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 T H E VARIABILITY IN FOREST S T R E A M C H A N N E L MORPHOLOGY by KRISTIE MARIE TRAINOR B.Sc, The University of British Columbia, 1995 A THESIS SUBMITTED IN PARTIAL F U L F I L L M E N T OF T H E REQUIREMENTS FOR T H E D E G R E E OF M A S T E R OF SCIENCE in T H E 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  T H E UNIVERSITY OF BRITISH C O L U M B I A 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, L W D jam spacing, L W D 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-  ii  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).  iii  Table of Contents Abstract  ii  Table of Contents  iv  List of Tables  vii  List of Figures  x  Acknowledgements  xii  Dedication  xiii  Chapter 1  Introduction  1  1.1 Research Objectives  3  1.2 Organization of the Study  5  Chapter 2  Drainage Basins  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 Chapter 3  29  Stream Channel Characteristics  31  3.1 Introduction  31  3.2 Selection of Stream Channel Characteristics  31  3.2.1 Channel Unit Frequency and Length  iv  31  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 A l l 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 5.4 Old-Growth vs. Managed Stream Channels 5.4.1 Uncoupled Reach Pair Combinations 5.5 Discussion  91 93 93 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 o f a Drainage Basin Table 2  11  Basin Characteristics Measured from 1:50 000 N T S 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 o f Old-Growth Q C I and V I Watersheds  20  Table 6  Drainage Basin/Stream Reach Similarity Assessment  23  Table 7  Additional Biogeophysical, Morphometric and Hydroclimatic Characteristics of Old-Growth Q C I and V I Watersheds  24  Table 8  General Characteristics of Managed Q C I 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 L W D  40  Classification  Table 14 Classification of the Span of a L W D  Jam  41  Table 15 Classification of L W D  Jam Integrity  41  Table 16 Classification of L W D  Jam Height  41  Table 17 Classification o f L W D  Jam Age  41  Table 18 Classification of L W D  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  ix  96  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  Figure 8  Illustration of a Fitted 2  nd  45  -Order Polynomial Regression Line  (Estimated Thalweg Elevation) Superimposed on a Longitudinal Profile (Actual Thalweg Elevation) Figure 9  48  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 Figure 11 Variance Plot for Government N B NF  50 52  Figure 12 Variance Plots for Government U M : Sections 800m - 1740m and 1000m- 1925m Figure 13 Variance Plot: Selected Sub-Reach Length for Government U M  53 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) A l l Old-Growth Reach Pairs, (B) A l l Uncoupled Old-Growth Reach Pairs, (C) A l l Coupled Old-Growth Reach Pairs, and (D) A l l Selected Old-Growth Reach Pairs  86  Figure 22 Frequency Distribution of Dissimilarity Values for A l l 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)  xi  95  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 unit  1  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.  Channel units consist o f various types o f pools and shallows that are the basic morphological components o f a reach (Hogan and C h u r c h , 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. 1  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 multistep 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).  3  "Managed  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 oldgrowth 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  characteristics for use in this project is explained.  specific  stream channel  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 19  th  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 L L  The total length o f the channel system w i t h i n a basin. The overall m a x i m u m basin length measured from the mouth. The length o f overland flow (map distance from a point on a divide orthogonally to the adjacent stream channel).  c  B E  X P  The belt o f no sheet erosion. The perimeter o f the drainage basin.  c  Num >er measures: The number o f streams o f a given order. U n d e r 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. The bifurcation ratio, ( R b = N / N ) . It is a dimensionless number varying between approximately 3.0 and 5.0 for networks formed in homogenous rocks; it is fairly insensitive to all but the most important structural controls.  N„ R  u  b  u+1  Meaningful areal variables used to define basin morphometry are: The area o f a drainage basin o f a given order. The total area o f a given drainage basin. The drainage density, equal to L / A (initially defined by Horton in 1945). It is found to be closely related to mean stream discharge, mean annual precipitation and sediment y i e l d . The source density, w h i c h is very closely related to drainage density. It is the number o f stream sources per unit area. B o t h D and D are very sensitive to possible map-to-map inconsistencies in the portrayal o f the drainage net. The peak density, equal to the number o f closed hilltop contours per unit area.  A A D  u  D  c  s  s  Dp F  The stream frequency, w h i c h expresses the number o f stream segments o f all orders per unit area (=2N /A). U  The circularity ratio ( = A / A , where A = the area o f a circle having the same perimeter P as the basin). The elongation ratio ( = d / L , where d = the diameter o f a circle o f area A ) .  Rc  C  Re  A  c  B  A  The relief of a basin may be described by: H H  Relief, w h i c h expresses the elevation difference between high and l o w points. This measure was first introduced b y Partsch (1911) and was first used in the E n g l i s h language in 1935 (Mark, 1975). The available relief. This concept was introduced by G l o c k in 1932 and was rephrased b y Johnson in 1933 ( M a r k , 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 . 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 w h i c h local relief is based are o f an appropriate size ( M a r k , 1975).  a  /  H  d  And inally, significant gradient measures include: The m a x i m u m slope o f the ground surface at a given point.  ^max S H/L  The m a x i m u m angle o f a given valley-side slope profile.  HHA  A dimensionless " r e l i e f ratio", actually a generalized gradient measure.  c  B  The slope o f a reach o f a stream channel at a point or averaged over a reach. A dimensionless " r e l i e f ratio", actually a generalized gradient measure. F r o m this and closely related to mean slope is the ruggedness number ( = H * D ) .  H/P  A dimensionless " r e l i e f ratio", actually a generalized gradient measure. Taken from M a r k , (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 N T S Maps Used to Classify and Compare Individual Watersheds Watershed characteristics: Area (km ) Perimeter (km) M e a n basin elevation (m) 2  Ice ( k m ) Lake (km ) 2  2  V a l l e y flat extent ( k m ) M e a n basin gradient (m/m) 2  Relief Elevation R e l i e f Ratio  K  w  Steepland area ( k m ) Shape Factor (km / k m ) 2  Channel characteristics: Drainage density ( k m / k m ) M a x . channel elevation (m) 2  Definitions: The drainage area o f the watershed. The length o f the perimeter o f the watershed. The mean elevation o f the drainage basin. The total ice covered area within the basin. The total lake covered area within the basin. Total area with a gradient less than 7 % and adjacent to the drainage network. The The The The  mean gradient o f the drainage basin. difference between the highest and lowest points in the basin. proportion o f highland and lowland with respect to mean elevation. circularity o f the basin in relation to basin area and basin length.  The total area with a gradient o f greater than 6 0 % .  Definitions: The total channel length divided b y the drainage area. 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) M a i n channel gradient  The lowest point on the channel Change in elevation over the main channel divided by channel length.  Total channel length (km) M a i n channel length ( k m )  The sum o f the length o f all channels in the drainage basin. The length o f the longest channel from its point o f origin to the outlet.  The total number o f first order streams. Magnitude Taken from C h e o n g , (1996) and B i r d , (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 Typel Type 2 Type 3 Type 4 Type 5 Type 6 Type 7  Relatively l o w proportions o f ice and valley flat, high proportions o f lake and steepland H i g h proportions o f ice and lake, l o w proportions o f steepland L o w proportions o f ice and very high proportions o f valley flat L o w proportions o f ice, lake, and steepland and high proportions o f valley flat Quite high proportions o f ice and relatively large amounts o f steepland H i g h proportions o f ice, l o w proportions o f steepland and lakes L o w proportions o f ice and lake, large extents o f steepland and l o w proportions o f valley flat L o w proportions o f ice and lake cover, and relatively large extents o f steepland and valley flat  Type 8 H i g h proportions o f ice and steepland Type 9 L o w proportions o f ice, lake, steepland and valley flat T y p e 10 L o w amounts o f ice, valley flat and steepland, high proportions o f lakes T y p e 11 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,  temperature) and hydrology (e.g., hydrographic analysis, flood frequency  snowfall, 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. summarizes the 3 levels of drainage basin similarity assessment.  14  Table 4  Table 4 Drainage Basin Similarity Assessment Level 2 Level 1 Similar geology based on Similar geology based on Geology 1:50 000 or smaller scale 1:250 000 or 1:50 000 maps maps Similar soils and terrain based on 1:50 000 maps Soils/Surficial Geology  Hydrology  Climate  Vegetation  Similarity  Similar hydrological zones, Similar hydrological zone similar hydrograph characteristics Similar precipitation intensity, total precipitation, snowfall Similar biogeoclimatic zone Similar biogeoclimatic from detailed mapping, zone 1:2 000 000 1:250 000 or 1:50 000 mapping mapping Similar basin type  Similar basin type and dissimilarity within group ranges  Level 3 Detailed bedrock geology analysis based on 1:20 000 maps or better Similar soils, terrain, surficial geology based on detailed 1:20 000 or better scale maps, mass wasting history Hydrograph analysis, flood frequency analysis Similar precipitation intensity, total precipitation, snowfall, frost free days Similarity from biogeclimatic mapping (1:20 000) or vegetative surveys Similar basin type and dissimilarity within group ranges and below critical similarity  Taken from C h e o n g , (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 oldgrowth 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  00-2 O ^  < co u u  "  CO  e  <*3  43  JO  8 U  '»  oo oo. t- oo oo  o t-~o*.woooovor-ro o o o o o o o o  a •V ii JS  d  r—;  •—;  r«-j  ON  t—;  oo d so » n  oo — so  v©  O o  Tfr  m  >n m oo  — — ~* o o o  O o  o\  ro  CN 00  CN —'  0O  l-~r~ocNCNroro("> tN(Nnviiciif)(StN  o\ O CN  cn  0\  co  O  ^  ON  _H  H  CN"— •—' ro  >—'  •—'  H  •—i  ro --^ CN CN ro  ii «  a  u  *  «  CN  O M O N  1*^  o  V© 43  I—(  u  a pfl H-<  as  o  o o  W  CQ "o  k  ii JS U  "« u a a  «  00  o  •3  1  CD  CO  CO  O U u a U  o o o o o o o o o o o o o o o o rorororororororo  o o o o _ o o o o 2 SO SO SO SO ro co ro ro JZ?  A  A  A  A  A  A  A  "o5 Io a o  A  A  A  A  T3  O.  ~  CL>  J3 00 43 CO  n  o  n  oi oi oi oi u>  = -a  I CO CD  -2 CO CD  to  53  U O U U g >  O h r-^ ^  CO  S3  co  o o o o o  c  o.  m  m X U X X oi oi oi oi oi oi  00  §  uuouuauu  A  OH  i  00 T3  &| .2 -S .Br  U  m o CN r CN ro so <o  o  t/3  on  «  o  oo  M ' ! O O > rr 2 t""  r - ro ro  :  CO  >  £  —  CO  CQ  CO CO  CQ  H Pi  u  PH  ^  t  60 O  X  E o  C  "33 •§ c  co  J=  U  b oo  IS S3 O  o  a  •a "E. 3  o o  c D  •co u  &  is  i-i  IS H  -a DH  <; -> w  c  J3  CO  U 13 +H  _o  .S c  « 2  o o  CD  60 O  5  a  oo c o  J3  U  00  o  o  CN  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 N B 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 , it is questionable whether a classification method based on 2  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:  22  It allows emphasis to be  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 2 Morphometric/Hydroclimatic Parameters:  Stage 1 Biogeophysical Filter Parameters:  • • • • • •  S i m i l a r geology based on 1:125 000 maps Similar soils and terrain based on 1:50 000 maps Same physiographic zone S i m i l a r total precipitation Same biogeoclimatic zone from 1:50 000 mapping Same hydrological zone (Church, 1997)  •  •  S i m i l a r mean basin elevation, ice cover, lake area, valley flat area, shape factor and steepland area (here defined as >60% slopes) S i m i l a r drainage density, channel gradient and stream magnitude (here defined as total number o f 1 order channels) Similar hydrograph characteristics st  •  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 Q C I and VI Watersheds Channel Configuration  Sub-basin  Soils  M e a n basin elevation (m)  Ice Cover (km ) 2  Stream L a k e Shape Drainage C o v e r Factor Density Magnitude (km/km ) (km ) 2  2  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 U p p e r M a i n  P P P P P P P P  207 257 213 454 467 484 342 324  0 0 0 0 0 0 0 0  0 0 0 0.3 0 0.3 0 0  3.43 1.45 4.19 0.82 1.12 0.83 1.22 1.80  0.924 0.923 0.999 0.481 0.474 0.408 0.397 0.443  2 1 1 2 1 1 1 5  Uncoupled  Govt. M a i n Govt. U p p e r M a i n Jason Upper Jason L o w e r Carmanah Upper  P P P P P  183 168 373 347 403  0 0 0 0 0  0 0 0 0 0.02  1.72 2.14 0.88 1.19 9.3  0.837 0.896 0.346 0.358 1.56  6 3 1 1 42  [Taken in part from Hogan et al, 1998] Soils  P =  Podzolic soils (mainly ferro-humic podzols).  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 N B 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. Upper has a contributing area  This could be due in part to the fact that Carmanah 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 N B 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 N B 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 N B 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  D C  o  "E  g S  T3 T3  o o  vo  v© cn  3  cn  oil  A  i3  vo </1  g e  CN  e1 -M2  cn  *  OH  cu , cu OH  i4  o o  Ov cn  cn  cn  A  cn  VO cn  ©  CN O  ©  o o  on  o  oo  OH  S  cn f*i  VOl  A I.  •s  u  Of  "a o o c  V WD A  a  o o CN  Ov  VO  o p  00 IT)  oo  VO cn  ©  A  12  .a  Pi  o o  a  A  u ;  VO cn  >* u  CN  cn  II  1/5 CJ  II  II  ^ >- u  CU  PH  o o  es  5i  ov  00  o o  Ov  vo  VO cn  o  "o  ©  A  CN  <o  u  O  03  a u c o N  00  S3  H  o o  <D  s  ia a -3  o  if . co m i  1)  c  IS  i'i  i3  CD CO  OH  IS  >  w  ,s  o o  , 3  00  CO CO  l  2  I  c <u u Q CO PH  s CO  c  .s  ea  CD  X  2 CO  o to  S3 S x: CD  U to  s  3 M  £  a oo  CO  IS  e CD  w  CD  ^  00  -=  00  5 g OPH0  o H3  O .9 CO  m  O  CO  oo  oo _o 'lo  JS OH  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 subbasins. 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%. A l l 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. A l l 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  Channel Configuration  Sub-Basin  Government Creek Government Government Government Government Government  Main Upper M a i n ( U M ) N o r t h Branch ( N B ) N o r t h Branch N o r t h Fork ( N B N F ) North Branch East Fork ( N B E F )  Uncoupled Uncoupled Coupled Coupled Coupled  Inskip Creek Inskip M a i n Inskip N o r t h Branch ( N B ) Inskip South Branch ( S B )  Coupled Coupled Coupled  Gregory Creek Gregory U p p e r M a i n ( U M ) Gregory North Branch ( N B ) Jason Creek Jason L o w e r Jason Upper  Coupled Coupled Uncoupled Uncoupled Uncoupled  Carmanah Creek Carmanah Upper  Table 10 Final Selection of Managed Sub-Basins for Project Database Watershed Mosquito Creek R i l e y Creek Tarundl Peel  Sub-Basin Mosquito M a i n Mosquito Upper Riley Lower Riley Middle Tarundl Peel  30  Channel Configuration Uncoupled Uncoupled Uncoupled Coupled Uncoupled Uncoupled  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 11.2%; Wood-Smith and Buffington (1996) define the riffle bed slope at 2-4%. likely explained by differences in overall stream gradient.  Nevertheless, there is  characteristic organization of channel units downstream following the downstream variation in gradient (see Figure 3).  32  This is  systematic  Stream 1 ^ ^ s t e p - p o o l Stream 2 ^  r a p i d - p o o l ^ ^ ^ ^ _^  Figure 3  riffle-pool  Downstream Organization of Stream Channel Units. While 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, w h i c h is c o m m o n i n Coastal B C watersheds. There are other watershed types (e.g., in Plateau areas o f B C , headwater streams are generally flat and steepen downstream where channels are incised). ( M . C h u r c h , 2000, personal c o m m u n . , D . H o g a n , 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 Channel U n i t Type  Channel Unit Characteristics  Pool  Topographic depressions with heterogeneous substratum .  Glide  Heterogeneous substratum. C o m m o n at pool-riffle and pool-rapid breaks.  Riffle  Primarily uniformly distributed alluvial gravel and cobbles. Relative roughness generally >1. C o m m o n in larger channels dominated by a sequence o f alternating bars with intervening crossovers and average channel gradient less than 1 degree.  >  Gentle  Slope  Rapid  1  Characterized by bed elements (most c o m m o n l y boulders and cobbles) arranged into irregular ribs oriented roughly perpendicular to the channel and exposed at l o w flow. Relative roughness is >1 at the ribs and <1 in the secondary pools.  Steep  <  Stone L i n e W h e n clusters (interacting individual clasts o f similar size) ramify into extended linear or arcuate features (Church, 1998). They form in the "partial transport regime" found i n headward channels, where the largest material present often is lag material delivered from overbank that rarely moves. L o g Step  Near-vertical to vertical steps typically created b y w o o d y debris oriented transverse to the channel.  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 typically steeper than boulder cascades. Relative 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.  T w o major pool types can be delineated: backwater pools and scour pools (Bisson et al., 1982; Sullivan, 1986). Backwater pools are caused by downstream obstructions and are generally shallow due to accumulation o f sediment. Scour pools are caused by f l o w convergence over or around an obstruction (Church, 1992) and are relatively deep. Variants o f both scour pools and backwater pools exist. 1  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 poolto-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  L W D jams.  Montgomery et al. (1995) found that, in forest channels, pool spacing  depended on L W D 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 L W D 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 ( L W D ) 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 L W D pieces and their arrangement or position are commonly examined (Hogan, 1986; Ralph et al., 1994; Montgomery et al., 1995). The mean volume of individual L W D pieces and mean spacing of L W D 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.). (Dsosurface/Dsosubsuface)  This surface-to-subsurface ratio  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  D95 face Sur  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 S t r e a m c h a n n e l characteristics Channel unit frequency Channel unit length P o o l spacing (scaled as multiples o f W ) b  Depth V a r i a b i l i t y (m/m) W i d t h V a r i a b i l i t y (m/m) L W D Spacing (scaled as multiples o f W ) b  L W D Volume (m /m ) 2  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 L W D jams, steps and individual pieces were inventoried every bankfull width. The volume of all in-channel L W D was determined by visual estimation based on the procedure established by Hogan (1989) (see Table 13):  Table ] 3 L W D classification Rank 1 2 3 4  D i a m e t e r (m) <0.1 0.1-0.3 0.4-0.7 0.7-1.2  5 >1.2 Based on Hogan, (1989)  Length (m) 1-5 5-10 10-15 15-20  # o f pieces <2 2-3 4-7 7-12  >20  >12  Orientation  //  1  Using the ranking scheme outlined in Table 13, the diameter, length, number of pieces, and orientation of L W D were rapidly estimated during the longitudinal profile survey.  Log jams (multiple, interacting L W D 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 L W D 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 1 2  Description of Span Completely crosses channel Incomplete, 0.75-1 W spanned, scour around end b  0.5 W spanned 3 0.25 W spanned 4 <0.25 W spanned 5 Based on H o g a n , (1989) b  b  b  Table 15 Classification of LWD Jam Integrity Rank 1 2 3 4  Description of Integrity V e r y solid, compact, b i g pieces, no rot, anchored S o l i d , compact, anchored Moderate, less compact (spaces), rot  Weak, poor anchor, rot V e r y weak, small pieces, no anchor 5 Based on H o g a n , (1989)  Table 16 Classification of LWD Jam Height Description of Height A b o v e local bank height V* -1 bank height  Rank 1 2 3  Vi bank height  4  Vi bank height < Vi bank height 5 Based on H o g a n , (1989)  Table 17 Classification of LWD Jam Age Rank 1 2 3 4  Description of Age V e r y recent, new trees, no nurse trees (<2 yrs) Recent (2-10 yrs old) Moderate (10-20 yrs old) Moderate to o l d (20-30 yrs old)  O l d (30-50 yrs old) 5 v. o l d , nurse trees, no bark (>50 yrs old) 5+ Based on H o g a n , (1989)  Table 18 Classification of LWD Jam Location Code LB  General Location  Right bank In middle  RB M B  Description of Location Left bank  In-channel  A l o n g channel  A t bend Bedrock knob  R  W a d or tree stump Trees (standing)  W T Based on H o g a n , (1989)  41  Table 19 Classification of L W D Jam Shape Code 1  /L II V  LWD j a m shape Perpendicular Diagonal Parallel A r c h e d (apex downstream)  A r c h e d (apex upstream) A Based on Hogan, (1989)  In addition to the longitudinal profile and L W D inventory, a scaled diagram was assembled which included the position of channel units, position and orientation of L W D , 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 is considered b  a conservative measure, and is based on the knowledge that in free-formed pool-riffle reaches, pool-to-pool spacing averages 5-7 W (Leopold et al., 1964). This ensures a b  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  Small Woody Debris  Large Woody Debris  O  P  o  o  o o  x  x  Boulders  Cobbles  Pebbles  Sand  Vegetation  Cross-Section Location 4= 200m Bank Erosion  Channel Unit Boundary  Riffle  t  <-10m->  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 L W D clustering could be  43  investigated more thoroughly. Hogan and Bird (1998) have determined that the largest L W D 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 R E A 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  0  1  4 00  1  800  "  I  1200  1600  I  200 0  i  240 0  2800  Figure 6 Mean Runoff Volume and Catchment Size. 1 pixel = 30 m . Taken from Wood et al., (1988). 2  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 Government M a i n Government U M Government N B  Length (m) 800 1860 793  Length (Wb) 25 101 39 82  Government N B E F  871 672  Inskip M a i n Inskip N B Inskip S B  649 270 700  28  Gregory U M  2165  100  Gregory N B Jason L o w e r  420  31 65 34 26  Government N B N F  Jason U p p e r Carmanah Upper  1205 720 1000  46,  40 18 48  Table 21 M a n a g e d Stream C h a n n e l R e a c h  Reach Mosquito M a i n Mosquito Upper Riley Lower Riley Middle Peel Tarundl  3.4.2 S t r e a m C h a n n e l S u b - R e a c h  Lengths  Length (m) 2022  Length (Wb) 70  943 5325 1481 1200 2910  29 210 43 69 176  Selection Procedure  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)  0.96  Section o f R e a c h U s e d to Calculate Variance 0m - 20m  Variance Of |Differences| 0.00378  1.00  0m - 40m  0.02567  1.03  0m - 60m  0.06584  7.09  0.89  0 m - 80m  0.12889  7.86  7.23  0.63  0 m - 100m  0.12622  7.75  7.37  0.38  0 m - 120m  0.13029  0m - 140m  0.14190  Horizontal Distance (m) 0  Actual Elevation (m) 7.92  Estimated Elevation (m) 6.96  |Difference| (Depth Deviation)  8  8.01  7.01  9  8.05  7.02  20  7.98  40 60 80  7.61  7.51  0.10  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) « c _o |  0.4 |  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 L W D 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 N B ,  Inskip SB and Government N B NF, plots for coupled reaches also stabilized at variance values of -0.05.  As Inskip N B 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 in Length b  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):  Selected Sub-Reach Inskip SB: 902m - 1399m  Entire Reach Inskip SB: 649m- 1399m .2 0.4  0.4  0.3  0.3  •3 0.2  • 0.2  CL  u  Q  o 0.1 0.0 600  0.1 750  900  0.0 850  1050 1200 1350 1500  Horizontal Distance (m)  Figure 10  950  1050 1150 1250 1350 1450 Horizontal Distance (m)  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 subreach 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 N B , Government N B EF, Inskip Main, Gregory N B , 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 N B NF, Government U M , Gregory U M and Jason Lower). A l l reaches in this group, apart from Government N B 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 finetuned based on visual examination.  51  Table 23 Old-Growth Sub-Reach Sections to be Analyzed  Government U M  8 0 0 m - 1740m 1000m-1925m 1225m-2140m 1405m-2340m 1600m-2520m 1780m-2660m  Result Further analysis required Discard Discard Discard Further analysis required Further analysis required Discard Discard Discard Discard  Gregory U M  3701m-4886m 4011m-5086m 4222m - 5306m 4446m - 5526m 4664m - 5746m 4886m - 5866m 0 m - 920m 200m - 1205m  Discard Discard Discard Discard Further analysis required Further analysis required Further analysis required Discard  R e a c h Sections T o B e A n a l y z e d 1593m-2143m Government N B N F 1723m-2243m 1823m-2343m 1923m-2464m  Jason L o w e r  The first section of Government N B N F (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 N B N F  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: 1000m - 1925m  Government Upper Main: 800m - 1740m  I  0.4 r  1 0.4  0.3 \  0.3  Q  fo.2 ID  0.2  r  0.1  0.1| '§ o.o — |> 700  E  1  900  1100 1300 1500 1700 Horizontal Distance (m)  Figure 12  1900  0.0 900  >  1100  1300 1500 1700 1900 Horizontal Distance (m)  2100  Variance Plots for Government U M : sections 800m - 1740m and 1000m - 1925m  Selected Sub-Reach Government Upper Main: 800m - 1925m 0.4  i  > 0.3 Q  •3  & 0.2 Q o  2 0.1 S3 >  0.0 600 1  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 0.4,  . . .  •S  Gregory Upper Main: 4886m - 5866m .  CO  I  Q  0.3  s >  4600  4900 5200 5500 Horizontal Distance (m)  g o.o L — — > 4900 5100  5800  1  5300 5500 5700 5900 Horizontal Distance (m)  Figure 14 Variance Plots for Gregory U M : Sections 4664m - 5746m and 4886m - 5866m  Selected Sub-Reach Gregory Upper Main: 4786m - 5866m 0.4 0.3 • Q  •B CL  0.2 •  a o.i  o.o  4600  4800  5000  5200  5400  5600  5800  6000  Horizontal Distance (m)  Figure 15  Variance Plot: Selected Sub-Reach Length for Gregory U M  54  ' 6100  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 —  0  200  400  Upstream Limit  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  of  unnecessary  calculations.  (Several  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 subreach.  56  Table 24 Managed Sub-Reach Sections to be Analyzed Reach  Sections T o B e Analyzed Om - 1500m 500m-2022m  Result  Riley Lower  0 m - 1275m 2 5 0 m - 1525m 5 0 0 m - 1775m 1525m-2800m 2000m-3275m 2550m-3825m 3825m-5100m 4000m - 5325m  Tarundl  0 m - 825m 4 0 5 m - 1410m 8 2 5 m - 1640m 1640m-2460m  Further analysis Further analysis Further analysis Further analysis Discard Discard Discard Discard Further analysis Further analysis Discard Discard  Mosquito M a i n  Peel  0m - 800m  Further analysis required Discard required required required required  required required  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  Riley Lower: 250m - 1525m  Riley Lower: 0m - 1275m  V30.4  0.4  C  o  >  Q  0.3  0.3  0.2  Q . 0.2 u O 'o i> 0.1  §  0.1  0.0  >  0  200  400  600  800  0.0 200 400 600 800 1000 1200 1400 1600 1800 Horizontal Distance (m)  1000 1200 1400  Horizontal Distance (m)  Riley Lower: 1525m-2800m  Riley Lower: 500m - 1775m 0.4 0.3 0.2 0.1  J 0.0 > 1400 1600 1800 2000 2200 2400 2600 2800 3000 Horizontal Distance (m)  0.0 400 600 800 1000 1200 1400 1600 1800 2000 Horizontal Distance (m)  Figure 17  Variance Plots for Riley Lower: sections 0m - 1275m, 250m 1525m, 500m - 1775m, and 1525m - 2800m.  Selected Sub-Reach Riley Lower 2: 1525m-2800m  Selected Sub-Reach Riley Lower 1: 0m - 1525m § 0.4  I  0.4 «0.3  0.3  •3  ! • 0.2  8-0.2 Q o 0.1  1 0.0  |  0  400 800 1200 Horizontal Distance (m)  Figure 18  0.0 1400 1600 1800 2000 2200 2400 2600 2800 3000  1600  Horizontal Distance (m)  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: 405m - 1410m  Tarundl: 0m - 825m 0.4  0.4  0.3 0.2 0.1 200 400 600 Horizontal Distance (m)  Figure 19  0.0 '300  800  500  700 900 1100 1300 1500 Horizontal Distance (m)  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).  Selected Sub-Reach Peel: 0m-800m 3 0.4 o  'S  0 3  Q % 0.2  S o.i  I  0.0  200  400  600  800  Horizontal Distance (m)  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 Initial Reach  Watershed Reach Government Creek Government Government Government Government Government  Main UM NB NB NF NB EF  Inskip Creek Inskip M a i n Inskip N B Inskip S B Gregory Creek Gregory U M Gregory N B Jason Creek Jason L o w e r Jason U p p e r Carmanah Creek Carmanah U p p e r  Om - 800m 800m - 2660m 8 0 0 m - 1593m 1593m-2464m 1593m-2265m 0 m - 649m 649m-919m  Selected S u b Reach 0m-800m 800m-1925m 800m-1593m 1593m-1963m 1593m-2265m  Sub-Reach Length ( W ) 25 60 b  40 34  Final Variance 0.05 0.05 0.06 0.05 0.06 0.05  40 29 0-649m Discard - only 18 W long & irregular b  34  0.14  3701m-5866m  902m-1399m 4786m-5866m  50  5866m - 6286m 0 m - 1205m 1 2 0 5 m - 1925m 0 m - 1000m  5866-6286m 0m-860m 1205m-1925m 0 m - 1000m  28 46 33 26  0.11 0.03 0.07 0.05 0.07  6 4 9 m - 1399m  Table 26 Managed Sub-Reach Selection Watershed Reach Mosquito Creek Mosquito M a i n Mosquito Upper R i l e y Creek R i l e y M i d d l e Riley Lower 1  Initial Reach 0m - 2022m 2022m-2965m 5325m - 6840m 0 m - 5325m  Selected S u b Reach 0 m - 1500m  Sub-Reach Length ( W „ ) 51  2022m-2965m 29 Discard - d i d not stabilize 0 m - 1525m  50  0.1 0.06 0.2 0.055  0 m - 5325m  1525m-2800m  Tarundl Creek Tarundl  0m-2910m  405m-1410m  50 62  Peel Creek Peel  0 m - 1200m  0 m - 800m  46  Riley Lower 2  60  Final Variance 0.1 0.04  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 dissimilarity of 1  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,  Eq. 1  Ewklxjk-Xjkl'  Ew  k  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  1/A  Eq.2  k  Stream channel similarity can also be used to compare stream channels. H o w e v e r , as the purpose o f this project is to determine the range o f variability that exists i n stream channels, the interest is focused primarily on how different, or dissimilar, various reaches are. Therefore dissimilarity is the preferable measure w h i c h w i l l be explored in this project. 1  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 )  Eq. 3  2  ik  (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. numerator in Equation 3 is squared, the denominator is not.  While the  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:  dijk  2  Xjk  Xjk  Oi  a.  Eq.4  64  Which reduces to: dijk = 2  (Xik-x )  Eq. 5  2  ik  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 Units  Information Type  # o f specific units / # o f all units m/m  Ratio Ratio  m/m m/m m/m  Ratio Ratio  Parameter Channel Unit Frequency Channel U n i t Length P o o l Spacing Depth Variability W i d t h Variability L W D Spacing L W D Volume Relative Roughness  Ratio Ratio Ratio Ratio  m/m  nvVm  3  m/m  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 )  Eq. 6  2  (Orfffle)  Where i represents Government Main, j represents Government Upper Main, x represents the riffle frequency value for a given reach, and a iffl is the standard deviation r  e  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:  dij-toto/  =  2  2  2  2  V2  (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). question,  In order to investigate this  the dissimilarity testing procedure was carried out for two cases:  measures standardized, and (2) all measures standardized.  (1) no  With no standardization,  L W D 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 dy :, wDspadng was significantly greater than the dissimilarity value of any other stream 2  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, L W D spacing values (scaled as multiples of Wb) fall in the range of 1 to 10.  In other words, the  dominance of the L W D 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 scalereferenced 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 Units  Parameter Channel U n i t Frequency Channel U n i t Length P o o l Spacing  # o f specific units / # o f all units m /m m /m m /m m /m  Depth Variability W i d t h Variability L W D Spacing  m /m m /m m /m M  L W D Volume Relative Roughness Mean Bankfull Width  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, oldgrowth 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 N B 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 N B N F are slightly higher.  The most problematic old-growth reach is Gregory N B . 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 L W D volume data or D95 sediment data exist for this reach. The L W D spacing data are also questionable as only two L W D jams are found in the reach. Therefore the L W D spacing value is based on one distance value, not an average like the other reaches which have multiple L W D jams.  70  I  (-1  CtH  o  vo  O  •a  2?  CD 43  60  43  3  '% O  a S  CD  43  co  CO 43  •  1^3  Q  pa Z & o  >  _ CD d>  PH - J PH _1 CD CD 5H OH PH  .S '  St  00  CT 8  CD CD  OO 0 0 4^ 4 *  C  2 '(2 4*i 44  u u u u 5 3 3 5 0H OH PH PH  oo  CD LH  o  4=  oo 4=  CO  1(2  g.: oo  T3  u  C  CD  %s  OH  1  <;  3 i_ CD L L ,  43  CO  CO  on 00 §  > kH  co  —  3  _o o  •« al a CD  i-  o  CD  Lt,  CD  00 s  t-  gl  VO CN  t,  PH  8 c 81 a 3  CD  —1  3 CD 3  _i  CT  gi  CD  LLL.  L.  PH -*-»  'S  CD  p 13  •P2^ CQ  Z u es  cu  o Z  o  P £ 43  C0  u Ii  o .„  cn O — i i fD  PQ oo  PH  3 3  co  -3  o  sis  olo  &  ^ »  CO  3  _2  "° Jp-3 "  — I I  CD  PH3  a oO  "33 s s  1,2  3 cr CO  o .S  CD  p  «  «*  M co P CO U CO co 2 ^ PH •9 .2 CD ta CO "c0 Z >> - p *a co .3 s •  PH  CD  1 i  s .2 «  <*H 00  L«  IP.  c!  &• s"i3 "JH  c  £  S  CO  U~i  3  60  o CD  to  CD  "O  00 4df  *i  -J  , p  1  §  CH  p £  O —1  o  CD _3  3  CD -a CO  3  p £  P  CfH  p  •ti al c CD  3  CD  CD  CD Xt  o  U  »  'S  p  CD  PH CD  M  00 CO  .5 _  HH  « S CO 2 O OO  00  u  00  P  T3 3 43  u  HH  -J 00  ^  H "  cd  5.2.1  All Reach Pair Combinations  Table 30 lists the summary dissimilarity results using all possible reach pair combinations:  Table 30 Dissimilarity Results - A l l Possible Reach Pair Combinations Reach Pairs G o v M a i n - G o v Upper M a i n G o v M a i n - Jason L o w e r Gov Gov Gov Gov  Main Main Main Main  - Jason Upper - Carmanah - Greg N B - Inskip M a i n  Gov Gov Gov Gov Gov Gov Gov  M a i n - Inskip S B M a i n - G r e g Upper M a i n Main - Gov N B E F Main - Gov N B N F Main - Gov N B U p p e r - Jason L o w e r U p p e r - Jason Upper  Gov Gov Gov Gov  Upper Upper Upper Upper  Dissimilarity Reach Pairs Dissimilarity 7.25 Jason Upper - Inskip S B 4.49 Jason Upper - Greg Upper M a i n 4.51 3.89 Jason Upper - G o v N B E F 4.05 4.40 6.33 Jason Upper - G o v N B N F 4.47 6.08 4.24 10.15 4.12 6.15 7.23 4.81 3.50 2.84  Jason Upper - G o v N B Carmanah - G r e g N B Carmanah - Inskip M a i n Carmanah - Inskip S B Carmanah - Greg Upper M a i n Carmanah - G o v N B E F Carmanah - G o v N B N F Carmanah - G o v N B G r e g N B - Inskip M a i n Greg N B - Inskip S B Greg N B - Greg U p p e r M a i n Greg N B - Gov N B E F Greg N B - G o v N B N F  3.78 6.65 6.33 10.92 5.57 7.93 8.90 7.05 6.44  Greg N B - G o v N B Inskip M a i n - Inskip S B  6.43 8.02  10.17 4.88 6.45 7.13  G o v U p p e r - Greg Upper M a i n Gov U p p e r - Gov N B E F  5.62 6.39 4.32 8.19 4.33 4.02  G o v Upper - G o v N B N F  6.30  Inskip M a i n - G r e g U p p e r M a i n  4.40  G o v Upper - G o v N B  2.73 5.16 6.37  Inskip M a i n - G o v N B E F  4.73  Inskip M a i n - G o v N B N F Inskip M a i n - G o v N B  5.90 4.35  6.55 5.18  Inskip S B - Greg Upper M a i n Inskip S B - G o v N B E F Inskip S B - G o v N B N F  9.25 6.96 8.32  Inskip S B - G o v N B Greg Upper M a i n - G o v N B E F  4.64  G r e g Upper M a i n - G o v N B N F G r e g Upper M a i n - G o v N B  5.50 4.10  -  Carmanah Greg N B Inskip M a i n Inskip S B  Jason L o w e r - Jason Upper Jason L o w e r - Carmanah Jason L o w e r - G r e g N B Jason L o w e r - Inskip M a i n Jason L o w e r - Inskip S B Jason L o w e r - G r e g U p p e r M a i n Jason L o w e r - G o v N B E F Jason L o w e r - G o v N B N F Jason L o w e r - G o v N B Jason Upper - Carmanah Jason U p p e r - G r e g N B Jason Upper - Inskip M a i n  10.22 3.54 4.89 6.14 3.20  8.56  5.36  Gov N B E F - Gov N B N F  5.29  5.96  Gov N B E F - Gov N B Gov N B N F - G o v N B  2.74 5.18  3.26  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: L W D volume (d LWDvoi = 11.65), mean bankfull width (d b 2  2  W  (d  2 D/rf  =  9.51), and relative roughness  =9.11).  The high value for d LWDvoi is a reflection of the L W D volume values for Inskip SB 2  and Carmanah (see Table 29). Inskip SB has the highest L W D volume in the old-growth database while Carmanah has the lowest. This is likely related to scale, as the d wb value 2  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 d o/d is also likely related to scale. 2  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 km ), whereas Carmanah is the largest. So, processes mediated by 2  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  characteristics: log step length (d i i 2  s en  for two  other  = 12.15) and cascade length ( d  stream channel  2 cte  „ = 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 ff= ri  frequency (d kn c  =  12.83), width variability (d  w v a r  = 11.18), and cascade length  11.03).  The high values for d ff and d / „ reflect the fact that Inskip SB has the highest 2  2  ri  c  e  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 d  2 wvar  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 value (Wb = 39.2). This high Wb value is b  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 N B 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 L W D 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 N B N F 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 N B N F has the lowest mean Wb value and contributing area.  The most similar reach pairs are Gov Upper - Gov N B and Gov N B EF - Gov NB: Gov U p p e r - G o v 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 N B (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 in „ = 1.93). g  relatively low.  e  However, this value is still  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 N B E F 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). A n analysis of variance determined that no significant difference exists between these two average dissimilarity values [F „ (4.07) > F(1.47); accept H ]. cr  0  Other reach pairs with significantly low dissimilarities (within the top 10% most similar) include: Gov Upper - Jason Upper, Jason Lower - Gov N B , Jason Upper Inskip Main, and Gov Upper - Jason Lower. These four reach pairs are not situated adjacent to each other as Gov Upper - Gov N B and Gov N B E F - Gov N B 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 As  Uncoupled Reach Pair Combinations 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 Pairs G o v M a i n - G o v Upper M a i n G o v M a i n - Jason L o w e r G o v M a i n - Jason U p p e r G o v M a i n - Carmanah G o v Upper M a i n - Jason L o w e r  Reach Pairs Dissimilarity G o v Upper M a i n - Jason Upper 6.15 G o v Upper M a i n - Carmanah 4.98  Dissimilarity 4.33 6.35  6.46  Jason L o w e r - Jason U p p e r  6.55  5.35  Jason L o w e r - Carmanah Jason Upper - Carmanah  6.16 7.17  3.96  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  stream channel characteristics: riffle length (d riflen  2  = 6.81), relative roughness (d o/d~  6.56), and cascade length (d / „ = 5.81). The high value of d o/d reflects the fact that 2  2  c  e  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. 2  2  Lower — Jason Upper, cascade (d i •= 5.81) and pool (d pooiien c en  For Jason  2  =  5.59, d oifreq  =  p0  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 d o/d is 0.09 for Gov Upper - Jason Lower. In 2  contrast, d /y</is 3.22 for Gov Upper - Jason Upper. Looking at Table 29 it is apparent 2  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). apparent in the overall ranking of the reach pairs.  79  Again, only slight changes are  Table 32 Dissimilarity Results - Coupled Reach Pair Combinations Reach Pairs G r e g N B - Inskip M a i n G r e g N B - Inskip S B G r e g N B - G r e g Upper M a i n Greg N B - G o v N B E F Greg N B - Gov N B N F Greg N B - G o v N B Inskip M a i n - Inskip S B Inskip M a i n - G r e g Upper M a i n Inskip M a i n - G o v N B E F Inskip M a i n - G o v N B N F Inskip M a i n - G o v N B  Dissimilarity Reach Pairs Inskip S B - G r e g U p p e r M a i n 6.10 Inskip S B - G o v N B E F 8.72  6.63 7.93 8.15 4.39  4.99 6.05 6.70 6.37  Inskip S B Inskip S B G r e g Upper Greg Upper  7.05 4.34  G r e g Upper M a i n - G o v N B  3.87  G o v N B E F - Gov N B N F Gov N B E F - Gov N B Gov N B N F - Gov N B  4.61  4.90 6.14  Gov N B N F Gov N B Main - Gov N B E F Main - Gov N B N F  Dissimilarity 8.41  5.43  2.75 4.94  4.74  The least similar reach pair combinations all involve Inskip SB. They include (in order of decreasing dissimilarity): Greg N B - Inskip SB, Inskip SB - Greg Upper, and Inskip SB - Gov N B . 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 L W D volume values, and low riffle frequency values.  It is curious that Greg N B shows up in the most dissimilar reach pair, as Greg N B has no data available on D/d or L W D volume. involving Greg NB, no values exist for d ^ o r  In other words, for any reach pair  & LWDVOI2  If the missing data for Greg N B  were available, the total dissimilarity value for Greg N B - Inskip SB would be even greater.  The  most  similar reach pair combinations include (in order of increasing  dissimilarity): Gov NB EF - Gov N B , Greg Upper - Gov N B , and Inskip Main - Greg Upper. As stated previously, the similarity between Gov N B E F and Gov N B 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 N B 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 L W D 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 N B 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 Pairs G o v M a i n - G o v Upper M a i n G o v M a i n - Jason L o w e r G o v M a i n - Jason Upper Gov Gov Gov Gov Gov Gov Gov  Main - Gov N B E F Main - Gov N B N F Main - Gov N B Upper - Jason L o w e r Upper - Jason Upper Upper - G o v N B E F Upper - G o v N B N F  G o v Upper - G o v N B  Dissimilarity 6.44 5.40 6.01 8.46 8.80 6.69 4.74 3.94 5.27 7.30  Jason Jason Jason Jason Jason  Reach Pairs L o w e r - Jason Upper Lower - Gov N B E F Lower - Gov N B N F Lower - Gov N B Upper - G o v N B E F  Dissimilarity 6.89 6.64 6.92 4.16 5.44  Jason Upper - G o v N B N F  7.71  Jason Upper Gov N B E F Gov N B E F Gov N B N F -  5.16 6.51 4.05 6.00  Gov N B Gov N B N F Gov N B Gov N B  3.33  The most dissimilar reaches (in order of decreasing dissimilarity) are: Gov Main Gov N B NF, and Gov Main - Gov N B EF.  The high total dissimilarity for Gov Main - Gov N B N F can be largely attributed to the mean bankfull width characteristic (d wt = 11 -08). With the exclusion of Carmanah, 2  Gov Main now has the largest Wb value (32.3 m).  In contrast Gov N B N F has the  smallest Wb value (10.6 m). Gov Main is an uncoupled, outlet reach while Gov N B N F 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 N B NF), greater L W D volume values (as is the case with Gov N B NF), and differences in channel unit characteristics. In the Gov Main - Gov N B N F case, pool length frequency is the most influential characteristic. Gov Main has a pool length frequency of 0.7, nearly twice that of Gov N B NF.  82  A similar situation applies to the Gov Main - Gov N B E F case. Similar to Gov NB NF, Gov N B E F is a smaller, coupled reach. Differences in mean Wb, L W D 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 N B .  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, L W D 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 1tank Correlation Test for Selected Old-Growth Reach Pairs Rank Rank Reach Pairs Dissimilarity Geographic Proximity (km) (Dissimilarity) (Geographic Proximity) 6.44 Gov Main - Gov Upper Main 2.00 12 7 Gov Main - Jason Lower 5.40 10.60 8 17 Gov Main - Jason Upper 6.01 11 21 12.40 Gov Main - Gov NB EF 8.46 20 11 2.95 Gov Main - Gov NB NF 8.80 1.70 21 5 Gov Main - Gov NB 6.69 1 15 0.50 Gov Upper - Jason Lower 4.74 5 12 8.55 Gov Upper - Jason Upper 3.94 10.40 2 16 Gov Upper - Gov NB EF 5.27 7 9 2.20 7.30 Gov Upper - Gov NB NF 18 6 1.95 Gov Upper - Gov NB 3.33 1 4 1.60 Jason Lower - Jason Upper 6.89 2.10 16 8 6.64 Jason Lower - Gov NB EF 14 13 9.15 Jason Lower - Gov NB NF 6.92 14 10.00 17 Jason Lower - Gov NB 4.16 4 10.15 15 5.44 Jason Upper - Gov NB EF 18 10.75 9 Jason Upper - Gov NB NF 7.71 11.65 19 19 Jason Upper - Gov NB 5.16 20 11.95 6 6.51 Gov NB EF - Gov NB NF 2 1.30 13 4.05 Gov NB EF - Gov NB 3 10 2.45 6.00 Gov NB NF - Gov NB 1.30 10 3  Table 35 Spearman Rank Order Correlations Dissimilarity & Geographic Proximity  Number of Reach Pairs 21  p-level 0.410  Spearman R -0.190  t(n-2) -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  (B)  (A) All Old-Growth Reach Pair Combinations  15  All Uncoupled Old-Growth Reach Pair Combinations  g 10  £ 6 1 O <g 4  I  Z  A  5  V*  1^  m  nflnn^ r*-  Ul  co  U>  a*  I'f  *-  2 0  UI  o  U>  I -  ^ - c u ^ i n f - c o c * * -  Dissimilarity (upper limits)  o  Dissimilarity (upper limits)  (C)  (D) All Selected Old-Growth Reach Pair Combinations  All Coupled Old-Growth Reach Pair Combinations 8 w  o  6  «> 6  o  o o  O Z 2 Ul  n, n v  Z III  V  V*  CO  <7>  *~  2 0  O  Dissimilarity (upper limits)  Figure 21  A  n Dissimilarity (upper limits)  Frequency Distribution of Dissimilarity Values:  (A) A l l Old-Growth Reach Pairs, (B) A l l Uncoupled Old-Growth Reach Pairs, (C) A l l Coupled O l d - G r o w t h Reach P a i r s , a n d ( D ) A l l selected O l d - G r o w t h R e a c h P a i r s  These data were imported i n to Statistica® i n order to evaluate the fit o f the observed data to a variety o f hypothesized distributions.  The ' A l l Old-Growth' data did not  deviate significantly from the standard lognormal distribution (Chi-Square = 1.90, d f = 5, p = 863) (see Figure 22).  This distribution fitting analysis was not performed on any o f 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)  |  2.2  2.9  3.6  4.3  5.0  5.7  6.4  7.1  7.8  8.5  9.2  Expected  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 o f 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 o f the managed reaches had stone line features delimited.  89  CD  3  3  PH  •3 ca CJ  CA  <»  JS  ca o  B  la  CU  CD  pf!  43,  •8 ?  S  >&  t/5  CL> ft ft  • ; o t ca ; ft' t/3 s  88 ©  U  —  "3 3  i>  08  o  I?  eg CD  a cu  CD  1  p £  58  PH  -S  .s CO  3  O  M  •a  H  PH PH  H  P 13 43  «  &  '3  O 60 O S . .  PH  XI  60 3 CD  IS a <"  3  CB  ta  PH  ^ P H  o  4>  H  Is -H  CD  CT  1  —  -a  O 3  CD [JL,  Q P  CD  43  1  p £  H  CD X) ca u  CD _3 "ca >  3 In CD LL,  a  « H  CD  S3  <  60  a  ca  XI 3  0)  op C o CD  u  PH  o £ c DH  P  u 3 CD 3 IT CD  PH  3 in CD tt,  HJ 5 3  r-  a  2 P  PH  CN  3  P 13 3  43  o  U  ca cu  Pi  CD XI O  4-»  •3  ft CD «J  c/3  ca CJ 00 co ca  s o  D H J U CD 3  II II  43 (/J  U  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 Mos Mos Mos Mos  Reach Pairs M a i n - M o s Upper Main - Riley Lower 1 Main - Riley Lower 2 M a i n - Tarundl  M o s M a i n - Peel M o s Upper - Riley Lower 1 Mos Upper - Riley Lower 2  Dissimilarity 5.91 4.57 4.57 4.91 5.42 8.30 7.11  Dissimilarity 6.39 4.02 6.51 6.39  Reach Pairs M o s Upper - Peel Riley Lower 1 - Riley Lower 2 R i l e y L o w e r 1 - Tarundl R i l e y L o w e r 1 - Peel  4.90 6.19 4.62  R i l e y L o w e r 2 - Tarundl R i l e y L o w e r 2 - Peel Tarundl - Peel  5.70  M o s U p p e r - Tarundl  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  characteristics: pool spacing (d 6.70).  poo  2  i  sp  = 9.58) and log step length frequency (d  „ =  fafe  The high total dissimilarity value for Mos Upper - Riley Lower 2 can be  attributed to three stream channel characteristics: width variability (d  2 wvar  = 7.98), depth  variability ( d ^ = 7.72), and large woody debris spacing (d wDs = 7.58). The high 2  2  L  values for d  2  and d ^ 2  w v a r  P  are related to the fact that Mosquito Upper has the largest  width and depth variability values in the managed database (see Table 36). 91  The high  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 o f 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 o f increasing dissimilarity) are: Riley Lower 1 - Riley Lower 2 and M o s 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 M o s M a i n and Riley Lower 2, one o f the •  2  key discriminators appears to be L W D spacing (d  =  LWDS  P  5.67). A s stated previously,  the L W D spacing value for Riley Lower 2 is relatively high and, more importantly, questionable. Four o f 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 o f logging disturbance, no attempt has been made to create a 'selected, managed' reach pair group. I f these factors were controlled, only two o f the reaches (Mos M a i n and M o s Upper) would remain in the database. W i t h 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 Mos Mos Mos Mos Mos Mos Mos  Main - Gov Main M a i n - G o v Upper M a i n - Jason L o w e r M a i n - Jason U p p e r Upper - G o v M a i n Upper - G o v Upper Upper - Jason L o w e r  Reach Pairs  Dissimilarity  Riley Lower 2 - Gov Main R i l e y L o w e r 2 - G o v Upper R i l e y L o w e r 2 - Jason L o w e r R i l e y L o w e r 2 - Jason U p p e r Tarundl - G o v M a i n Tarundl - G o v Upper Tarundl - Jason L o w e r  4.47 6.00 5.89 6.08 8.36 7.02  9.04  M o s Upper - Jason Upper Riley Lower 1 - G o v Main  6.66 4.17  R i l e y L o w e r 1 - G o v Upper R i l e y L o w e r 1 - Jason L o w e r R i l e y L o w e r 1 - Jason Upper  5.51  Tarundl - Jason Upper Peel Peel Peel Peel  4.11 6.19  -  Gov Main G o v Upper Jason L o w e r Jason Upper  Dissimilarity 5.48 4.93 4.65 6.27 5.84 4.56 6.00 6.23 6.19 4.93 6.44 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 oldgrowth 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  Selected old-growth uncoupled  M e a n Dissimilarity Value 5.40  B  Managed uncoupled  5.73  Standard Deviation 1.13 1.13  C  Selected old-growth (uncoupled) vs. Managed (uncoupled)  5.83  1.20  Group  Reach P a i r Combination Groups  A  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  Box & Whisker Plots, Groups A-C 7.6 7.0 6.4  '5  I-  5 8  to  S  5.2 _L_ ±Std.Dev. 4.6 4.0  Figure 24  Box  I Group A  Group B  Group C  and W h i s k e r Plot for G r o u p s A - C :  I ±Std. Err. •  Mean  A = Selected O l d - G r o w t h  (uncoupled), B = M a n a g e d ( u n c o u p l e d ) , a n d 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 o f 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 Pairs Jason Upper - Jason L o w e r Gov Gov Gov Gov Gov  M a i n - Jason Upper M a i n - G o v Upper M a i n - Jason L o w e r Upper - Jason Upper Upper - Jason L o w e r  Dissimilarity 6.83 6.69 6.59 5.33 4.55 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 Pairs Jason L o w e r - M o s U p p e r G o v M a i n - M o s Upper G o v Upper - M o s Upper Jason Upper - M o s Upper Jason U p p e r - Jason L o w e r G o v M a i n - Jason Upper G o v M a i n - G o v Upper G o v Upper - Jason Upper G o v M a i n - Jason L o w e r  Dissimilarity  G o v Upper - Jason L o w e r  3.42  8.09 7.31 6.42 6.21 5.91 5.65 5.06 4.37 4.24  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 o f 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 o f 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. dissimilarity  values may  Those stream channel characteristics exhibiting l o w 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 I n this study a m e t h o d o f q u a n t i f y i n g stream c h a n n e l v a r i a b i l i t y and d e f i n i n g undesirable states (thereby  q u a n t i f y i n g stream c h a n n e l impact) has been d e v e l o p e d .  T h r e e p r e l i m i n a r y steps were i n v o l v e d i n d e v e l o p i n g this m e t h o d :  (1) s e l e c t i o n o f s u i t a b l y s i m i l a r drainage basins (2) s e l e c t i o n o f suitable stream c h a n n e l characteristics; and (3) s e l e c t i o n o f suitable stream reaches  In a l l o f these steps important issues were i d e n t i f i e d and addressed.  T h e selection o f s u i t a b l y s i m i l a r drainage basins was based o n a m e t h o d from  C h e o n g (1992,  thorough,  systematic  1996).  C h e o n g ' s b a s i n c l a s s i f i c a t i o n procedure  approach  m o r p h o m e t r i c characteristcs.  available,  does  configuration.  not  appear  both  is the  biogeophysical  most and  H o w e v e r , it is not w e l l suited to the s m a l l e r - s i z e d sub-  basins w h i c h are the focus o f this project. procedure  incorporating  adapted  to  I n particular, the C h e o n g b a s i n c l a s s i f i c a t i o n  effectively  characterize  differences  in  channel  It is not clear that c h a n n e l c o n f i g u r a t i o n ( w h i c h m a y be reach specific)  c a n be r e s o l v e d f r o m maps or s y s t e m a t i c a l l y correlated w i t h v a l l e y flat area.  The  underlying problem  with  utilizing  a  rather r i g o r o u s  basin classification  procedure is that it i s v e r y d i f f i c u l t to get a sizable sample group w h i c h meets a l l the assessment c r i t e r i a (e.g. T a b l e 6). T h e absence o f f o r m a l b a s i n c l a s s i f i c a t i o n procedures i n m a n y c o m p a r a t i v e stream c h a n n e l studies i s l i k e l y related to this issue. I n this study, a r e l a t i v e l y 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, L W D 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 oldgrowth' 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. A n 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 ofAmerica Bulletin, 75: 767-774. Cheong, A . (1992). Quantifying drainage basin comparisons within a knowledgebased systemframework.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): 14131430. 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 steppedbed morphology in high-gradient streams, Western Cascades, Oregon. Geological Society ofAmerica 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 ofBritish 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 ofAmerica 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 ofAmerica Bulletin, 109(5): 596611. 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 ofFisheries Management, 15: 49-53. Schumm, S. A . (1977). The fluvial system: New York. John Wiley and Sons, 338 pp. Statistica © 1 9 9 9 by StatSoft Inc. Strahler, A . N . (1964). Quantitative geomorphology of drainage basins and channel networks. In Chow, V . T., editor, Handbook ofApplied 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 (m ) / channel length (m) Mean channel cross-sectional area = mean bankfull x mean channel width (m) depth (m) 3  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)  200  400  600  800  Variance Plot: Government Upper Main (Entire Reach)  1000  Horizontal Distance (m)  1000 2000 Horizontal Distance (m)  Variance Plot: Jason Lower (Entire Reach)  Variance Plot: Jason Upper (Entire Reach)  500 1000 Horizontal Distance (m)  1500  1000  Variance Plot: Carmanah (Entire Reach)  I  0 4  'S •g 0.3 a | 0.2  500  1000  1500  Horizontal Distance (m)  110  3000  1200 1400 1600 1800 2000 Horizontal Distance (m)  Coupled, Old-Growth Reaches Variance Plot: Gregory NB (Entire Reach)  Variance Plot: Inskip Main (Entire Reach) 0.4  -I  1 " I ' 03  Q  02  5900  6000 6100 6200 6300 Horizontal Distance (m)  o J 0  6400  200 400 600 Horizontal Distance (m)  200  400 600 800 Horizontal Distance (m)  500 1000 Horizontal Distance (m)  1000  s o  0.4  -I  •  .  .-  0.3 0.2 0.1 -  ^  04000  6000  8000  1000  Horizontal Distance (m)  Variance Plot: Government NB NF (Entire Reach)  1000 2000 Horizontal Distance (m)  1500  Variance Plot: Government NB EF (Entire Reach)  Variance Plot: Gregory Upper Main (Entire Reach)  2000  800  Variance Plot: Inskip SB (Entire Reach)  Variance Plot: Inskip NB (Entire Reach)  Variance o!F Depth Deviati  5800  p  Variance  o  —  -  1500 2000 Horizontal Distance (m)  2500  Variance Plot: Government NB (Entire Reach)  500  3000  1000  1500  Horizontal Distance (m) 111  2000  Managed Reaches  Variance Plot: Mosquito Main* (Entire Reach)  500  1000  1500  2000  Variance Plot: Mosquito Upper* (Entire Reach)  2500  1000  Horizontal Distance (m)  2000  3000  4000  Horizontal Distance (m)  Variance Plot: Riley Lower* (Entire Reach)  Variance Plot: Peel* (Entire Reach) 0.4 0.3 0.2 0.1  \  500  1000  2000  1500  Variance Plot: Riley Middle (Coupled) (Entire Reach)  2000  4000  6000  4000  6000  Horizontal Distance (m)  Horizontal Distance (m)  Variance Plot: Tarundl* (Entire Reach)  1000  8000  2000  3000  Horizontal Distance (m)  Horizontal Distance (m)  * Uncoupled  112  4000  

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