 2.4.2 Selection of Managed Sub-Basins Table 8 summarizes the biophysical, morphometric, and hydroclimatic characteristics of the 6 managed sub-basins. It also provides historical information on the nature of the logging disturbance. All areas in question are located in the Coastal Western Hemlock biogeoclimatic zone and fall under the same major subdivision in the Western System subcategory of the Outer Mountain Area physiographic region (Holland, 1965). Physiographic zone, 26 "E 3 i3 on I. u Of V WD A a 1/5 CJ CU es u "a o o c T 3 T3 cu , cu OH •s 12 .a o o v© cn A cn vo oil vo </1 OH © CN Pi u ; a i4 o o cn A o o VO cn A o o VO cn A o o VO cn A cn oo Ov cn © cn OH VO 00 IT) PH o o VO cn A vo 00 Ov © cn VOl oo CN O Ov o p © CN cn ov o o CN o cn f*i D C o g S g e 1 -2 e M * S >* u II II II ^ >- u 5i o "o <o O CN 03 a 00 S3 H s ia a -3 o . co m i u c o N if , 3 i ' i i3 OH 1) c IS CO o o IS l > w ,s CO CO u CO PH -= 00 c <u Q .s ea o O .9 CO m c IS CD 2 I CO s s CO £ 00 H3 O CO oo <D 00 w 2 CD CO o to S3 S x: CD U to 3 M CD ^ a oo 5 PH g O0 oo _o ' lo JS OH o o e CD X o o CD 00 o s biogeoclimatic zone, hydrological zone, total precipitation and soils meet the Stage 1 assessment criteria. However, the geology does not. The Karmutsen formation, composed primarily of 'hard' volcanics which are not prone to rapid erosion, is found in the Mosquito and Peel sub-basins. The bedrock geology in the Riley basin is dominated by the Yakoun formation, which is composed primarily of 'soft' volcanics that are prone to erosion. In addition, the Tarundl basin is composed of mainly Cretaceous shale. In Stage 2 the mean basin elevation, valley flat area and hydrograph characteristics are all reasonably similar. Ice cover and lake area are also effectively zero for all the sub-basins. Furthermore, the shape factor values all fall within one order of magnitude of each other. However, some discrepancies are apparent. The proportion of steepland area varies from 18% to as high as 71%. All drainage density values fall within two standard deviations of the mean. Another concern is the variation in level of disturbance between basins. The '% basin logged' values range from 12% to 65%. In addition, the 'time since logging stopped' values vary from 40 years to 9 years ago. However, the time interval required for slope destabilization to occur due to tree root decay is believed to be approximately 5 years. Therefore all of the selected stream channels may potentially have been affected by logging-related slope instabilities. In addition, logging methods were fairly uniform for all the basins. All basins were logged to the channel bank, and the dominant yarding method used was high lead. 28 While some geologic, morphometric and logging disturbance discrepancies were noted in several of the sub-basins, they were judged not significant enough to warrant expulsion from the database at this time. 2.5 Summary It is very difficult to find any formal, comprehensive procedure with which to classify drainage basins based on both basin morphometry and biogeophysical parameters. The basin classification procedure developed by Cheong (1992, 1996) is the most thorough, systematic approach available that addresses the key concerns that many researchers have voiced over the use of representative basins. Unfortunately, it is not well suited to the smaller-sized sub-basins which are the focus of this project. As a result, a 2-stage comparison method was developed to assess sub-basins for inclusion in the project database (see Table 6). From nine watersheds, 19 sub-basins were initially delimited. They represent the largest surveyed grouping of small to intermediate-sized watersheds located in the same biogeoclimatic zone. Final selection of suitable sub-basins was based on a detailed assessment of both biogeophysical parameters and morphometric indices (as outlined in Table 6). While some geologic and morphometric discrepancies were noted in several sub-basins, they were judged not sufficiently significant to warrant expulsion from the database. Tables 9 and 10 summarize the final selection of sub-basins. 29 Table 9 Final Selection of Old-Growth Sub-Basins for Project Database Watershed Sub-Bas in C h a n n e l Con f igu ra t ion Government Creek Government M a i n Government Upper Ma in ( U M ) Government North Branch (NB) Government North Branch North Fork ( N B N F ) Government North Branch East Fork ( N B E F ) Uncoupled Uncoupled Coupled Coupled Coupled Inskip Creek Inskip M a i n Inskip North Branch (NB) Inskip South Branch (SB) Coupled Coupled Coupled Gregory Creek Gregory Upper Ma in ( U M ) Gregory North Branch (NB) Coupled Coupled Jason Creek Jason Lower Jason Upper Uncoupled Uncoupled Carmanah Creek Carmanah Upper Uncoupled Table 10 Final Selection of Managed Sub-Basins for Project Database Watershed Sub-Bas in C h a n n e l Con f igu ra t ion Mosqui to Creek Mosqui to M a i n Uncoupled Mosqui to Upper Uncoupled R i ley Creek Ri ley Lower Uncoupled Ri ley Midd le Coupled Tarundl Tarundl Uncoupled Peel Peel Uncoupled 30 Chapter 3; Stream Channel Characteristics 3.1 Introduction The purpose of this chapter is to address current issues and problems associated with selecting and sampling key stream channel characteristics. As no clear, quantitative approach exists for determining which variables best characterize stream channels, the task is a difficult and subjective process. Discriminant function analysis has been used to determine which variables best discriminate between old-growth and disturbed stream channels (Wood-Smith and Buffington, 1996), but that is not the first question at hand. As stated previously, it can be argued that the idea of discriminating between disturbed and undisturbed stream channels is not properly founded without knowledge of the natural variability that exists. 3.2 Selection of Stream Channel Characteristics The morphology exhibited by a stream is, in essence, a reflection of the processes that are, or have been, occurring. While these processes are probably the key issue when considering the idea of characterizing stream channels, it is more expedient to focus on the stream channel morphology for several reasons: (1) it can be assessed more readily, and (2) it reflects aquatic habitat quality directly. 3.2.1 Channel Unit Frequency and Length Channel units consist of various types of pools and shallows that are the basic morphological components of a reach (Hogan and Church, 1989). They are also important descriptors of aquatic habitat (Bisson et al., 1982; Sullivan, 1986). Channel 31 unit characteristics (which include the proportions, spacing, slope and shape) are often used as indicators of a stream's response to land-use changes. They reflect sediment input characteristics (Church and Jones, 1982), are independent of both channel pattern and channel materials (Keller and Melhorn, 1978), and represent an important scale for understanding stream dynamics (Grant et al., 1990). Stream classifications which involve channel units are founded on the perception that they are discrete and can be delineated (Naiman et al., 1992). However, consistently identifying these channel units in the field can be difficult (Roper and Scarnecchia, 1995) due in part to the substantial variability that exists within each channel unit type. Headward, non-alluvial channels may have much poorer channel unit distinctiveness compared to downstream channels due to the fact that flow is less able to impose characteristic alluvial organization. One key problem regarding channel unit identification is the lack of universal criteria. Standardized nomenclature and accurate descriptions of structural and functionally distinct morphological features are required. While no set criterion exist for identifying or discriminating channel units, bed slope is one measure commonly used. Channel units generally have distinct slope means and medians (Grant et al., 1990). However, the slope values affixed to each channel unit often differ between studies: Grant et al., (1990) define the average riffle bed slope at 1-1.2%; Wood-Smith and Buffington (1996) define the riffle bed slope at 2-4%. This is likely explained by differences in overall stream gradient. Nevertheless, there is characteristic organization of channel units downstream following the systematic downstream variation in gradient (see Figure 3). 32 Stream 1 ^ ^ s t e p - p o o l St ream 2 ^ r a p i d - p o o l ^ ^ ^ ^ _ ^ r i f f l e - p o o l Figure 3 Downstream Organization of Stream Channel Units. Whi le stream 1 is steeper than stream 2, the downstream organization o f stream channel units is the same. Stream 1 may be smaller, or have relatively larger bed material (or both). This model assumes similar drainage area, and that gradient is decreasing in a downstream direction, which is common in Coastal B C watersheds. There are other watershed types (e.g., in Plateau areas of B C , headwater streams are generally flat and steepen downstream where channels are incised). ( M . Church, 2000, personal commun., D. Hogan, 2000, personal commun.) Montgomery and Buffington (1997) suggest that relative roughness (the ratio between grain diameter and flow depth, D/d) and bed slope together differentiate alluvial reach types. While bed slope is often used, this value can be misleading. For example, it is possible to measure adverse gradients in pools. As an alternative to measuring bed slope, measuring the water surface slope may prove advantageous. In this study, channel units were identified primarily by their topographic, sedimentological, and hydraulic characteristics (Table 11). Some unit types listed in Table 11 are not conventionally defined as channel units. Cascades can be viewed as a compound unit made up of successive steps and pools. Stone lines and log steps are generally viewed as channel unit elements within the cascade channel unit. The unit types listed in Table 11 were selected primarily in order to match those morphological units delineated in the FFIP surveys database. 33 Table 11 Channel Unit Types and their Associated Characteristics Slope Channe l Uni t Type Channe l Un i t Charac ter is t i cs Steep < > Gentle Pool Topographic depressions with heterogeneous substratum 1. Steep < > Gentle G l ide Heterogeneous substratum. Common at pool-riff le and pool-rapid breaks. Steep < > Gentle Riff le Primari ly uniformly distributed al luvial gravel and cobbles. Relative roughness generally >1. Common in larger channels dominated by a sequence of alternating bars with intervening crossovers and average channel gradient less than 1 degree. Steep < > Gentle Rapid Characterized by bed elements (most commonly boulders and cobbles) arranged into irregular ribs oriented roughly perpendicular to the channel and exposed at low flow. Relative roughness is >1 at the ribs and <1 in the secondary pools. Steep < > Gentle Stone Line When clusters (interacting individual clasts o f simi lar size) ramify into extended linear or arcuate features (Church, 1998). They form in the "partial transport regime" found in headward channels, where the largest material present often is lag material del ivered from overbank that rarely moves. Steep < > Gentle L o g Step Near-vertical to vertical steps typical ly created by woody debris oriented transverse to the channel. Steep < > Gentle Cascade M a y be defined by boulders (composed o f step-pool sequences) or by bedrock (where water flows directly on rock). Bedrock cascades are typical ly steeper than boulder cascades. Relat ive roughness is high (>1). Based on Grant el al . , (1990) and Church (1992) One characteristic commonly examined when investigating stream channel morphology is the relative proportion of stream area occupied by a given channel unit (Bisson et al., 1982; Hogan, 1986; Ralph et al., 1994; among others). While this can be a useful measure, it is important to note that the proportions of stream area occupied by different channel units can change with flow. Therefore bias may be introduced if units are identified while the river stage fluctuates. For example, Hogan and Church (1989) found that two similar streams with riffle-pool sequences at low flow generally became more riffle-like as the discharge increased. 1 Two major pool types can be delineated: backwater pools and scour pools (Bisson et al . , 1982; Sul l ivan, 1986). Backwater pools are caused by downstream obstructions and are generally shallow due to accumulation o f sediment. Scour pools are caused by f low convergence over or around an obstruction (Church, 1992) and are relatively deep. Variants o f both scour pools and backwater pools exist. 34 The two channel unit characteristics selected for this project are: (1) channel unit frequency, and (2) channel unit length (see Appendix A). Unit frequency (expressed as a percentage of the total number of channel units) provides a more objective measure than areal proportion (Wood-Smith and Buffington, 1996), although there is a potential for observer bias. For example, identification of scour or plunge pools can be difficult if flow-forming conditions are not active at the time of survey. However, the potential for unit frequency to be a key discriminator between stream channels outweighs the disadvantages associated with the measure. Channel unit length (expressed as a percentage of the total channel length) is perhaps not as robust a measure as channel unit frequency, yet provides important information on the nature of a stream. For example, Jason Lower has a pool unit frequency of 49% (i. e. 49% of the observed channel units in the reach were pools). While this characteristic provides key information, we can better understand the behavior of Jason Lower if we also know that 70% of the reach length is occupied by pools. 3.2.2 Pool Spacing The frequency of pools in a forest stream is a fundamental aspect of channel morphology (Montgomery et al., 1995). In free-formed pool-riffle reaches, pool-to-pool spacing averages 5-7 channel widths (Leopold et al., 1964). For forest streams the pool-to-pool spacing shortens to 3-5 channel widths, and in steeper step-pool systems pool spacing varies between 1 and 4 channel widths (Grant et al., 1990), with substantial variability around the mean values. It has been suggested that this variance is related to the frequency of nonalluvial pool-forming features along the channel margin, such as 35 LWD jams. Montgomery et al. (1995) found that, in forest channels, pool spacing depended on LWD loading, channel type, slope and width. As pool spacing is related to fundamental stream channel characteristics it is an important attribute worth exploring in this project. 3.2.3 Width and Depth Variability Although width and depth variability appear to be significant stream channel characteristics, they are not commonly investigated. Hogan (1986) investigated width and depth variability in relation to LWD occurrence. In zones with abundant debris oriented diagonally across the channel, the width and depth variability increased. Sediment input to stream channels from landslides and debris torrents was also found to initially reduce depth, width and sediment texture variability (Hogan, 1989). As these measures express valuable information relating to the channel structure and its complexity, they will be investigated in this project. 3.2.4 Large Woody Debris (LWD) Channel development in forest streams can be profoundly influenced by the presence of large woody debris (Keller and Swanson, 1979; Nakamura and Swanson, 1993; Abbe and Montgomery, 1996; Wood-Smith and Swanson, 1997; among others). When used as an indicator of stream condition, the total number, volume and/or size classes of LWD pieces and their arrangement or position are commonly examined (Hogan, 1986; Ralph et al., 1994; Montgomery et al., 1995). The mean volume of individual LWD pieces and mean spacing of LWD jam structures will be investigated in this project. 36 3.2.5 Sediment (Relative Roughness) Sediment texture data can be related to stream channel stability. Bed material (that which forms the bed and lower banks of the channel) determines the form of the channel; the stability of the channel is determined by the ease with which bed material can be remobilized. Thus channel disturbance (or instability) should theoretically be reflected in the bed material. Classical bedload formulae postulate that an equilibrium exists between the supply and transport of material through the reach. Furthermore, the grain size distribution of the supply is assumed to be the same as that of the mobilized sediment and output from the reach. However, the bed surface size distribution is not necessarily the same, as it adjusts to match supply size and rate with mobilized sizes and rates. If flow stresses on the bed are modest, relatively few large stones move, and the fine material is either removed or 'hidden' (in the lee of larger stones or in interstices between them). Thus a pavement of coarse material is left on the bed surface. If the transport rate is high, all sizes are moved, and the bed surface texture will become similar to the bed load texture. Therefore the contrast between the surface texture and the bulk texture of the material underneath gives some measure of the relative stability of the bed (M. Church, 1998, personal commun.). This surface-to-subsurface ratio (Dsosurface/Dsosubsuface) is not commonly investigated although it has the potential to provide valuable information. Most of the stream channel data being used in this project is from the Fish-Forestry Interaction Program (FFIP). As these channel surveys were carried out under different research objectives, some limitations exist regarding the available data on sediment size. 37 Unfortunately, only Dgssurface was measured in the FFIP. As channel depth is also known the D95Surface data may be used to calculate relative roughness (the ratio, D /d , between grain diameter and flow depth). 3.2.6 Summary The choice of key stream channel characteristics was somewhat limited because much of the data was originally collected for a different research project which had different objectives. Table 12 summarizes the stream channel characteristics selected for study in this project (see Appendix A for measurement methodology): Table 12 Selected Stream Channel Characteristics Stream channel character ist ics Channel unit frequency Channel unit length Poo l spacing (scaled as multiples o f W b ) Depth Variabi l i ty (m/m) Width Variabi l i ty (m/m) L W D Spacing (scaled as multiples o f W b ) L W D Vo lume (m 2 /m 2 ) Relative Roughness (m/m) 3.3 Field Methodology Of the nine watersheds selected for this project, eight have previously been surveyed for the FFIP. Carmanah Creek was surveyed by myself and two assistants in August 1999. In order to maintain consistency, the survey technique developed and used in the FFIP studies was employed in Carmanah Creek: Bankfull width was initially estimated through use of a regional discharge-bankfull width (Q x Wb) relation. (This was not done for Carmanah Creek - bankfull width was 38 estimated by averaging bankfull width measurements recorded randomly along the selected reach length). Longitudinal profiles were surveyed with an automatic level and stadia rod, and distances were measured with a surveyor's hip chain. Thalweg, water surface, bar and bank elevations were measured at a set interval of one bankfull width (as determined from the regional Q x W b relation). Morphological features ( e.g. breaks separating channel units, the deepest point of a pool) were added as supplementary survey points. Channel units were identified in the field using topographical, sedimentological and hydraulic criteria as outlined in Table 9. At a set interval of five bankfull widths a channel cross-section was surveyed. A fibre tape was strung horizontal and perpendicular to the banks and bankfull width was measured. Horizontal distances were also recorded at significant points in the cross-section (e.g. top of the bank, water surface edge, edge of vegetation). This information, along with the elevation data from the longitudinal profile, was then incorporated into a sketch of the cross-section (see Figure 4). Figure 4 Completed Sketch of Cross-Section at 560m - Carmanah Creek Survey 39 All LWD jams, steps and individual pieces were inventoried every bankfull width. The volume of all in-channel LWD was determined by visual estimation based on the procedure established by Hogan (1989) (see Table 13): Table ] 3 L W D classification R a n k Diameter (m) Length (m) # o f pieces Or ien ta t ion 1 <0.1 1-5 <2 // 2 0.1-0.3 5-10 2-3 3 0.4-0.7 10-15 4-7 4 0.7-1.2 15-20 7-12 1 5 >1.2 >20 >12 Based on Hogan, (1989) Using the ranking scheme outlined in Table 13, the diameter, length, number of pieces, and orientation of LWD were rapidly estimated during the longitudinal profile survey. Log jams (multiple, interacting LWD pieces that influence channel morphology by controlling sediment transport either presently or at some time in the past) were classified according to procedures developed by Hogan (1989) and Hogan and Bird (1998). Jam age was determined from the ages of nursed trees and bar and bank vegetation. If nursed trees were not present, jam age was approximated from the decay characteristics of individual LWD pieces. The number of flood channels and the sediment storage associated with LWD jams were also estimated. The span, integrity, height, age, location, and shape of jams were estimated using the following ranking scheme (see Tables 14 through 19): 40 Table 14 Classification of the Span of a LWD Jam Rank Description of Span 1 Completely crosses channel 2 Incomplete, 0.75-1 W b spanned, scour around end 3 0.5 W b spanned 4 0.25 W b spanned 5 <0.25 W b spanned Based on Hogan, (1989) Table 15 Classification of LWD Jam Integrity Rank Description of Integrity 1 Very sol id, compact, b ig pieces, no rot, anchored 2 Sol id , compact, anchored 3 Moderate, less compact (spaces), rot 4 Weak, poor anchor, rot 5 Very weak, small pieces, no anchor Based on Hogan, (1989) Table 16 Classification of LWD Jam Height Rank Description of Height 1 Above local bank height 2 V* -1 bank height 3 Vi bank height 4 Vi bank height 5 < Vi bank height Based on Hogan, (1989) Table 17 Classification of LWD Jam Age Rank Description of Age 1 Very recent, new trees, no nurse trees (<2 yrs) 2 Recent (2-10 yrs old) 3 Moderate (10-20 yrs old) 4 Moderate to old (20-30 yrs old) 5 O l d (30-50 yrs old) 5+ v. o ld, nurse trees, no bark (>50 yrs old) Based on Hogan, (1989) Table 18 Classification of LWD Jam Location Code General Location Description of Location L B In-channel Left bank R B Right bank M In middle B A l o n g channel A t bend R Bedrock knob W Wad or tree stump T Trees (standing) Based on Hogan, (1989) 41 Table 19 Classification of L W D Jam Shape Code LWD jam shape 1 Perpendicular /L Diagonal II Parallel V Arched (apex downstream) A Arched (apex upstream) Based on Hogan, (1989) In addition to the longitudinal profile and LWD inventory, a scaled diagram was assembled which included the position of channel units, position and orientation of LWD, and the position of cross-section surveys (see Figure 5). Additional notes were taken regarding bank materials and profile, occurrence of bank erosion and apparent cause, and other features of interest. 3.4 Stream Channel Sub-Reach Selection In the stream morphology literature, the term 'reach' is used in different contexts. In the strict sense, a reach is defined as a homogenous unit within which the controlling factors do not change appreciably (Church, 1992). However, it is common to use the term 'reach' to describe any length of channel being studied (in most cases this length of channel is located within a formally defined reach). As no standards exist, surveyed reach lengths tend to vary between studies. A reach length of 50 to 70 W b is considered a conservative measure, and is based on the knowledge that in free-formed pool-riffle reaches, pool-to-pool spacing averages 5-7 W b (Leopold et al., 1964). This ensures a well represented sample of channel units (i. e. roughly 10 units each). However, many studies look at much shorter reach lengths. For example, Montgomery and Buffington 42 4= 200m O o P o o o x x t <-10m-> Small Woody Debris Large Woody Debris Boulders Cobbles Pebbles Sand Vegetation Cross-Section Location Bank Erosion Channel Unit Boundary Riffle Figure 5 Scaled Diagram of Carmanah Creek (120 m - 240 m) (1997) used reach lengths of 10 to 20 channel widths, and Wood-Smith and Buffington (1996) used reach lengths of roughly 20 channel widths. Hogan (1986) used reach lengths of approximately 30 channel widths, but recommended longer reaches for future studies, in order that certain channel features such as LWD clustering could be 43 investigated more thoroughly. Hogan and Bird (1998) have determined that the largest LWD jams can influence sedimentation and ultimately channel morphology for distances exceeding 100 channel widths. Arbitrarily choosing a reach length for comparison between streams is problematic. In order to characterize the variability between streams, the reach lengths and measurement intervals used for a given stream must be representative of that system. 3.4.1 Representative Reach Lengths Determining a reach length at which the variance of a stream channel characteristic stabilizes can be viewed as somewhat analogous to the Representative Elementary Area concept (REA) of Wood et al. (1988). They applied this concept to catchment hydrology, and found that the variance of mean runoff volume, mean infiltration volume and rainfall volume all displayed a similar pattern (Figure 6). Beyond a threshold catchment area the variances stabilized; 1 km was deemed to represent the REA for their study catchment. The variances stabilized with increasing catchment size because at low resolution (spatially extended average) the small scale variability approached randomness, and thus could be represented stochastically. Further complicating this representative reach length issue is the question of representative intervals of measurement. For two of the selected stream channel characteristics (width and depth variability), some investigation into the appropriate interval length for sampling channel widths and depths (e. g. every Wb, or every 5 Wb) 44 0.6 2 0 1 1 1 " I I i 0 4 00 800 1200 1600 200 0 240 0 2800 Figure 6 Mean Runoff Volume and Catchment Size. 1 pixel = 30 m2. Taken from Wood et al., (1988). must be undertaken. Figure 7 illustrates the 2 sampling lengths in question. As the REA concept is based on sample size, these two sampling lengths are intimately connected to each other. Presumably one could either have a long representative reach length with a long interval of measurement, or a short representative reach length with a more intensive interval of measurement. However, there is likely a characteristic reach length within which full variability is expressed. Varying the measurement spacing would vary the precision with which variability is defined. Lengthening the reach could improve the precision merely because the sample size grows. Representative Reach Length > Figure 7 Illustration of the 2 Primary Sampling Lengths that Require Definition: Reach Length and Measurement Interval Length 45 In order to explore the behavior of these 2 sampling lengths it is necessary to have access to a data set with a relatively concentrated measurement regime over a considerable distance. The FFIP database satisfies nearly all these requirements: Channel surveys start at stream outlets and continue uninterrupted to the headwaters. Therefore the survey data are comprised of entire reaches (where reach is defined as the maximum channel length within which governing conditions do not change). Tables 20 and 21 summarize the channel lengths associated with each reach. With these survey data it is possible to quantitatively explore the issue of what constitutes a representative reach length. However, depth measurements were surveyed at an interval less than or equal to 1 Wb and width measurements were surveyed at an interval of 5 Wb. With such a moderately large minimum interval of measurement (particularly for widths) it is not feasible to explore the behavior of measurement intervals. Rather, the interval of measurement used in the FFIP studies will be held constant as representative reach lengths are determined. Considering the direct relation between representative reach length and representative measurement interval, this will not adversely affect the results obtained. Table 20 Old-Growth Stream Channel Reach Lengths Reach Length Leng th (m) (Wb) Government Ma in 800 25 Government U M 1860 101 Government N B 793 39 Government N B N F 871 82 Government N B E F 672 40 Inskip M a i n 649 28 Inskip N B 270 18 Inskip S B 700 48 Gregory U M 2165 100 Gregory N B 420 31 Jason Lower 1205 65 Jason Upper 720 34 Carmanah Upper 1000 26 46, T a b l e 2 1 M a n a g e d S t r e a m C h a n n e l R e a c h L e n g t h s Reach Leng th Leng th (m) (Wb) Mosqui to Ma in 2022 70 Mosqui to Upper 943 29 Ri ley Lower 5325 210 Ri ley M idd le 1481 43 Peel 1200 69 Tarundl 2910 176 3 . 4 . 2 S t r e a m C h a n n e l S u b - R e a c h S e l e c t i o n P r o c e d u r e The FFIP depth data, from which depth variability will be calculated, are best suited for calculating representative reach lengths. There are two main reasons for this: (1) Depths were surveyed over the longest possible distances (i. e. entire reaches) using a reasonably rigorous measurement interval (less than or equal to 1 Wb), and (2) the longitudinal depth profile reflects important structural elements of the channel and is related to channel unit types (which reflect aquatic habitat quality). For each reach longitudinal profiles were constructed. Second-order polynomial regression lines (which estimated the average thalweg elevations) were then fitted to the longitudinal profiles (see Figure 8), and depth deviations, actual thalweg elevation -estimated thalweg elevation, were calculated. Once depth deviations were calculated, the variance of those deviations could be calculated for increasingly large groupings of the data (see Table 22). 47 23.5 Horizontal Distance (m) Figure 8 Illustration of a Fitted 2 -Order Polynomial Regression Line (Estimated Thalweg Elevation) Superimposed on a Longitudinal Profile (Actual Thalweg Elevation) Table 22 Sample Spreadsheet Illustrating Method of Calculating Variances (Jason Lower) Hor i zon ta l A c t u a l Es t imated |Difference| Sect ion o f Va r i ance Distance E levat ion E levat ion (Depth Reach Used to O f (m) (m) (m) Deviat ion) Ca lcu la te V a r i a n c e |Differences| 0 7.92 6.96 0.96 0m - 20m 0.00378 8 8.01 7.01 1.00 0m - 40m 0.02567 9 8.05 7.02 1.03 0m - 60m 0.06584 20 7.98 7.09 0.89 0m - 80m 0.12889 40 7.86 7.23 0.63 0 m - 100m 0.12622 60 7.75 7.37 0.38 0 m - 120m 0.13029 80 7.61 7.51 0.10 0m - 140m 0.14190 93 7.31 7.61 0.30 0 m - 160m 0.12672 100 8.00 7.66 0.34 0 m - 180m 0.11831 106 8.01 7.71 0.30 1205 21.51 22.78 1.27 0 m - 1205m 0.10642 Once the entire reach length is used to calculate the variance of depth deviations, the results can be plotted (Figure 9). Variance Plot of Jason Lower (Om - 1205m) 0.4 | « c _o | 0.3 -Q •S & 0.2 -Q 1400 Horizontal Distance (m) Figure 9 Plot of Horizontal Distance vs. Variance of Depth Deviation -Jason Lower As can be seen from Figure 9, the variance of depth deviation generally decreases with longer reach lengths, similar to the results of Wood et al., (1988). The upstream end of the Jason Lower variance plot shows an abrupt rise at 875m. This is likely due to the presence of several large LWD jams immediately downstream of this point. The calculations outlined in Table 22 and illustrated in Figure 9 were carried out for all old-growth and managed reaches (see Appendix B). 3.4.3 Selection of Old-Growth Stream Channel Sub-Reaches With the exception of Jason Lower, plots for all the uncoupled old-growth reaches generally stabilized at variance values of -0.05. With the exceptions of Inskip NB, Inskip SB and Government NB NF, plots for coupled reaches also stabilized at variance values of -0.05. As Inskip NB exhibited high variance values (-0.4), behaved erratically and was only 18 Wb in length it was discarded from the database. 49 Reaches were then classified into two general groups: (1) reaches less than 50 Wb in length, and (2) reaches greater than 50 Wb in length. Reaches Less Than 50 W b in Length Reaches under 50 Wb in length whose variance plots were stable were considered representative and suitable for this project. The only reach in this group which did not conform to this general behavior was Inskip SB. However, looking at the variance plot for Inskip SB it is apparent that sub-sections within the reach do stabilize to a degree. Based on this visual examination a sub-reach section was delimited and the calculations outlined in Table 22 were again carried out (Figure 10): .2 0.4 •3 CL u Q o 0.3 0.2 0.1 0.0 Entire Reach Inskip SB: 649m- 1399m 600 750 900 1050 1200 1350 1500 Horizontal Distance (m) 0.4 0.3 • 0.2 0.1 0.0 Selected Sub-Reach Inskip SB: 902m - 1399m 850 950 1050 1150 1250 1350 1450 Horizontal Distance (m) Figure 10 Variance Plots: Entire Reach Length and Selected Sub-Reach Length for Inskip SB Removing the downstream portion of Inskip SB did improve the behavior of the reach: The sub-reach plot for Inskip SB stabilized at variance values of -0.14. This sub-reach represents a more homogenous reach within Inskip SB. 50 It can be argued that this sub-reach analysis could be used for those reaches in which variance plots were initially stable (Government Main, Government NB, Government NB EF, Inskip Main, Gregory NB, Jason Upper, and Carmanah Upper); variance plots for sub-reaches could possibly stabilize at lower variance values. However, these reaches are less than 50 Wb in length and are already considered representative. The 50 Wb length criterion was arbitrarily chosen as the minimum reach length required for further analysis of this kind. Reaches Greater Than 50 W b in Length A similar analysis was undertaken for those reaches greater than 50 Wb in length (Government NB NF, Government U M , Gregory U M and Jason Lower). All reaches in this group, apart from Government NB NF, stabilized. However it can be argued that by further analyzing all of these reaches more representative sub-reaches could be delimited. The four reaches in question were systematically split into sections approximately 50 Wb in length (Table 23). These sections were delimited by moving a 50 Wb sample length upstream at an interval of approximately 10 Wb. Once again, the calculations outlined in Table 22 and illustrated in Figure 9 were carried out and the results were plotted up. Those sections in which variance plots showed no improvement or worsened were then discarded. The sections in which plots appeared to improve were then fine-tuned based on visual examination. 51 Table 23 Old-Growth Sub-Reach Sections to be Analyzed Reach Sections T o B e Ana l yzed Resul t Government N B N F 1 5 9 3 m - 2 1 4 3 m Further analysis required 1 7 2 3 m - 2 2 4 3 m Discard 1 8 2 3 m - 2 3 4 3 m Discard 1 9 2 3 m - 2 4 6 4 m Discard Government U M 8 0 0 m - 1740m Further analysis required 1 0 0 0 m - 1 9 2 5 m Further analysis required 1 2 2 5 m - 2 1 4 0 m Discard 1 4 0 5 m - 2 3 4 0 m Discard 1 6 0 0 m - 2 5 2 0 m Discard 1 7 8 0 m - 2 6 6 0 m Discard Gregory U M 3 7 0 1 m - 4 8 8 6 m Discard 4 0 1 1 m - 5 0 8 6 m Discard 4222m - 5306m Discard 4446m - 5526m Discard 4664m - 5746m Further analysis required 4886m - 5866m Further analysis required Jason Lower 0m - 920m Further analysis required 200m - 1205m Discard The first section of Government NB NF (1593m - 2143m) was selected for further analysis. In this section, an abrupt rise in variance values is apparent in the upstream portion of the variance plot (Figure 11). This upstream portion was removed, resulting in a final sub-reach (1593m - 1963m) in which variance values stabilized at ~0.05. Government NB NF: 1593m-2143m — Variance Plot — Upstream Limit y — ; • • 'l550 1650 1750 1850 1950 2050 2150 2250 Horizontal Distance (m) Figure 11 Variance Plot for Government NB NF 52 The first two sections of Government U M (800m - 1740m; 1000m - 1925m) were selected for further analysis (Figure 12). These two sections were combined to form a final sub-reach (800m - 1925m) in which variance values stabilized at ~0.05 (Figure 13). Government Upper Main: 800m - 1740m 0.4 r 0.3 \ I Q f o . 2 r ID 0.1| '§ o.o 1 — |> 700 900 1100 1300 1500 1700 1900 Horizontal Distance (m) Government Upper Main: 1000m - 1925m 1 0.4 0.3 0.2 0.1 E 0.0 > 900 1100 1300 1500 1700 1900 2100 Horizontal Distance (m) Figure 12 Variance Plots for Government U M : sections 800m - 1740m and 1000m - 1925m 0.4 i > 0.3 Q •3 & 0.2 Q o 2 0.1 S3 > 0.0 1 600 Selected Sub-Reach Government Upper Main: 800m - 1925m 800 1000 1200 1400 1600 1800 2000 2200 Horizontal Distance (m) Figure 13 Variance Plot: Selected Sub-Reach Length for Government U M 53 The final two sections of Gregory U M (4664m - 5746m; 4886m - 5866m) were selected for further analysis (Figure 14). The downstream portion of the 4664m -5746m variance plot showed an abrupt rise in variance values. The final sub-reach (4786m - 5866m) excluded this downstream portion, resulting in variance values stabilizing at-0.1 (Figure 15). Gregory Upper Main: 4664m - 5746m g 0,41 , . . . • S > 4600 4900 5200 5500 5800 Horizontal Distance (m) Gregory Upper Main: 4886m - 5866m S 0.4, . CO I 0.3 Q s g o.o L — — 1 ' > 4900 5100 5300 5500 5700 5900 6100 Horizontal Distance (m) Figure 14 Variance Plots for Gregory U M : Sections 4664m - 5746m and 4886m - 5866m 0.4 Q •B CL 0.3 • 0.2 • a o.i o.o Selected Sub-Reach Gregory Upper Main: 4786m - 5866m 4600 4800 5000 5200 5400 5600 5800 6000 Horizontal Distance (m) Figure 15 Variance Plot: Selected Sub-Reach Length for Gregory U M 54 The Om - 920m section of Jason Lower was selected for further analysis (Figure 16). From this section a final sub-reach (0m - 860m) was delimited. Removal of the upstream portion of this section resulted in variance values stabilizing at ~0.06. Jason Lower: 0m - 920m —<—« • i • .—•—>—i—i Variance Plot — Upstream Limit 0 200 400 600 800 1000 Horizontal Distance (m) Figure 16 Variance Plot for Jason Lower 3.4.4 Selection of Managed Stream Channel Sub-Reaches As with the old-growth stream channels, representative reach lengths need to be delineated for the managed stream channels. Unlike the old growth stream channels, the variance plots for the managed reaches did not stabilize at a consistent variance value. However, with the exception of Riley Middle, they did stabilize. Reaches were classified into two general groups: (1) reaches less than 50 Wb in length, and (2) reaches greater than 50 Wb in length. 55 Reaches Less Than 50 W b in Length Only two of the managed reaches were under 50 Wb in length. Mosquito Upper, whose variance plot was stable, was considered representative and suitable for this project. In contrast, the variance plot for Riley Middle behaved erratically. As only a 9 Wb sub-section within the reach appeared to stabilize, it was discarded from the database. Reaches Greater Than 50 W b in Length All reaches in this group, apart from Peel, stabilized. However, by further analyzing all of these reaches more representative sub-reaches may be delimited. The four reaches in question were split into sections approximately 50 Wb in length (Table 24). Unlike the old growth streams, these sections were delimited based on visual examination of the original variance plots. This method was used in order to minimize unnecessary calculations. (Several of the managed reaches were approximately 200 Wb in length, and the differences between variance plots shifted in 10 Wb intervals were judged to be insignificant.) The calculations outlined in Table 22 and illustrated in Figure 9 were carried out for these sub-reach sections and the results were plotted. Those sections in which variance plots showed no improvement or worsened were then discarded. The sections in which plots appeared to improve were then fine-tuned based on visual examination. The first section of Mosquito Main (0m - 1500m) stabilized at ~0.1. As no anomalies were apparent in the variance plot, this section was selected as the final sub-reach. 56 Table 24 Managed Sub-Reach Sections to be Analyzed Reach Sections T o Be Ana l yzed Resul t Mosqui to M a i n Om - 1500m 5 0 0 m - 2 0 2 2 m Further analysis required Discard Ri ley Lower 0 m - 1275m 2 5 0 m - 1525m 5 0 0 m - 1775m 1 5 2 5 m - 2 8 0 0 m 2 0 0 0 m - 3 2 7 5 m 2 5 5 0 m - 3 8 2 5 m 3 8 2 5 m - 5 1 0 0 m 4000m - 5325m Further analysis required Further analysis required Further analysis required Further analysis required Discard Discard Discard Discard Tarundl 0m - 825m 4 0 5 m - 1410m 8 2 5 m - 1640m 1 6 4 0 m - 2 4 6 0 m Further analysis required Further analysis required Discard Discard Peel 0m - 800m Further analysis required The first four sections of Riley Lower (0m - 1275m; 250m - 1525m; 500m -1775m; 1525m - 2800m) were selected for further analysis (Figure 17). While all four variance plots stabilized, section 1525m - 2800m stabilized at the lowest variance value (-0.06). This reach was selected as a final sub-reach. However, as the first three sections all stabilized (at -0.1) and covered a significant length of reach not included in section 1525m - 2800m, a second representative reach was delineated from Riley Lower. The first two sections were combined to form a second final sub-reach (0m -1525m) in which variance values stabilized at -0.09. Section 0m - 1525m was renamed Riley Lower 1, and section 1525m - 2800m was renamed Riley Lower 2 (Figure 18). 57 V3 C 0.4 o > 0.3 Q Q . u 0.2 O 'o i> 0.1 § 0.0 > 0.0 Riley Lower: 0m - 1275m 0 200 400 600 800 1000 1200 1400 Horizontal Distance (m) Riley Lower: 500m - 1775m 400 600 800 1000 1200 1400 1600 1800 2000 Horizontal Distance (m) Riley Lower: 250m - 1525m 0.4 0.3 0.2 0.1 0.0 200 400 600 800 1000 1200 1400 1600 1800 Horizontal Distance (m) Riley Lower: 1525m-2800m 0.4 0.3 0.2 0.1 J 0.0 > 1400 1600 1800 2000 2200 2400 2600 2800 3000 Horizontal Distance (m) Figure 17 Variance Plots for Riley Lower: sections 0m - 1275m, 250m -1525m, 500m - 1775m, and 1525m - 2800m. § 0.4 I 0.3 ! • 0.2 1 0.0 | 0 Selected Sub-Reach Riley Lower 1: 0m - 1525m 400 800 1200 Horizontal Distance (m) 1600 0.4 « 0 . 3 •3 8-0.2 Q o 0.1 0.0 Selected Sub-Reach Riley Lower 2: 1525m-2800m 1400 1600 1800 2000 2200 2400 2600 2800 3000 Horizontal Distance (m) Figure 18 Final Variance Plots for Riley Lower 1 (0m - 1525m), and Riley Lower 2 (1525m - 2800m). 58 The first two sections of Tarundl (Om - 825m; 405m - 1410m) were selected for further analysis (Figure 19). While both sections stabilized, section 405m - 1410m stabilized at the lowest variance value (-0.2) and was selected as the final sub-reach. Tarundl: 0m - 825m 0.4 200 400 600 Horizontal Distance (m) 800 Tarundl: 405m - 1410m 0.4 0.3 0.2 0.1 0.0 '300 500 700 900 1100 1300 1500 Horizontal Distance (m) Figure 19 Variance Plots for Tarundl (0m - 825m, and 405m - 1410m). The original variance plot for the entire Peel reach showed a rise in the upstream portion of the reach. Once this portion was removed, the reach length for Peel was approximately 50 Wb in length. As the variance plot for this sub-reach stabilized at -0.055, it was selected as the final sub-reach for Peel (Figure 20). 3 0.4 o 'S 0 3 Q % 0.2 S o.i I 0.0 Selected Sub-Reach Peel: 0m-800m 200 400 600 Horizontal Distance (m) 800 Figure 20 Final Variance Plot for Peel (0m - 800m). 59 3.5 Summary Representative reach lengths were determined based on depth characteristics. The behavior of the variance of depth deviations was examined over increasingly longer reach lengths. Once the variance values stabilized the reach length was deemed representative. This approach was based on the Representative Elementary Area concept introduced by Wood et al., (1988). Tables 25 and 26 summarize the selected sub-reaches that will be used in the following chapters to quantify stream channel variability. Table 25 Old-Growth Sub-Reach Selection Watershed Reach In i t ia l Reach Selected Sub- Sub-Reach F i n a l Reach Leng th ( W b ) V a r i a n c e Government Creek Government M a i n Om - 800m 0m-800m 25 0.05 Government U M 800m - 2660m 800m-1925m 60 0.05 Government N B 8 0 0 m - 1593m 800m-1593m 40 0.06 Government N B N F 1 5 9 3 m - 2 4 6 4 m 1593m-1963m 34 0.05 Government N B E F 1 5 9 3 m - 2 2 6 5 m 1593m-2265m 40 0.06 Inskip Creek Inskip M a i n 0m - 649m 0-649m 29 0.05 Inskip N B 6 4 9 m - 9 1 9 m Discard - only 18 W b long & irregular Inskip S B 6 4 9 m - 1399m 902m-1399m 34 0.14 Gregory Creek Gregory U M 3 7 0 1 m - 5 8 6 6 m 4786m-5866m 50 0.11 Gregory N B 5866m - 6286m 5866-6286m 28 0.03 Jason Creek Jason Lower 0 m - 1205m 0m-860m 46 0.07 Jason Upper 1 2 0 5 m - 1925m 1205m-1925m 33 0.05 Carmanah Creek Carmanah Upper 0 m - 1000m 0 m - 1000m 26 0.07 Table 26 Managed Sub-Reach Selection Watershed Reach In i t ia l Reach Selected Sub -Reach Sub -Reach Length ( W „ ) F i n a l V a r i a n c e Mosqui to Creek Mosqui to M a i n 0m - 2022m 0 m - 1500m 51 0.1 Mosqui to Upper 2 0 2 2 m - 2 9 6 5 m 2 0 2 2 m - 2 9 6 5 m 29 0.04 Ri ley Creek R i ley M idd le 5325m - 6840m Discard - d id not stabilize R i ley Lower 1 0m - 5325m 0 m - 1525m 50 0.1 Ri ley Lower 2 0m - 5325m 1 5 2 5 m - 2 8 0 0 m 50 0.06 Tarundl Creek Tarundl 0 m - 2 9 1 0 m 4 0 5 m - 1 4 1 0 m 62 0.2 Peel Creek Peel 0 m - 1200m 0m - 800m 46 0.055 60 Chapter 4: A Method For Stream Channel Comparison 4.1 Introduction Attempts to compare forest stream channels commonly focus on: (1) channel unit characteristics (the proportions, spacing, slope and shape of channel units); and (2) changes in large woody debris (Hogan, 1986 & 1989; Andrus et al., 1988; Grant et al., 1990; Ralph et al., 1994; Wood-Smith and Buffington, 1996). While these techniques provide useful information, they do not offer a general method of quantitatively comparing stream channels based on a variety of stream channel characteristics. More comprehensive methods of stream channel comparisons do exist: In British Columbia, the current method used to assess stream channels is known as the Channel Assessment Procedure (CAP). The intent of the CAP is to identify disturbed channels (if they exist) in a "consistent and repeatable" process (Hogan et al., 1996). Wood-Smith and Buffington (1996) used discriminant function analysis to develop an objective geomorphic discrimination of pristine and disturbed channel conditions. While their method does encompass a variety of stream channel characteristics, it has the specific purpose of identifying or discriminating between old-growth and disturbed channels. As stated previously, such a purpose is not the sole aim of this project. And, indeed, it is arguable that the notion of discriminating or identifying disturbed channels is not well founded without knowledge of the range of variability that exists in undisturbed channels. 61 The objective method developed by Cheong, (1992) for basin comparison was adapted here for use in stream channels. This method is based on the concept of dissimilarity. 4.2 Dissimilarity One way to compare stream channel reaches is by calculating the dissimilarity1 of two reaches based on key stream channel characteristics (as selected in section 3.2). If this dissimilarity can be calculated for a large enough data set, a frequency distribution of the dissimilarity values can be constructed. The frequency distribution would express the range of variability present in the streams analyzed. Several methods may be employed to analyze the dissimilarity between two objects. Most procedures incorporate some form of Euclidean distance measure in order to calculate the 'proximity' between two objects [Gordon, 1981]. For example, Ewklxjk-Xjkl' E w k Eq. 1 where Wk (k=l...p) is a set of weights, i represents the first object, j represents the second object, and k is the k* characteristic (equivalent to weighted root mean squared statistic W.R.M.S.). From this a general dissimilarity index may be achieved, D..W= Zwklxjk-Xjkl' Sw k 1/A Eq.2 1 Stream channel similarity can also be used to compare stream channels. However, as the purpose o f this project is to determine the range o f variabil i ty that exists in stream channels, the interest is focused primari ly on how different, or dissimilar, various reaches are. Therefore dissimilarity is the preferable measure which w i l l be explored in this project. 62 where X > 0 and higher values of X give relatively more emphasis to the larger differences |xik-Xjk| (Gordon, 1981; Cheong, 1992). 4.3 Cheong's Dissimilarity Testing Procedure Several difficulties are associated with dissimilarity indices. These difficulties are related to the characteristics, or variables, used in the calculations. For example, incompatible units between variables may pose a problem. This can be overcome by standardizing the variables. This is achieved by dividing each variable by its standard deviation or range. This standardization technique can also be used for variables with a relatively large range of variation. (A variable with a large range of variation may skew calculations.) Difficulties arise when variables exist at different levels of mensuration (e.g. ordinal, interval, ratio). Gordon (1981) suggests three possible approaches to this problem: (1) convert all variables to the most common data type; (2) employ a general similarity coefficient which can incorporate information from different data types; or (3) carry out separate analyses based on data type. Cheong (1992) developed a procedure to overcome this problem. This procedure can also be utilized for stream channel comparison, and is adopted here: 1. Conduct a general test for interval type data. This test, similar to equation 1, calculates a dissimilarity distance between reaches i andj: dijk 2 = (Xik-x i k ) 2 Eq. 3 (0 .25*R) 63 Here k represents the k characteristic, R = range and X = 2 so that greater differences are reflected in the measure. The information is standardized in order to eliminate any effects of the unit or range of measurement. In this test one quarter of the range is used as a surrogate for the standard deviation. 2. The second distance calculation is for ratio information and is similar to the first test with one exception: The distance is standardized by using the standard deviation. The dissimilarity measure is achieved by combining these two tests. The total difference between the two stream reaches is represented by the square root of the sums over all characteristics of the interval and ratio type tests. However, the Cheong method of standardization is questionable. While the numerator in Equation 3 is squared, the denominator is not. In other words, each variable is standardized by dividing by the square root of the standard deviation. This creates greater emphasis on those characteristics with larger values. A proper standardization method would involve separately dividing each variable by the standard deviation: dijk2 Xjk Oi Xjk a. Eq.4 64 Which reduces to: dijk2 = (Xik-x i k) 2 Eq. 5 4.4 Stream Channel Dissimilarity Testing Procedure Table 27 lists the stream channel characteristics to be used in this project. For each characteristic the dissimilarity equation requires a single, or summary, statistic. The majority of the variables can easily be expressed by a single value. The methodology used to calculate values for these characteristics is outlined in Appendix A. Table 27 Stream Channel Characteristics - Units, Information Type Paramete r Un i ts In format ion Type Channel Uni t Frequency # o f specific units / # o f al l units Ratio Channel Uni t Length m / m Ratio Poo l Spacing m / m Ratio Depth Variabi l i ty m / m Ratio Width Variabi l i ty m / m Ratio L W D Spacing m / m Ratio L W D Vo lume nvVm3 Ratio Relative Roughness m / m Ratio The stream channel dissimilarity testing procedure involves two steps. The first step is to calculate the dissimilarity of each stream channel characteristic for any given reach pair combination. For example, the riffle frequency dissimilarity between Government Main and Government Upper Main is calculated as: difrifflrec, = (Xj - X j ) 2 Eq. 6 (Orfffle) Where i represents Government Main, j represents Government Upper Main, x represents the riffle frequency value for a given reach, and ariffle is the standard deviation of all the riffle frequencies in the sample (i.e., all 12 old-growth reaches). 65 At this point total dissimilarity can be calculated. For a given reach pair combination, the total dissimilarity is the square root of the sum of all the dissimilarity values for each stream channel characteristic: 2 2 2 2 V2 dij-toto/ = (dij riffreq^ dij rifflength+ dy poolfreq ••• + dy D/d) ' Eq. 7 As can be seen from Table 27 and Appendix A, all of the selected stream channel characteristics are scale free. This places the focus on the intensive measures of the system. As scale free variates are dimensionless, it can be argued that the measures need not be standardized (as illustrated in Equations 5 and 6). In order to investigate this question, the dissimilarity testing procedure was carried out for two cases: (1) no measures standardized, and (2) all measures standardized. With no standardization, LWD spacing emerges as the most influential variate in determining dissimilarity between reach pairs. In other words, for any given reach pair combination, the value of dy2:, wDspadng was significantly greater than the dissimilarity value of any other stream channel characteristic. This is simply a reflection of the input data range. While the majority of the stream channel characteristics fall in the range of 0 to 1, LWD spacing values (scaled as multiples of Wb) fall in the range of 1 to 10. In other words, the dominance of the LWD spacing variate is simply an artifact of the calculation method, which places greater emphasis on variables having higher values. When all variates are standardized, no single variable dominates and the bias is removed. As the purpose of the dissimilarity testing procedure is to determine how different any two stream channels are, the removal of scale from the calculations may not be 66 appropriate. It is possible that a scale-referenced comparison may be more effective. Two scaled variates to consider adding to the stream channel characteristics listed in Table 27 are (1) contributing drainage area, and (2) mean bankfull width. Mean bankfull width relates directly to the physical scale of the system in question. In contrast, contributing drainage area would better reflect the general hydrology of the system. For the purposes of this study, mean bankfull width is the most appropriate variate. The question arises then whether mean bankfull width should be standardized. In order to explore this question the dissimilarity testing procedure was carried out for two cases: (1) mean bankfull width not standardized, and (2) mean bankfull width standardized. For the case when mean bankfull width is not standardized, it is by far the most influential variate in determining dissimilarity between reach pairs. The values for dy wb are so great that all other values are inconsequential in the calculations. As explained above, this is related to the fact that the calculation method places greater emphasis on variables having higher values. In contrast, when mean bankfull width is standardized (along with all the other variates) it is only occasionally the most influential measure. While standardizing mean bankfull width does diminish the importance of scale in the calculations, it is judged to be the best way to incorporate the scale-referenced characteristic into the calculations. The use of weights (as shown in Equation 2), could ultimately be employed to increase the effect of mean bankfull width on the dissimilarity calculations. However, it is beyond the scope of this present study to determine appropriate weights for stream channel characteristics, which presumably would depend on the particular purposes of an individual problem. 67 4.5 Summary There is a lack of quantitative techniques available for stream channel comparison. A method adapted from Cheong (1992), based on analysis of dissimilarity, offers a comprehensive and objective way to quantitatively compare stream channels. Table 28 lists the final selection of stream channel characteristics to be used in the stream channel dissimilarity testing procedure. Table 28 Selected Stream Channel Characteristics Paramete r Uni ts Channel Uni t Frequency # o f specific units / # o f al l units Channel Un i t Length m / m Poo l Spacing m / m Depth Variabi l i ty m / m Width Variabi l i ty m / m L W D Spacing m / m L W D Vo lume m / m Relative Roughness m / m Mean Bankfu l l Width M 68 Chapter 5; Results and Discussion 5.1 Introduction Having selected suitable stream channel reaches in Chapter Three, the relevant stream channel characteristics required for each reach (as outlined in Table 28 and Appendix A) can now be extracted from the channel surveys. These summary stream channel data are used to calculate dissimilarity values for various reach pair combinations. Five general types of reach pair combinations are constructed: • Old-growth vs. Old-growth (channel configuration ignored) • Old-growth (uncoupled) vs. Old-growth (uncoupled) • Old-growth (coupled) vs. Old-growth (coupled) • Managed (uncoupled) vs. Managed (uncoupled) • Managed (uncoupled) vs. Old-growth (uncoupled) As no coupled, managed reaches exist in the database, only the uncoupled, old-growth reaches are included in the 'Managed vs. Old-growth' reach pair combination type. In order to build a basis for comparison, the old-growth reach pair combinations will be studied first. 5.2 Old-Growth Stream Channels Table 29 summarizes the relevant stream channel characteristics for each reach. Several problems arose during this phase of data extraction which were primarily related to the morphological classification used in the FFIP surveys. Of the twelve old-growth reaches, only Gregory Upper Main had step-pool features delineated. As discussed in section 3.2.1, cascades can be viewed as a compound unit made up of successive steps 69 and pools. Considering that the step-pool features in Gregory Upper Main accounted for only 1% of the total length of the reach, they were merged with the cascade features. Only two reaches (Inskip SB and Gov NB NF) have stone line features defined, and they account for only a small percentage of the total reach lengths. It could be argued that these features be treated like the step-pool features in Gregory Upper Main and merged with the cascade data. However, as stone line features were delineated in more than one reach, this feature was not removed from the matrix table. In addition, the stone line data do help to characterize the reaches to a small degree. While the dissimilarity values for reach pairs which do not have stone line features are not affected by their absence, the dissimilarity values for reach pairs involving Inskip SB and/or Gov NB NF are slightly higher. The most problematic old-growth reach is Gregory NB. Half of the reach length is bedrock, and no morphological data were recorded in the bedrock channel section. This results in low frequency percentages for the channel units and channel unit lengths. In addition, no LWD volume data or D95 sediment data exist for this reach. The LWD spacing data are also questionable as only two LWD jams are found in the reach. Therefore the LWD spacing value is based on one distance value, not an average like the other reaches which have multiple LWD jams. 70 (-1 .S ' St 1^3 • Q > oo 1(2 g . : oo % s1 3 i_ CD L L , •ti al c CD p £ 3 kH •« al a CD p £ 3 i-CD L t , - J P £ CD O 3 t -—1 "O 60 gl , p S t , IP. « PH I—I 8 c 81 1,2 P £ CD 43 _ i a gi 3 L -CD L L . —1 CD •P2^ u es cu CQ Z o . „ fD cn O i—i PQ oo PH sis olo VO CN T3 C CD OH <; CD _3 > 3 _o o o c CD CD L« PH 00 s CD p "33 s s C0 u Ii o 3 CD 3 CT CD L . PH -*-» 'S p 13 3 3 co - 3 o I CtH o O •a CD 43 60 3 '% O CD 43 co CO 43 pa Z & o oo CD LH o o Z CD X t o U » 'S p T3 3 43 u vo 2? 43 a S PH _1 OH PH CD CD CT CD 00 8 _ d> PH - J CD CD C 5H OO 00 2 '(2 4^ 4 * 4*i 44 u u u u 5 3 3 5 0H OH PH PH 4= CO 4= u CD 43 p 3 on § CD CfH CD -a CO o CO CO o 00 00 § 4df — 1 CO co c! s CH «* *i U~i PH &• s <*H 00 M co U CO 2 ^ •9 .2 CD ta -p *a Z >> o .S a o CD 1 i .2 « to & 3 ^ c r » CO 3 CO 3 _2 "i3 "JH P CO co PH CO "c0 co .3 s • "° Jp-3 " ^ cd O H " CD .5 P H CD M 00 CO _ 00 « S P CO 2 HH O OO 00 u -J HH 00 5.2.1 All Reach Pair Combinations Table 30 lists the summary dissimilarity results using all possible reach pair combinations: Table 30 Dissimilarity Results - All Possible Reach Pair Combinations Reach Pa i r s D iss imi la r i t y Reach Pa i r s D iss im i la r i t y G o v M a i n - G o v Upper Ma in 4.49 Jason Upper - Inskip S B 7.25 G o v M a i n - Jason Lower 3.89 Jason Upper - Greg Upper M a i n 4.51 G o v M a i n - Jason Upper 4.40 Jason Upper - G o v N B E F 4.05 G o v M a i n - Carmanah 4.47 Jason Upper - Gov N B N F 6.33 G o v M a i n - Greg N B 6.08 Jason Upper - Gov N B 3.78 G o v Ma in - Inskip M a i n 4.24 Carmanah - Greg N B 6.65 G o v M a i n - Inskip S B 10.15 Carmanah - Inskip M a i n 6.33 G o v M a i n - Greg Upper M a i n 4.12 Carmanah - Inskip S B 10.92 Gov Ma in - Gov N B E F 6.15 Carmanah - Greg Upper M a i n 5.57 Gov Ma in - G o v N B N F 7.23 Carmanah - G o v N B E F 7.93 G o v M a i n - G o v N B 4.81 Carmanah - G o v N B N F 8.90 G o v Upper - Jason Lower 3.50 Carmanah - G o v N B 7.05 G o v Upper - Jason Upper 2.84 Greg N B - Inskip M a i n 6.44 G o v Upper - Carmanah 5.62 Greg N B - Inskip S B 10.17 G o v Upper - Greg N B 6.39 Greg N B - Greg Upper Ma in 4.88 G o v Upper - Inskip M a i n 4.32 Greg N B - Gov N B E F 6.45 G o v Upper - Inskip S B 8.19 Greg N B - G o v N B N F 7.13 G o v Upper - Greg Upper M a i n 4.33 Greg N B - G o v N B 6.43 G o v U p p e r - G o v N B E F 4.02 Inskip M a i n - Inskip S B 8.02 G o v Upper - G o v N B N F 6.30 Inskip Ma in - Greg Upper M a i n 4.40 G o v Upper - G o v N B 2.73 Inskip Ma in - G o v N B E F 4.73 Jason Lower - Jason Upper 5.16 Inskip Ma in - Gov N B N F 5.90 Jason Lower - Carmanah 6.37 Inskip Ma in - Gov N B 4.35 Jason Lower - Greg N B 6.55 Inskip S B - Greg Upper M a i n 9.25 Jason Lower - Inskip M a i n 5.18 Inskip S B - G o v N B E F 6.96 Jason Lower - Inskip S B 10.22 Inskip S B - G o v N B N F 8.32 Jason Lower - Greg Upper M a i n 3.54 Inskip S B - G o v N B 8.56 Jason Lower - G o v N B E F 4.89 Greg Upper M a i n - G o v N B E F 4.64 Jason Lower - G o v N B N F 6.14 Greg Upper M a i n - G o v N B N F 5.50 Jason Lower - Gov N B 3.20 Greg Upper Ma in - G o v N B 4.10 Jason Upper - Carmanah 5.36 G o v N B E F - Gov N B N F 5.29 Jason Upper - Greg N B 5.96 Gov N B E F - G o v N B 2.74 Jason Upper - Inskip M a i n 3.26 Gov N B N F - G o v N B 5.18 There are no notable outliers in this table. The least similar reach pairs are Carmanah - Inskip SB and Jason Lower - Inskip SB: 72 Carmanah - Inskip SB The high total dissimilarity value for the Carmanah - Inskip SB reach pair can be attributed to the high dissimilarity values of three stream channel characteristics: LWD volume (d2LWDvoi = 11.65), mean bankfull width (d2Wb = 9.51), and relative roughness (d2D/rf=9.11). The high value for d2LWDvoi is a reflection of the LWD volume values for Inskip SB and Carmanah (see Table 29). Inskip SB has the highest LWD volume in the old-growth database while Carmanah has the lowest. This is likely related to scale, as the d2wb value is also high. Carmanah, at Wb = 38.9 m, is an uncoupled reach that is too wide for logs to fully span the channel. While wood is still an effective element in the stream channel reach, large channel spanning logjams (which represent a large volume of wood) are not present. In contrast, Inskip SB is a smaller, coupled reach (Wb = 14.6 m) with multiple channel spanning log jams. The high value for d2o/d is also likely related to scale. Carmanah has the lowest relative roughness (Did = 0.63) in the old-growth database while Inskip SB has the highest (D/d = 2.53). This is expected, as in smaller channels the ratio, D/d, is usually greater than one. In intermediate channels, relative roughness usually falls in the range 1.0>D/d>0.1 (Church, 1992). Carmanah, an intermediate-sized channel, is situated in the transport zone of the drainage basin. The sediment in this zone has already received some organization (through prior alluvial transport) and flows are therefore competent. In contrast, Inskip SB is a smaller, coupled reach which experiences active mass-wasting 73 and therefore receives sediment directly from the hillslopes. It is situated closer to the headward zone of the drainage basin, which implies the presence of material larger than the usual ability of a stream to transport. The Carmanah - Inskip SB reach pair is a good example of differences that arise across extreme variations in scale. Inskip SB is among the smaller basins in the comparison (Ad = 5.0 km2), whereas Carmanah is the largest. So, processes mediated by position in the basin have their greatest effect in comparisons such as this. A critical operational comparison would probably endeavor to hold scale variations well within an order of magnitude. High dissimilarity values were also noted for two other stream channel characteristics: log step length (d2isien = 12.15) and cascade length (d 2 c t e „ = 11.03). This is related to the fact that Inskip SB has the largest values for both log step length frequency ( LS length = 0.04) and cascade length frequency (C length = 0.2). As Inskip SB is a steep, small channel with relatively large amounts of wood present, the higher frequency of log steps is not surprising. However, the high cascade length value for Inskip SB is significantly higher than all other reaches in the old-growth database. This is not wholly unexpected, as Inskip SB is one of the steepest reaches in the old-growth database (see Table 5). In addition, this could be related to the presence of a lake in the headwaters region (see Figure 1), which could be regulating stream flow to some degree. Jason Lower - Inskip SB The high total dissimilarity value for the Jason Lower - Inskip SB reach pair can be attributed to the high dissimilarity values of three stream channel characteristics: riffle 74 unit frequency (d riff= 12.83), width variability (d w v a r = 11.18), and cascade length frequency (d ckn = 11.03). The high values for d2riff and d 2 c/ e„ reflect the fact that Inskip SB has the highest cascade length frequency value and the lowest riffle unit frequency value (see Table 29). In contrast, the cascade length frequency value for Jason Lower is zero, and the riffle unit frequency value is relatively high. The issues of scale (found in the Carmanah -Inskip SB case) do not exist here as both contributing area and bankfull width values are comparable between Jason Lower and Inskip SB. However, similar to the Carmanah -Inskip SB case, the channel configuration is different. Jason Lower is an uncoupled, relatively flat reach, while Inskip SB is both coupled and relatively steep. This is reflected in the channel unit characteristics. Cascades are associated with steep channels such as Inskip SB while riffles are found in flatter reaches such as Jason Lower (which has a relatively high riffle unit frequency value). The high value for d2wvar is related to the width variability values for Jason Lower and Inskip SB. Jason Lower has a relatively low width variability value (0.148) while Inskip SB has the highest (0.692). The high width variability calculated for Inskip SB can be attributed to one unusually high W b value (Wb = 39.2). This high Wb value is recorded at the site of a relatively large channel spanning logjam. Other reach pairs with significantly high dissimilarities (within the top 10% least similar) include: Gregory NB - Inskip SB, Government Main - Inskip SB, Inskip SB -Gregory Upper, and Carmanah - Gov NB NF. Inskip SB is found in three of these four reach pair combinations. Once again the high dissimilarity values are largely related to 75 Inskip SB's high cascade length frequency values, high LWD volume values, and low riffle frequency values. Overall, Inskip SB emphasizes the importance of topography and processes in mediating comparisons. It also stresses the importance of coupling as a filter factor. The high total dissimilarity value between Carmanah and Gov NB NF can be attributed to the high dissimilarity value of the mean bankfull width characteristic (d wb = 12.82). This is related to scale. Carmanah has the highest mean Wb value and contributing area, while Gov NB NF has the lowest mean Wb value and contributing area. The most similar reach pairs are Gov Upper - Gov NB and Gov NB EF - Gov NB: Gov Upper -Gov NB The high degree of similarity between these two reaches is likely related to the fact that Gov Upper is located immediately adjacent to Gov NB (see Figure 1). Looking closely at the stream channel characteristics for both Gov Upper and Gov NB it is apparent that there are no significant differences between the two reaches. The highest dissimilarity value is for glide length (d gine„ = 1.93). However, this value is still relatively low. In addition, mean bankfull width values and contributing areas are remarkably similar for Gov Upper and Gov NB. Gov NB E F - Gov NB Again, the similarity between these two reaches is related to the fact that Gov NB EF is situated immediately upstream of Gov NB. These two reaches also have similar mean bankfull width values and contributing areas. 76 A question that occurs here is whether the average dissimilarity amongst reach pairs within the same drainage basin (e.g., Government Creek) is less than that amongst reach pairs between basins (e.g., one reach from Government Creek, one reach from a different drainage basin). An analysis of variance determined that no significant difference exists between these two average dissimilarity values [F c r„ (4.07) > F(1.47); accept H 0]. Other reach pairs with significantly low dissimilarities (within the top 10% most similar) include: Gov Upper - Jason Upper, Jason Lower - Gov NB, Jason Upper -Inskip Main, and Gov Upper - Jason Lower. These four reach pairs are not situated adjacent to each other as Gov Upper - Gov NB and Gov NB EF - Gov NB were. However, two of these four reach pairs have the same channel configuration, and all have similar contributing areas and mean bankfull width values. 5.2.2 Uncoupled Reach Pair Combinations As it appears that channel configuration may play an influential role when calculating dissimilarities, a closer examination of both the coupled and uncoupled reach pair combinations is necessary. This involves constructing a new general dissimilarity index, as values for standard deviations change with different sample sizes. The dissimilarity of each stream channel characteristic and the total dissimilarity were calculated for all uncoupled reach pair combinations. Table 31 lists the summary dissimilarity results using all possible uncoupled reach pair combinations: Table 31 Dissimilarity Results - Uncoupled Reach Pair Combinations Reach Pa i r s D iss imi la r i t y Reach Pa i r s D iss im i la r i t y Gov M a i n - G o v Upper Ma in 6.15 Gov Upper M a i n - Jason Upper 4.33 G o v M a i n - Jason Lower 4.98 G o v Upper Ma in - Carmanah 6.35 G o v M a i n - Jason Upper 6.46 Jason Lower - Jason Upper 6.55 G o v M a i n - Carmanah 5.35 Jason Lower - Carmanah 6.16 G o v Upper M a i n - Jason Lower 3.96 Jason Upper - Carmanah 7.17 77 Table 31 (uncoupled reach pair combinations) has a smaller range of dissimilarity values than Table 30 (all reach pair combinations). It is also worth noting that the dissimilarity values in Table 31 differ from those in Table 30. For example, Gov Main -Gov Upper Main has a total dissimilarity value of 4.49 in Table 30, but a total dissimilarity value of 6.15 in Table 31. This is related to the change in sample size, which alters the standard deviation values. If sufficiently large sample sizes were available, the standard deviation values would be stable and the changes in dissimilarity values would not likely occur. While the dissimilarity values are different, there are only slight changes in the overall ranking of the reach pairs. For example, Jason Upper - Carmanah has a higher dissimilarity value than Gov Upper - Jason Lower in both Table 30 and Table 31. The most dissimilar reach pair combination in Table 31 is Jason Upper - Carmanah, followed by Jason Lower - Jason Upper. The high total dissimilarity value for the Jason Upper - Carmanah reach pair can be attributed to the high dissimilarity values of three 2 2 stream channel characteristics: riffle length (d riflen = 6.81), relative roughness (d o/d~ 6.56), and cascade length (d 2 c/ e„ = 5.81). The high value of d2o/d reflects the fact that Carmanah has the lowest relative roughness value. This is related to its location within the watershed (in the transport zone) where the bed material has received some organization through prior alluvial transport and is relatively fine-grained. For Jason 2 2 2 Lower — Jason Upper, cascade (d cien•= 5.81) and pool (d pooiien = 5.59, d p0oifreq = 5.55) stream channel characteristics have high dissimilarity values. While Jason Upper has a relatively high cascade length frequency value (0.06), no cascades exist in Jason Lower. 78 This is related to the fact that Jason Upper has a steeper channel gradient than Jason Lower (steeper channel gradients are associated with cascade features). The most similar reach pair combination is Gov Upper - Jason Lower, followed by Gov Upper - Jason Upper. While there are not many substantial differences between the two most similar reach pairs, the relative roughness characteristic appears to be one of the key discriminators. The value for d2o/d is 0.09 for Gov Upper - Jason Lower. In contrast, d2/y</is 3.22 for Gov Upper - Jason Upper. Looking at Table 29 it is apparent that the relative roughness values for Gov Upper and Jason Lower are very similar (0.81 and 0.74, respectively), while the relative roughness for Jason Upper is larger (1.21). This could possibly be explained by the fact both Jason Lower and Gov Upper are located within the transport zones of their watersheds, where it is expected that the bed material size is generally finer and better sorted than further upstream (e.g., Jason Upper). 5.2.3 Coupled Reach Pair Combinations The same calculation procedures as explained in sections 4.4 and 5.2.2 were carried out for all coupled reach pair combinations. Table 32 lists the summary dissimilarity results: Similar to the results for uncoupled reach pair combinations, there is a smaller range of dissimilarity values in Table 32. In addition, dissimilarity values in Table 32 differ from those in Table 30 (all reach pair combinations). Again, only slight changes are apparent in the overall ranking of the reach pairs. 79 Table 32 Dissimilarity Results - Coupled Reach Pair Combinations Reach Pa i r s D iss imi la r i t y Reach Pa i r s D iss im i la r i t y Greg N B - Inskip M a i n 6.10 Inskip S B - Greg Upper M a i n 8.41 Greg N B - Inskip S B 8.72 Inskip S B - G o v N B E F 6.63 Greg N B - Greg Upper M a i n 4.99 Inskip S B - G o v N B N F 7.93 Greg N B - G o v N B E F 6.05 Inskip S B - G o v N B 8.15 Greg N B - Gov N B N F 6.70 Greg Upper M a i n - G o v N B E F 4.39 Greg N B - G o v N B 6.37 Greg Upper M a i n - G o v N B N F 5.43 Inskip M a i n - Inskip S B 7.05 Greg Upper M a i n - G o v N B 3.87 Inskip M a i n - Greg Upper M a i n 4.34 G o v N B E F - G o v N B N F 4.61 Inskip M a i n - G o v N B E F 4.90 G o v N B E F - G o v N B 2.75 Inskip M a i n - G o v N B N F 6.14 G o v N B N F - G o v N B 4.94 Inskip M a i n - G o v N B 4.74 The least similar reach pair combinations all involve Inskip SB. They include (in order of decreasing dissimilarity): Greg NB - Inskip SB, Inskip SB - Greg Upper, and Inskip SB - Gov NB. The high total dissimilarity values for these three reach pair combinations can be largely attributed to Inskip SB's high cascade length frequency values, high LWD volume values, and low riffle frequency values. It is curious that Greg NB shows up in the most dissimilar reach pair, as Greg NB has no data available on D/d or LWD volume. In other words, for any reach pair involving Greg NB, no values exist for d ^ o r &2LWDVOI- If the missing data for Greg NB were available, the total dissimilarity value for Greg NB - Inskip SB would be even greater. The most similar reach pair combinations include (in order of increasing dissimilarity): Gov NB EF - Gov NB, Greg Upper - Gov NB, and Inskip Main - Greg Upper. As stated previously, the similarity between Gov NB EF and Gov NB is related to the fact that these two reaches are situated adjacent to each other. While there are not many substantial differences between Greg Upper - Gov NB and Inskip Main - Greg Upper, the pool spacing characteristic appears to be the key discriminator. This is 80 related to the fact that Greg Upper has the greatest pool spacing value (2.52) while Gov NB and Inskip Main have small pool spacing values (0.92 and 1.25, respectively). This may be related to Greg Upper's low LWD volume and relatively flat gradient. 5.2.4 Selected Reach Pair Combinations After examining the initial dissimilarity results one problematic reach stands out: Inskip SB. Interestingly, the final drainage sub-basin selection process in Section 2.4 pinpointed this sub-basin as not meeting all the assessment criteria summarized in Table 6. The steep gradient of the reach, coupled with the moderate-sized lake in the headwaters, are judged to be acceptable reasons for discarding this reach from the database. Four other sub-basins were also selected in Section 2.4 for not meeting all the assessment criteria: Inskip Main, Carmanah, Gregory Upper Main and Gregory NB. The geologic structure of Gregory NB and Gregory Upper Main (specifically rock strength) was noted as being different to the rock strength of all the other sub-basins in the project database. In addition, Carmanah was highlighted due to it's extreme stream magnitude value. Inskip Main was selected due to it's relatively high lake area. With the possible exception of Inskip SB, it is not clear that any of these five reaches are unduly affecting the overall dissimilarity results. However, a strict analysis removing all the previously selected "questionable" reaches may prove to be useful. 81 Table 33 lists the summary dissimilarity results using all possible reach pair combinations excluding Inskip SB, Inskip Main, Carmanah, Gregory Upper Main and Gregory NB: Table 33 Dissimilarity Results - Selected Reach Pair Combinations Reach Pa i r s D iss imi la r i t y Reach Pa i r s D iss im i la r i t y G o v M a i n - G o v Upper M a i n 6.44 Jason Lower - Jason Upper 6.89 G o v Ma in - Jason Lower 5.40 Jason Lower - G o v N B E F 6.64 G o v M a i n - Jason Upper 6.01 Jason Lower - G o v N B N F 6.92 G o v M a i n - Gov N B E F 8.46 Jason Lower - G o v N B 4.16 G o v M a i n - G o v N B N F 8.80 Jason Upper - G o v N B E F 5.44 G o v M a i n - G o v N B 6.69 Jason Upper - G o v N B N F 7.71 G o v Upper - Jason Lower 4.74 Jason Upper - G o v N B 5.16 G o v Upper - Jason Upper 3.94 G o v N B E F - G o v N B N F 6.51 G o v Upper - G o v N B E F 5.27 G o v N B E F - G o v N B 4.05 G o v Upper - G o v N B N F 7.30 G o v N B N F - Gov N B 6.00 G o v Upper - G o v N B 3.33 The most dissimilar reaches (in order of decreasing dissimilarity) are: Gov Main -Gov NB NF, and Gov Main - Gov NB EF. The high total dissimilarity for Gov Main - Gov NB NF can be largely attributed to the mean bankfull width characteristic (d2wt = 11 -08). With the exclusion of Carmanah, Gov Main now has the largest Wb value (32.3 m). In contrast Gov NB NF has the smallest Wb value (10.6 m). Gov Main is an uncoupled, outlet reach while Gov NB NF is a steeper, coupled reach. As discussed previously, it is possible to infer certain general reach characteristics from the channel configuration. Coupled reaches tend to have greater relative roughness values (as is the case with Gov NB NF), greater LWD volume values (as is the case with Gov NB NF), and differences in channel unit characteristics. In the Gov Main - Gov NB NF case, pool length frequency is the most influential characteristic. Gov Main has a pool length frequency of 0.7, nearly twice that of Gov NB NF. 82 A similar situation applies to the Gov Main - Gov NB EF case. Similar to Gov NB NF, Gov NB EF is a smaller, coupled reach. Differences in mean Wb, LWD volume, relative roughness, and certain channel unit characteristics (in this case cascade length and log step frequency) have resulted in the high total dissimilarity value for this reach pair. Unlike the results from Table 30 (all reach pair combinations), channel configuration appears to play a more influential role here in the dissimilarity calculations. The most similar reach pair combinations (in order of increasing dissimilarity) are: Gov Upper - Gov NB, and Gov Upper - Jason Upper. As discussed in Section 5.2.1, the high degree of similarity between these two reaches is likely related to the fact that Gov Upper is located immediately adjacent to Gov NB. Although in this case channel configuration is different, mean bankfull widths and channel gradient are similar. While Gov Upper and Jason Upper are not located adjacent to each other, they do have the same channel configuration and have similar contributing areas and mean bankfull width values. 5.2.5 Discussion Several characteristics stand out as being somewhat influential in the dissimilarity calculations: relative roughness, LWD characteristics, and certain channel unit characteristics. Differences that exist between reaches (based on these influential stream channel characteristics) are often best explained by considering the stream reach position 83 within the watershed. This frequently relates to channel configuration and, ultimately, sediment characteristics. This raises the question of whether sediment characteristics (and sediment-related characteristics) should be dominant. While the reliability of the original D95 data is somewhat questionable, the importance of sediment in characterizing stream channels is undeniable. One of the principal governing conditions for stream channels is the magnitude and time distribution of sediment supplied to the channel from the land surface. The calibre of the sediment is also important for it determines the mobility of the sediment once in the channel. In other words, the stability of the channel is determined by the ease with which bed material can be remobilized. Thus channel disturbance (or instability) should theoretically be reflected in the bed material. Another issue to consider is whether there is a relation between geographic proximity and dissimilarity. To investigate this a Spearman Rank correlation test was performed on all selected old-growth reach pair combinations. Reach pairs were given a rank for both their dissimilarity and their geographic proximity (see Tables 34 and 35): 84 Table 34 Spearman 1 tank Correlation Test for Selected Old-Growth Reach Pairs Reach Pairs Dissimilarity Geographic Proximity (km) Rank (Dissimilarity) Rank (Geographic Proximity) Gov Main - Gov Upper Main 6.44 2.00 12 7 Gov Main - Jason Lower 5.40 10.60 8 17 Gov Main - Jason Upper 6.01 12.40 11 21 Gov Main - Gov NB EF 8.46 2.95 20 11 Gov Main - Gov NB NF 8.80 1.70 21 5 Gov Main - Gov NB 6.69 0.50 15 1 Gov Upper - Jason Lower 4.74 8.55 5 12 Gov Upper - Jason Upper 3.94 10.40 2 16 Gov Upper - Gov NB EF 5.27 2.20 7 9 Gov Upper - Gov NB NF 7.30 1.95 18 6 Gov Upper - Gov NB 3.33 1.60 1 4 Jason Lower - Jason Upper 6.89 2.10 16 8 Jason Lower - Gov NB EF 6.64 9.15 14 13 Jason Lower - Gov NB NF 6.92 10.00 17 14 Jason Lower - Gov NB 4.16 10.15 4 15 Jason Upper - Gov NB EF 5.44 10.75 9 18 Jason Upper - Gov NB NF 7.71 11.65 19 19 Jason Upper - Gov NB 5.16 11.95 6 20 Gov NB EF - Gov NB NF 6.51 1.30 13 2 Gov NB EF - Gov NB 4.05 2.45 3 10 Gov NB NF - Gov NB 6.00 1.30 10 3 Table 35 Spearman Rank Order Correlations Number of Reach Pairs p-level Spearman R t(n-2) Dissimilarity & Geographic Proximity 21 0.410 -0.190 -0.842 The Spearman Rank Order Correlation results reveal that the correlation is not significant. In other words, the reach pairs closest in geographical proximity are not necessarily the most similar. This increases the value of the dissimilarity analysis, as comparisons can be made over some distances. The overall objective of this project was to quantify the variability of stream channel morphology. This can be accomplished by constructing frequency distributions of the dissimilarity values for various groups (e.g., all old-growth, uncoupled old-growth, coupled old-growth, and selected old-growth) (Figure 21). 85 (A) 15 g 10 I 5 (B) All Old-Growth Reach Pair Combinations nflnn^ V* 1^ U l U> I ' f m r*- co a* *- o Dissimilarity (upper limits) All Uncoupled Old-Growth Reach Pair Combinations £ 6 1 O A <g 4 Z 2 0 U I U> I - -^ - c u ^ i n f - c o c * * - o Dissimilarity (upper limits) (C) (D) All Coupled Old-Growth Reach Pair Combinations w 6 o O Z 2 n, n U l v III V V* CO <7> *~ O Dissimilarity (upper limits) All Selected Old-Growth Reach Pair Combinations 8 «> 6 o A o o Z 2 0 n Dissimilarity (upper limits) Figure 21 Frequency Distribution of Dissimilarity Values: (A) A l l O l d - G r o w t h Reach Pa i r s , (B) A l l Uncoup led O l d - G r o w t h Reach Pa i r s , (C) A l l C o u p l e d O l d - G r o w t h Reach Pa i r s , and (D) A l l selected O l d - G r o w t h Reach Pa i r s These data were imported in to Statistica® in order to evaluate the fit o f the observed data to a variety of hypothesized distributions. The ' A l l Old-Growth' data did not deviate significantly from the standard lognormal distribution (Chi-Square = 1.90, df = 5, p = 863) (see Figure 22). This distribution fitting analysis was not performed on any of the other reach pair groups due to insufficient sample sizes. 86 All Old-Growth Reach Pairs with Lognormal Distribution Superimposed Chi-Square: 1.900865, df = 5, p = .8626809 (df adjusted) | Expected 2.2 2.9 3.6 4.3 5.0 5.7 6.4 7.1 7.8 8.5 9.2 9.9 10.6 11.3 12.0 Dissimilarity (upper limits) Figure 22 Frequency Distribution of Dissimilarity Values for All Old-Growth Reach Pair Combinations (Lognormal Distribution Superimposed). It is now possible to go back to the underlying question of this study: Can a target state be defined? Looking at Figures 21 and 22, the answer to that question is no. The conditions are simply too scattered for a unique target state to be identified. However, the distribution illustrated in Figure 22 may be used to define high dissimilarity, thereby establishing a basis for identifying "pathological" (undesirable) states. Two basic approaches may be taken to define high dissimilarity. Empirically, the upper modal group (e.g., high dissimilarity > 9.20) could be used. This isolates 5 reach pairs, all involving Inskip SB. Statistically, the upper 10% of the reach pairs could be selected. This isolates 7 pairs whose dissimilarity values are greater than or equal to 8.56. Six out of these seven reach pairs involve Inskip SB. The 'all old-growth' reach pair group serves to illustrate how dissimilarity testing may be used to identify reaches which are potentially undesirable. In the case above, reach pairs involving Inskip SB consistently achieve dissimilarity values judged to be significantly high. Therefore Inskip SB would be a reach of concern. However, this 87 database does not by any means constitute a reference set for the Queen Charlotte Islands region of British Columbia. An ideal reference set would have all reaches meeting the assessment criteria outlined in Table 6, and have a sample size large enough so that the standard deviation values of the stream channel characteristics (used to standardize the variates) would remain stable. The selected old-growth reach pair group, although very small, is at present the best reference set available. While the Chi-square distribution fitting analysis cannot be used on this group (due to insufficient sample size), a Kolmogorov-Smirnov test can be performed. The Kolmogorov-Smirnov test is more a technique to detect gross deviations from a theoretical distribution. The results from this test are illustrated in Figure 23: Selected Old-Growth Reach Pairs with Normal Distribution Superimposed Kohnogorov-Smirnov d = .0264312, p = n.s. 7 , ; Figure 23 Frequency Distribution of Dissimilarity Values for Selected Old-Growth Reach Pair Combinations (Normal Distribution Superimposed). Based on both visual inspection and the Kolmogorov-Smirnov test, the distribution for 'selected old-growth' reach pairs does not deviate significantly from the theoretical normal distribution. Using this distribution as the reference set for the Queen Charlotte Islands, new reaches (which pass the assessment criteria outlined in Table 6) could be 88 assessed. One could simply introduce a new reach into this group, recalculate dissimilarity values for all reach pair combinations, and assess the dissimilarity values for reach pairs involving the new reach. I f reach pairs involving the new reach consistently achieved dissimilarity values judged to be significantly high, the new reach would be considered undesirable. Alternatively, reach pairs with what are judged to be low dissimilarity values could be used as paired experiment/control reaches. 5.3 Managed Stream Channels In order to further explore the efficacy of dissimilarity testing it is necessary to add managed stream channels to the project database. Forestry activities may affect stream channels by altering sediment delivery rates, manipulating both riparian vegetation and instream L W D structure, and modifying flood flow characteristics (Hogan et al., 1996). The managed stream channels w i l l provide a contrast to the old growth stream channel dissimilarity results. Table 36 summarizes the relevant stream channel characteristics for each sub-reach. Unlike the old-growth streams, no major problems arose during the data extraction. The stone line morphological class was excluded from this matrix table, as none of the managed reaches had stone line features delimited. 89 <» JS CU p f ! S t/5 a 88 © — « H H 08 — 3 58 H CD 3 3 PH l a B •8 ? > & 60 • s ; o t ca ; ft' t/3 U 3 In CD LL, a i>1 p £ I? CD [JL, a cu1 p £ Is1 ^ P H IS a <" Q P PH - H -S O 60 O S . . PH « PH M o £ c DH P PH 3 in CD tt, HJ 5 3 a 2 P PH U ca cu Pi r- C N X I 3 CL> ft ft < CD _3 "ca > "3 3 o eg CD O 3 CD CT H PH 60 3 CD '3 P 13 3 43 u •a u 3 CD 3 I T CD 3 P 13 3 43 o •3 ca CJ CA ca o CD 43, ca CD S3 CD X ) ca u CD 43 -a CD 4> CB ta X I & PH PH H .s op o CD CO C 0) O CD X I ft O CD «J c/3 ca 4-» CJ • 3 00 co s o ca D H J U II II CD 3 43 (/J U o 5.3.1 Uncoupled Reach Pair Combinations A general dissimilarity index comprised of the summary stream channel data and the standard deviation of each stream channel characteristic was constructed. For any given reach pair combination, this index was then used to calculate dissimilarity values for each stream channel characteristic as well as the total dissimilarity. Table 37 lists the summary dissimilarity results using all managed reach pair combinations: Table 37 Dissimilarity Results - All Managed Reach Pair Combinations Reach Pa i r s D iss imi la r i t y Reach Pa i r s D iss im i la r i t y M o s M a i n - M o s Upper 5.91 M o s Upper - Peel 6.39 M o s M a i n - R i ley Lower 1 4.57 Ri ley Lower 1 - R i ley Lower 2 4.02 M o s Ma in - R i ley Lower 2 4.57 Ri ley Lower 1 - Tarundl 6.51 M o s M a i n - Tarundl 4.91 R i ley Lower 1 - Peel 6.39 M o s M a i n - Peel 5.42 Ri ley Lower 2 - Tarundl 4.90 M o s Upper - R i ley Lower 1 8.30 Ri ley Lower 2 - Peel 6.19 Mos Upper - R i ley Lower 2 7.11 Tarundl - Peel 4.62 M o s Upper - Tarundl 5.70 The most dissimilar reach pairs (in order of decreasing dissimilarity) are: Mos Upper - Riley Lower 1 and Mos Upper - Riley Lower 2. The reach pair combinations with the most dissimilar results involve Mos Upper. The high total dissimilarity value for Mos Upper - Riley Lower 1 can largely be attributed to two stream channel 2 2 characteristics: pool spacing (d p o o i s p = 9.58) and log step length frequency (d fafe„ = 6.70). The high total dissimilarity value for Mos Upper - Riley Lower 2 can be attributed to three stream channel characteristics: width variability (d2wvar = 7.98), depth variability ( d 2 ^ = 7.72), and large woody debris spacing (d2LwDsP = 7.58). The high values for d 2 w v a r and d2^ are related to the fact that Mosquito Upper has the largest width and depth variability values in the managed database (see Table 36). The high 91 dissimilarity value for the L W D spacing characteristic is related to the fact that the L W D spacing value for Riley Lower 2 is over twice that of any other managed reach. The reason for Riley Lower 2 having such a large L W D spacing value is unknown. It is particularly curious as both Riley sub-reaches (Riley Lower 1 and Riley Lower 2) were selected from within the same reach, but have significantly different values for L W D spacing. The most similar reach pairs (in order of increasing dissimilarity) are: Riley Lower 1 - Riley Lower 2 and Mos M a i n - Riley Lower 2. It is not surprising that the most similar reach pair is Riley Lower 1 - Riley Lower 2, as these two sub-reaches are situated adjacent to each other and are also located within the same reach. While there are not many substantial differences between Mos Main and Riley Lower 2, one of the • 2 key discriminators appears to be L W D spacing (d LWDSP = 5.67). A s stated previously, the L W D spacing value for Riley Lower 2 is relatively high and, more importantly, questionable. Four of the managed reaches were pinpointed in Section 2.4.2 as not meeting the assessment criteria in Table 6. While there exists substantial variation in both geology and intensity of logging disturbance, no attempt has been made to create a 'selected, managed' reach pair group. I f these factors were controlled, only two of the reaches (Mos M a i n and Mos Upper) would remain in the database. With only one reach pair combination, no analysis would be possible. 92 5.4 Old Growth vs. Managed Stream Channels 5.4.1 Uncoupled Reach Pair Combinations In order to analyze differences between the managed and old-growth reaches, the dissimilarity tests were performed for all possible reach pair combinations where one reach was old-growth and one reach was managed. As all the managed reaches were uncoupled, only the selected, uncoupled old-growth reaches were included in this analysis. The results are presented in Table 38. Table 38 Dissimilarity Results - Selected Old Growth vs. Managed Reach Pair Combinations Reach Pairs Dissimilarity Reach Pairs Dissimilarity M o s M a i n - G o v M a i n 4.47 Ri ley Lower 2 - Gov M a i n 5.48 M o s M a i n - G o v Upper 6.00 R i ley Lower 2 - G o v Upper 4.93 M o s M a i n - Jason Lower 5.89 R i ley Lower 2 - Jason Lower 4.65 M o s M a i n - Jason Upper 6.08 Ri ley Lower 2 - Jason Upper 6.27 M o s Upper - G o v M a i n 8.36 Tarundl - G o v M a i n 5.84 M o s Upper - G o v Upper 7.02 Tarundl - G o v Upper 4.56 M o s Upper - Jason Lower 9.04 Tarundl - Jason Lower 6.00 M o s Upper - Jason Upper 6.66 Tarundl - Jason Upper 6.23 Ri ley Lower 1 - G o v M a i n 4.17 Peel - G o v M a i n 6.19 Ri ley Lower 1 - G o v Upper 5.51 Peel - G o v Upper 4.93 R i ley Lower 1 - Jason Lower 4.11 Peel - Jason Lower 6.44 R i ley Lower 1 - Jason Upper 6.19 Peel - Jason Upper 4.88 The most dissimilar reach pair combinations all involve Mosquito Upper. This was expected, considering Mosquito Upper has the highest depth variability value and one of the highest width variability values in the database. The most similar reach pair combinations (in order of increasing dissimilarity) are: Riley Lower 1 - Jason Lower, Riley Lower 1 - Gov Main, and Mosquito Main - Gov Main. It is worth noting that the smallest total dissimilarity value for the 'old-growth vs. managed' group is greater than the smallest total dissimilarity value for any other reach pair combination group (e.g., old-growth uncoupled, old-growth coupled, selected old-growth, and managed uncoupled). 93 5.5 Discussion It is now possible to examine differences in the dissimilarity results between old-growth and managed reach pairs. As all the managed streams are uncoupled, only the selected, uncoupled old-growth reach pairs are required for this analysis. Three reach groups are compared: • Selected old-growth (uncoupled) • Managed (uncoupled) • Selected old-growth (uncoupled) vs. Managed (uncoupled) Due to insufficient sample sizes, these three reach pair groups are assessed by comparing means and standard deviations (Table 39 and Figure 24): Table 39 Comparison of Mean Dissimilarity Values and Standard Deviations Between Different Reach Pair Combination Groups G r o u p Reach P a i r Comb ina t i on G r o u p s M e a n D iss im i la r i t y V a l u e S tandard Deviat ion A Selected old-growth uncoupled 5.40 1.13 B Managed uncoupled 5.73 1.13 C Selected old-growth (uncoupled) vs. Managed (uncoupled) 5.83 1.20 As can be seen in Table 39, the lowest mean dissimilarity value is found within Group A. This is expected, as the selected old-growth stream channels have the most homogenous basin characteristics. 94 7.6 7.0 6.4 '5 I 5-8 to S 5.2 4.6 4.0 Box & Whisker Plots, Groups A-C Group A Group B Group C _L_ ±Std.Dev. I I ±Std. Err. • Mean F i g u r e 2 4 B o x a n d W h i s k e r P l o t f o r G r o u p s A - C : A = Selected O l d - G r o w t h (uncoupled), B = Managed (uncoupled), and C = Selected O l d - G r o w t h (uncoupled) vs. M a n a g e d (uncoupled). Figure 24 illustrates the differences between Groups A through C rather clearly. Group C (selected old-growth vs. managed) has the highest dissimilarity. This is not surprising, as Group C should theoretically have the greatest contrasts between reaches. It is important to note that there is overlap between groups in both the standard errors and standard deviations. This is expected, as the sample sizes are relatively small and the ranges of dissimilarity values are relatively large. The discussion in Section 5.2.5 included a description of the method by which a reach could be compared with a reference set. The following example illustrates how this method would work. In this example, the selected, uncoupled old-growth reach pair group will be used as the reference set. 95 Table 40 Dissimilarity Results - Reference Set (Selected, uncoupled old-growth Reach Pa i r s D iss im i la r i t y Jason Upper - Jason Lower 6.83 G o v M a i n - Jason Upper 6.69 G o v M a i n - G o v Upper 6.59 G o v M a i n - Jason Lower 5.33 G o v Upper - Jason Upper 4.55 G o v Upper - Jason Lower 4.46 The range of dissimilarity values in Table 40 is small, with the highest dissimilarity value being 6.83. A reach which will provide the greatest contrast to this reference set will best illustrate how the dissimilarity method for assessing stream reaches works. Mosquito Upper was selected for this example as all reach pairs involving Mosquito Upper were judged to have 'high' dissimilarity values. Table 41 lists the new dissimilarity results for the reference set (now including Mosquito Upper): Table 41 Dissimilarity Results - Reference Set with Mosquito Upper Reach Pa i r s D iss imi la r i t y Jason Lower - M o s Upper 8.09 G o v M a i n - M o s Upper 7.31 G o v Upper - M o s Upper 6.42 Jason Upper - M o s Upper 6.21 Jason Upper - Jason Lower 5.91 G o v M a i n - Jason Upper 5.65 G o v M a i n - G o v Upper 5.06 G o v Upper - Jason Upper 4.37 G o v M a i n - Jason Lower 4.24 G o v Upper - Jason Lower 3.42 As previously discussed, the small size of the reference set results in unstable standard deviation values. Thus the addition of Mosquito Upper to the reference set does change the dissimilarity values of reach pairs found in both Tables 40 and 41. At any rate, the reach pairs involving Mosquito Upper have the greatest dissimilarity values. With a larger reference set (and thus stable standard deviation values), reach pairs involving Mosquito Upper would likely be even more dissimilar. 96 Reach pairs involving Mosquito Upper have what are judged to be high dissimilarity values. A s a result, Mosquito Upper could be called 'pathological', or even 'severely impacted' (within this system of comparison). It follows that the general restoration goal for Mosquito Upper is to lower its dissimilarity values. In other words, the dissimilarity values for reach pairs involving Mosquito Upper should ideally be comparable to the rest of the reference set (in this example, dissimilarity values less than or equal to 6.83). More particularly, by analyzing the dissimilarity values for individual stream channel characteristics it may be possible to indicate where particular problem areas do or do not exist. Those stream channel characteristics exhibiting low dissimilarity values may not require attention, while characteristics with high dissimilarity values most likely would. Thus the dissimilarity testing procedure could be a powerful tool to help guide restoration efforts. 97 Chapter 6; Conclusions In this study a method o f quant i fying stream channel var iab i l i ty and def ining undesirable states (thereby quant i fying stream channel impact) has been developed. Three pre l iminary steps were i nvo lved i n developing this method: (1) select ion o f suitably s imi l a r drainage basins (2) select ion o f suitable stream channel characteristics; and (3) select ion o f suitable stream reaches In a l l o f these steps important issues were identif ied and addressed. The selection o f suitably s imi la r drainage basins was based o n a method adapted from C h e o n g (1992, 1996). Cheong ' s bas in classif icat ion procedure is the most thorough, systematic approach avai lable , incorporat ing both b iogeophys ica l and morphometr ic characteristcs. However , it is not w e l l suited to the smaller-s ized sub-basins w h i c h are the focus o f this project. In particular, the C h e o n g basin classif icat ion procedure does not appear to effectively characterize differences i n channel configurat ion. It is not clear that channel configurat ion ( w h i c h m a y be reach specific) can be resolved f rom maps or systematical ly correlated w i t h va l l ey flat area. The under ly ing p rob lem w i t h u t i l i z ing a rather r igorous basin c lass i f icat ion procedure is that it is very diff icul t to get a sizable sample group w h i c h meets a l l the assessment cr i ter ia (e.g. Table 6). The absence o f formal bas in c lass i f icat ion procedures i n many comparat ive stream channel studies is l i k e l y related to this issue. In this study, a relat ively large percentage o f the candidate streams do not meet a l l the assessment criteria. 98 The process of selecting suitable stream channel characteristics is difficult because no clear, quantitative approach exists for determining which variables best characterize stream channels. While stream channel processes are probably the key issue when considering the idea of characterizing stream channels, the stream channel morphology is focussed on here as it can be assessed more readily and reflects aquatic habitat quality directly. Several important issues arose during the selection of suitable stream channel reaches. First and foremost, no set standards exist for identifying suitable stream channel reach lengths. While the term 'reach' in the strict sense is defined as a homogenous unit within which the controlling factors do not change appreciably (Church, 1992), it is often used to describe any length of channel being studied. As arbitrarily choosing a reach length for stream channel comparison is problematic, a quantitative method for determining suitable reach lengths was developed. This method was based on the Representative Elementary Area concept, introduced by Wood et al. in 1988. Representative reach lengths were determined by assessing the variance of depth deviations over increasingly long reach lengths. Once variance values stabilized, a reach was considered representative. This resulted in a range of acceptable reach length values (25 to 62 Wb). A reach length of 25 Wb is relatively short considering the fact that features such as LWD jams may influence sedimentation and, ultimately, channel morphology for distances exceeding 100 Wb (Hogan and Bird, 1998). However, some reaches (in the strict sense of the word) are simply not very long. This is the case for all 99 those selected sub-reaches less than 30 Wb in length. For those sub-reaches, the entire available reach lengths were used. Stream channels were compared by adapting the method developed by Cheong (1992) for basin comparison. This method is based on the concept of dissimilarity, and incorporates a euclidean distance measure in order to calculate the 'proximity' between two objects (Gordon, 1981). While the comprehensive dissimilarity testing procedure developed by Cheong (1992) was sound, adaptations to the actual dissimilarity formula were required. This related to the method of standardizing variables in order to remove bias from the results. Furthermore, as all the selected stream channel characteristics were scale free, it was uncertain that standardization was required at all. In order to investigate this question several dissimilarity testing procedures were carried out. The results from this investigation made clear the importance of standardizing each variable (scale free or not). Without standardizing each variable, the calculation method places greater emphasis on those variables which have higher numeric values. Although the stream channel characteristics derived from the FFIP channel surveys are fairly comprehensive, further investigation into ways to characterize sediment characteristics would be a worthwhile endeavor. There is the question of how reliable the relative roughness values are, as they are based on the original D95 data which were determined through a quick visual estimate by a field worker. There is no guarantee of the accuracy of this measure, unlike a rigorous procedure such as bulk sampling. The results from comparing stream channels using the dissimilarity testing procedure are promising. Reach pairs exhibiting high dissimilarity values tend to have significant 100 differences in several key stream channel characteristics. The key stream channel characteristics vary between reach pairs, but commonly include relative roughness, LWD characteristics and average bankfull width (used as a surrogate for scale). A Spearman Rank Order correlation revealed that reach pairs closest in geographical proximity are not necessarily the most similar. This increases the value of the dissimilarity analysis, as it shows that comparisons can be made over some distances. Dissimilarity values varied between reach pairs depending on the reach pair group involved. This is related to changes in sample sizes, which subsequently alter the standard deviation values used to standardize variables. If sufficiently large sample sizes were available, the standard deviation values would be stable and the changes in dissimilarity values would not occur. High dissimilarity was judged to be greater than or equal to 8.56 for the 'all old-growth' reach pair group. This value does not by any means constitute a reference dissimilarity value for the Queen Charlotte Islands region, as the sample size is simply too small and both coupled and uncoupled reach pairs are involved in the group. An ideal reference set would have all reaches meeting the assessment criteria outlined in Table 6 (including similar channel configuration) and also be large enough so that the standard deviation values would remain stable. Reach pairs with dissimilarities judged to be high (compared to a suitable reference set) would require closer examination. Those reaches consistently appearing in reach pairs with high dissimilarity values could be considered undesirable, or 'severely impacted' (within this system of comparison). Thus the dissimilarity method of comparing stream channel reaches enables definition of 101 undesirable states and quantification of impact. Conversely, reach pairs with low dissimilarity values could be considered 'similar'. Hence the dissimilarity method of comparing stream channel reaches may also be used to identify reaches suitable for experimental treatment/control studies. This study represents a first attempt at quantifying the range of natural states found in old-growth forest streams (represented by frequency distributions of dissimilarity values). These distribution can be used to define undesirable states and quantify impact. While these results are promising, they are disappointingly limited by sample size. Even if the basin assessment criteria (Table 6) did not play a role in diminishing the sample size, the FFIP database in itself is not large enough for the resulting dissimilarity values to be stable. As stated previously, of the remaining intact, old-growth drainage basins left in British Columbia, most are found in remote locations that are not easily accessible. Of those that have been surveyed, it is difficult to find a substantial grouping with suitably similar basic governing conditions. Ideally, a regional reference set should include a minimum of 20 old-growth stream reaches in order to better characterize the natural range of states. With 20 reaches, the standard deviation values for stream channel characteristics would likely remain stable with the addition of a test reach's stream channel characteristics. In addition, 20 reaches would produce 190 reach pair combinations. With a sample size this large, it would be possible to more accurately determine a reference dissimilarity value (e.g., an upper limit, or 'high dissimilarity' value). 102 References Abbe, T. B. and Montgomery, D. R. (1996). Large woody debris jams, channel hydraulics and habitat formation in large rivers. Regulated Rivers: Research & Management, 12: 201-221. Andrus, C. W., Long, B. A., and Froehlich, H. A. (1988). Woody debris and its contribution to pool formation in a coastal stream 50 years after logging. Canadian Journal of Fisheries and Aquatic Sciences, 45: 2080-2086. ASPECT Editorial (August 2000). ASPECT- The DEGIFSNewsletter (Division of Engineers and Geoscientists in the Forest Sector, APEGBC), 5 (2): 2-3. Banner, A., Pojar, J., and Trowbridge, R. (1983). Ecosystem classification of the Coastal Western Hemlock zone, Queen Charlotte Island Subzone (CWHg), Prince Rupert Forest Region, British Columbia. Unpublished Report, British Columbia Ministry of Forests, Smithers, B. C. 235 pp. Bisson, P. A., Nielson, J. L. , Palmason, R. A., and Grove, L . E. (1982). A system of naming habitat types in small streams, with examples of habitat utilization by salmonids during low streamflow. In Armantrout, N. B., editor, Proceedings of a Symposium on Acquisition and Utilization of Aquatic Habitat Inventory Information: Portland Oregon, Western Division of the American Fisheries Society, p. 62-73. Bird, S. (1998). Stream channels, large woody debris and biogeoclimatic zones, year-two progress report. Unpublished Report, Research Branch, British Columbia Ministry of Forests, 62 pp. Bowden, K. L. , and Wallis, J. R. (1964). Effect of stream-ordering technique on horton's laws of drainage composition. Geological Society of America Bulletin, 75: 767-774. Cheong, A. (1992). Quantifying drainage basin comparisons within a knowledge-based system framework. Unpublished M.Sc. Thesis, Department of Geography, The University of British Columbia, 133 pp. Cheong, A. (1996). Classifying and comparing drainage basins in British Columbia. Watershed Restoration Guidebook. Fisheries Branch, British Columbia Ministry of Environment, Lands and Parks. Chorley, R. J., Schumm, S. A., and Sugden, D. E. (1984). Geomorphology, Methuen, London, 605 pp. 103 Church, M . (1983). Concepts of sediment transfer and transport on the Queen Charlotte Islands. Unpublished report, Fish/Forestry Interaction Program, 30 pp. Church, M . (1992). Channel morphology and typology. In Calow, P. and Petts, G. E. , editors, The Rivers Handbook. Oxford, Blackwell. Church, M . (1997). Regionalised hydrological estimates for British Columbia: First approximation of scale effect. Resources Inventory and Data Management Branch, British Columbia Ministry of Environment, Lands and Parks. Victoria, B . C . (Contract #CEF6156). Church, M . and Jones, D. (1982). Channel bars in gravel-bed rivers. In Hey, R. D., Bathurst, J. C , and Thorne, C. R., editors, Gravel-bed rivers. Chichester, England, John Wiley and Sons, p. 291-324. Cronan, C , Driscoll, C , Newton, R., Kelly, J., Schofield, C , Bartlett, R., and April, R. (1990). A comparative analysis of aluminum biogeochemistry in a northeastern and a southeastern forested watershed. Water Resources Research, 26(7): 1413-1430. Davis, W. M . (1899). The geographical cycle. Geographical Journal 14:81-504. de Villiers, A. B. (1986). A multivariate evaluation of a group of drainage basin variables - a South African case study. International Geomorphology Part II, John Wiley and Sons, p. 21-32 Ebisemiju, F. S. (1986). Environmental constraints of the interdependence of drainage basin morphometric properties. International Geomorphology Part II, John Wiley and Sons, p. 3-20. Gardiner,V., and Park, C. C. (1978). Drainage basin morphometry: review and assessment. Progress in Physical Geography, 2(1): 1-35. Gordon, A. D. (1981). Classification: methods for the exploratory analysis of multivariate data. Chapman and Hall, London. 193 pp. Grant, G. E. , Swanson, F. J., and Wolman, M . G. (1990). Pattern and origin of stepped-bed morphology in high-gradient streams, Western Cascades, Oregon. Geological Society of America Bulletin, 102: 340-352. Hewlett, J. D. (1971). Review of representative and experimental basins. Bulletin of the American Meteorological Society, 52: 892-893. Hogan, D. L . (1985). Stream channel morphology: Comparison of logged and unlogged watersheds in the Queen Charlotte Islands. Unpublished M.Sc. Thesis, Department of Geography, The University of British Columbia, 220 pp. 104 Hogan, D. L . (1986). Channel morphology of unlogged, logged, and debris torrented streams in the Queen Charlotte Islands. British Columbia Ministry of Forests and Lands. Land Management Report 49, 94 pp. Hogan, D. L. (1989). Channel response to mass wasting in the Queen Charlotte Islands, British Columbia: Temporal and spatial changes in stream morphology. In Proceedings of Watersheds '89, A conference on the stewardship of soil, air and water resources, Juneau, Alaska, March 21-23, 1989, United States Department of Agriculture, Forest Service, Alaska Region, R10-MB-77, p. 125-144. Hogan, D. L. , and Church, M . (1989). Hydraulic geometry in small, coastal streams: Progress toward quantification of salmonid habitat. Canadian Journal of Fisheries and Aquatic Science, 46: 844-852. Hogan, D. L. , Bird, S. A., and Wilford, D. J. (1996). Channel conditions and prescriptions assessment (interim methods). Watershed Restoration Program, British Columbia Ministry of Environment, Lands and Parks and Ministry of Forests. Hogan, D. L. , and Bird, S. A. (1998). Classification and assessment of small coastal stream channels. In Hogan, D. L. , Tschaplinski, P. J., and Chatwin, S., editors, Carnation Creek and Queen Charlotte Islands Fish/Forestry workshop: Applying 20 years of coastal research to management solutions. British Columbia Ministry of Forests, Research Program, Land Management Handbook No. 41. Hogan, D. L. , Bird, S. A., and Hassan, M . A. (1998). Spatial and temporal evolution of small coastal gravel-bed streams: Influence of forest management on channel morphology and fish habitats. In Klingeman, P. C , Beschta, R. L. , Komar, P. D., and Bradley, J. B., editors, Gravel-Bed Rivers in the Environment. Water Resources Publications, L L C . Holland, S. S. (1965). Landforms of British Columbia, a physiographic outline. Bulletin 48, Department of Mines and Petroleum Resources, 138 pp. Horton, R. E. (1945). Erosional development of streams and their drainage basins: Hydrophysical approach to quantitative morphology. Geological Society of America Bulletin 56: 275-370. Keller, E. A., and Melhorn, W. N. (1978). Rhythmic spacing and origin of pools and riffles. Geological Society of America Bulletin, 89:723-730. Keller, E. A., and Swanson, F. J. (1979). Effects of large organic material on channel form and fluvial processes. Earth Surface Processes, 4:361-380. 105 Keller, E. A. , and Tally, T. (1979). Effects of large organic debris on channel form and fluvial processes in the coastal redwood environment. In Rhodes, D. D., and Williams, G. P., editors, Adjustments of the fluvial system. Kendall-Hunt, Dubuque, Iowa, p. 169-197. Leopold, L. B., Wolman, M . G. and Miller, J. P. (1964). Fluvial processes in geomorphology. W. H. Freeman and Company, San Francisco, 522 pp. Lewis, L. A. (1969). Analysis of surficial landform properties - the regionalization of Indiana into units of morphometric similarity. Proceedings of the Indiana Academy of Science, 78:317-328. Mark, D. M . (1975). Morphometric parameters: A review and evaluation. Geografiska Annaler, 57A(3-4): 165-177. Melton, M . A. (1957). An analysis of relations amongst elements of climate, surface properties, and geomorphology. Technical Report 11, United States Office of Naval Research. 32 pp. Montgomery, D. R., Buffington, J. M . , Smith, R. D., Schmidt, K. M . , and Pess, G. (1995). Pool spacing in forest channels. Water Resources Research, 31(4): 1097-1105. Montgomery, D. R., and Buffington, J. M . (1997). Channel-reach morphology in mountain drainage basins. Geological Society of America Bulletin, 109(5): 596-611. Naiman, R. J., Lonzarich, D. G., Beechie, T. J., and Ralph, S. C. (1992). General principles of classification and the assessment of conservation potential in rivers. In Boon, P. J., Calow, P., and Petts, G. E. , editors, River Conservation and Management, p. 93-123. Nakamura, F. and Swanson, F. J. (1993). Effects of coarse woody debris on morphology and sediment storage of a mountain stream system in western Oregon. Earth Surface Processes and Landforms, 18: 43-61. Playfair (1802). Illustrations of the Huttonian theory of the earth. William Creech, Edinburgh, 528 pp. Ralph, S. C , Poole, G. C , Conquest, L. L . , and Naiman, R. J. (1994). Stream channel morphology and woody debris in logged and unlogged basins of western Washington. Canadian Journal of Fisheries and Aquatic Sciences, 51: 37-51. Rice, S. (1990). The spatial variation of bed material texture in coupled basins on the Queen Charlotte Islands. Unpublished M.Sc. Thesis, Department of Geography, The University of British Columbia, 128 pp. 106 Rierkirk, H. (1989). Influence of silvicultural practices on the hydrology of pine flatwoods in Florida. Water Resources Research, 25(4): 713-719. Rodda, J. C. (1976). Chapter 10: Basin studies. In J. C. Rodda, editor, Facets of Hydrology, John Wiley and Sons, London, 368 pp. Roper, B. B., and Scarnecchia, D. L. (1995). Observer variability in classifying habitat types in stream surveys. North America Journal of Fisheries Management, 15: 49-53. Schumm, S. A. (1977). The fluvial system: New York. John Wiley and Sons, 338 pp. Statistica ©1999 by StatSoft Inc. Strahler, A. N. (1964). Quantitative geomorphology of drainage basins and channel networks. In Chow, V. T., editor, Handbook of Applied Hydrology, 4.39-4.79. Sullivan, K. (1986). Hydraulics andfish habitat in relation to channel morphology. Ph.D. Thesis, Johns Hopkins University, Baltimore, Maryland. 407 pp. Sutherland Brown, A. (1968). Geology of the Queen Charlotte Islands. Bulletin 54, British Columbia Department of Mines and Petroleum Resources. 226 pp. Tarboton, D. G., Bras, R. L . and Rodriguez-Iturbe, I. (1989). The analysis of river basins and channel networks using digital terrain data. Report Number 326, Ralph M . Parsons Laboratory, Department of Civil Engineering, Massachusetts Institute of Technology, 251 pp. Valentine, K., Sprout, P., Baker, T., and Lavkulich, L. , editors (1978). The Soil landscapes of British Columbia. Resource Analysis Branch, Ministry of Environment, Victoria, B. C , 197 pp. Wood, E. F., Sivapalan, M . , Beven, K., and Band, L. (1988). Effects of spatial variability and scale with implications to hydrological modeling. Journal of Hydrology, 102: 29-47. Wood-Smith, R. D., and Buffington, J. M . (1996). Multivariate geomorphic analysis of forest streams: implications for assessment of land use impacts on channel condition. Earth Surface Processes and Landforms, 21: 377-393. Wood-Smith, R. D., and Swanson, F. J. (1997). The influence of large woody debris on forest stream geomorphology. In Proceedings of the Conference on Management of Landscapes Disturbed by Channel Incision, May 19-23, 1997. Oxford Campus, The University of Mississippi. Zavoianu, I. (1985). Morphometry of drainage basins. Developments in Water Science, 20, Elsevier, Amsterdam, 238 pp. 107 Appendix A: Measurement Methodology Channel Unit Frequency: Channel unit frequency was calculated for each channel unit type: e.g. Pool frequency = number of pools / total number of channel units Riffle frequency = number of riffles / total number of channel units Channel Unit Length Frequency (m/m): Channel unit length frequency was calculated for each channel unit type: e.g. Pool length frequency = total length of pools(m) / total length of reach(m) Riffle length frequency = total length of riffles(m) / total length of reach(m) Pool Spacing (Wb): The average distance (m) between pool outlets in a reach, standardized by dividing by the mean bankfull width (m). Average Bankfull Width (m): The mean value of all recorded bankfull widths in any given reach. Depth Variability (m/m): Depth variability was calculated by taking the standard deviation of all standardized depth deviations (depth deviation = actual thalweg elevation - estimated thalweg elevation (where the estimated thalweg elevation = second-order polynomial regression fit to the actual thalweg data)). Depth deviations were standardized by dividing by the mean channel depth*. The resulting value is a scale free measure: e.g. Depth variability = st.dev [all standardized depth deviations (m/m)] *Mean depth deviation would theoretically be a better measure with which to standardize the depth deviation values. However, some reaches had depth deviation values fairly equally scattered on both sides of the estimated longitudinal profile, resulting in a very small value for mean depth deviation. Extremely small mean depth deviation values would artificially inflate the depth variability values. Width Variability (m/m): Width variability was calculated by taking the standard deviation of all standardized bankfull width values. Bankfull width values were standardized by dividing by the average bankfull width.: e.g. Width variability = st. dev [all standardized bankfull widths (m/m)] 108 LWD Spacing (Wb): Average distance between LWD jams, standardized by dividing by the average bankfull width: e.g. LWD spacing = mean distance between LWD jams (m) / mean channel width(m) LWD Volume: LWD volume per length of channel, standardized by dividing by the mean channel cross-sectional area: e.g. LWD vol per length of channel = total LWD vol (m3) / channel length (m) Mean channel cross-sectional area = mean bankfull x mean channel width (m) depth (m) Relative Roughness: The average value of all calculated relative roughness ratios in a reach, (relative roughness ratio = D95 value (m) / channel depth (m)). 109 Appendix B: Variance Plots for Old-Growth and Managed Reaches Uncoupled, Old-Growth Reaches Variance Plot: Government Main (Entire Reach) I 0 4 'S •g 0.3 a | 0.2 200 400 600 800 Horizontal Distance (m) 1000 Variance Plot: Jason Lower (Entire Reach) 500 1000 Horizontal Distance (m) Variance Plot: Carmanah (Entire Reach) 500 1000 Horizontal Distance (m) 1500 1500 Variance Plot: Government Upper Main (Entire Reach) 1000 2000 Horizontal Distance (m) 3000 Variance Plot: Jason Upper (Entire Reach) 1000 1200 1400 1600 1800 2000 Horizontal Distance (m) 110 Coupled, Old-Growth Reaches Variance Plot: Gregory NB (Entire Reach) 5800 5900 6000 6100 6200 6300 6400 Horizontal Distance (m) Variance Plot: Inskip NB (Entire Reach) 200 400 600 800 Horizontal Distance (m) 1000 Variance Plot: Gregory Upper Main (Entire Reach) 2000 4000 6000 Horizontal Distance (m) 8000 Variance Plot: Government NB NF (Entire Reach) 1000 2000 Horizontal Distance (m) 3000 Variance Plot: Inskip Main (Entire Reach) 0.4 -I 1 03 " Q I 02 ' o Variance p Variance p o J 0 200 400 600 800 Horizontal Distance (m) Variance Plot: Inskip SB (Entire Reach) 500 1000 Horizontal Distance (m) 1500 Variance Plot: Government NB EF (Entire Reach) 0.4 -I s o • . .-F Depth Deviati 0.3 -0.2 -Variance o! 0.1 -0 - ^ — -1000 1500 2000 Horizontal Distance (m) 2500 Variance Plot: Government NB (Entire Reach) 500 1000 1500 Horizontal Distance (m) 2000 111 Managed Reaches Variance Plot: Mosquito Main* (Entire Reach) 500 1000 1500 2000 Horizontal Distance (m) 2500 Variance Plot: Peel* (Entire Reach) 0.4 0.3 0.2 0.1 \ 500 1000 Horizontal Distance (m) 1500 Variance Plot: Riley Middle (Coupled) (Entire Reach) 2000 4000 6000 Horizontal Distance (m) 8000 Variance Plot: Mosquito Upper* (Entire Reach) 1000 2000 3000 Horizontal Distance (m) Variance Plot: Riley Lower* (Entire Reach) 2000 4000 Horizontal Distance (m) Variance Plot: Tarundl* (Entire Reach) 1000 2000 3000 Horizontal Distance (m) 4000 6000 4000 * Uncoupled 112 

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:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0089911/manifest

Comment

Related Items