Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The spatial variation of bed material texture in coupled basins on the Queen Charlotte Islands Rice, Stephen Philip 1990

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

Item Metadata


831-UBC_1990_A6_7 R52.pdf [ 8.8MB ]
JSON: 831-1.0098567.json
JSON-LD: 831-1.0098567-ld.json
RDF/XML (Pretty): 831-1.0098567-rdf.xml
RDF/JSON: 831-1.0098567-rdf.json
Turtle: 831-1.0098567-turtle.txt
N-Triples: 831-1.0098567-rdf-ntriples.txt
Original Record: 831-1.0098567-source.json
Full Text

Full Text

T H E SPATIAL VARIATION OF BED MATERIAL T E X T U R E IN C O U P L E D BASINS ON T H E Q U E E N C H A R L O T T E ISLANDS by Stephen Philip Rice B.Sc. (Hons.), Oxford University, 1988 A THESIS SUBMITTED IN PARTIAL F U L F I L L M E N T OF T H E R E Q U R E M E N T S FOR T H E D E G R E E OF MASTER 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 COLUMBIA October 1990 © Stephen Philip Rice, 1990 In presenting this thesis in partial fulfilment 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 G f i D G r R A P H V The University of British Columbia Vancouver, Canada Date Cfcrog,F_re 11 WRO DE-6 (2/88) ii ABSTRACT Whether one is interested in the geomorphology, hydrology, or ecology of a river, the nature of the bed sediments is of major importance. Despite a long history of interest our ability to predict local grain size is poor, which is unfortunate given the labour and costs associated with bed material sampling. A preliminary model of sediment texture variation at the drainage basin scale, which makes a fundamental distinction between coupled and uncoupled hillslope-channel units, is presented. It is hypothesised that grain size variations in strongly coupled rivers are unstructured as a result of overbank colluvial inputs and special storage elements. These preclude the development of the systematic downstream structure commonly associated with fluvial sorting and abrasion processes in uncoupled channels. This conjecture is assessed empirically using data collected in two rivers on the Queen Charlotte Islands. It is found that distance alone does not explain changes in grain size, and that local variations are dominated by large organic debris jams. Impermeable jams are associated with upstream fining and downstream coarsening but the effect diminishes as the jams become more permeable, often with age. Jam placement is random, but frequent, and consequently at the drainage basin scale, grain size changes dramatically and unpredictably over very short distances. No deterministic structure is apparent. Further analysis reveals that the observed variations of surface median grain size and Fredle index are best regarded as stochastic phenomena. Sampling criteria are then determined which enable the accurate characterisation of such variation, once a stream has been classified by land use and position relative to hillslopes. iii CONTENTS ABSTRACT ii LIST OF TABLES vii LIST OF FIGURES ix ACKNOWLEDGEMENTS xi 1. INTRODUCTION 1 1.1 Overview 1 1.2 Significance of Bed Material 2 1.2.1 Hydraulics 2 1.2.2 Geomorphology 3 1.2.3 Fisheries Ecology 4 1.3 The Need For a Predictive Model 6 1.4 Previous Work 6 1.4.1 Downstream changes in bed material size and heterogeneity 6 1.4.2 The "Sternberg" Model of downstream diminution 7 1.4.3 Local scale variability 9 1.4.4 Longitudinal grain size discontinuities 11 1.4.5 Downstream changes in sediment heterogeneity 12 1.5 Model Development 13 1.5.1 Link subsystems 14 1.5.2 Source subsystems in headward areas 17 1.5.3 Transfer subsystems downstream 17 1.5.4 Confluence mixing 18 iv 1.6 Source Subsystems in the Pacific Northwest 19 1.7 Study Objectives 20 2. STUDY AREA AND METHODOLOGY 23 2.1 Queen Charlotte Islands 23 2.2 Study Design 25 2.2.1 Operationalizing the objectives 25 2.2.2 Study creeks 27 2.3 Longitudinal Survey 29 2.3.1 Site selection 29 2.3.2 Sampling procedure 31 2.3.3 Laboratory analysis 34 2.3.4 Within site variance 34 2.4 High Resolution Photographic Surveys 42 2.4.1 Field technique 44 2.4.2 Calibration of particle size and number 45 2.5 Log Jam Studies 48 2.5.1 Bed material sampling 49 2.5.2 Permeability measurements 50 3. EVALUATION OF THE STERNBERG MODEL 53 3.1 Longitudinal Survey Data 53 3.1.1 Construction of synthetic bulk samples 53 3.1.2 Results and analysis 56 3.2 Photographic Surveys 62 3.3 Summary 68 V 4. THE EFFECT OF LOG JAMS ON BED MATERIAL TEXTURE 69 4.1 A Priori Ideas 69 4.2 Study Jams 71 4.3 Confirmation of Differential Permeability 73 4.3.1 Tracer recovery 74 4.3.2 Downstream tracers 74 4.3.3 Upstream tracers 77 4.3.4 Scour chains 77 4.4 Bed Material Characteristics 78 4.4.1 Surface and subsurface samples 78 4.4.2 Photographic survey data 84 4.5 Summary 89 5. ALTERNATIVE PREDICTIVE MODELS 90 5.1 Grain Size and Jam Proximity ,. 90 5.2 Grain Size, Sample Position in Relation to Log Jams and Distance Downstream 100 5.3 Summary 103 6. OPERATIONAL RECOMMENDATIONS 104 6.1 The Effect of Land Use and Confinement 104 6.1.1 The Fredle Index 104 6.1.2 Median surface grain size 108 6.2 A Test of Randomness: Runs Up and Down 110 6.2.1 Rationale of the test 110 6.2.2 Analysis 110 vi 6.3 Sampling Recommendations 113 6.3.1 Normality 113 6.3.2 Sample size 114 6.3.3 Sampling strategy 115 6.4 Summary 117 7. CONCLUSION 119 R E F E R E N C E S 121 APPENDIX 1 128 vii List of Tables 1. Previous studies of downstream diminution 10 2. Study basin characteristics 28 3. Replicate sample characteristics for GLS 1 90 38 4. Replicate sample characteristics for GLS REP 39 5. Bulk sample within-site variances 41 6. Photographic survey calibration data 46 7. Surface and subsurface at GLS 1 90 55 8. Longitudinal survey data 57 9. Within-link variance of surface materials 60 10. Surface grain size estimates a) Riley Creek 63 b) Gregory Creek 64. 11. Within-reach analysis of variance for surface D 5 0 67 12. Study jam characteristics 72 13. Tracer stone movements (1989-1990) 76 14. Grain size and proximity to log jams a) Riley Creek 91 b) Gregory Creek 92 15. Logjam characteristics of the photo reaches a) Riley Creek 94 b) Gregory Creek 95 16. Fredle Indices and various within group statistics 106 17. Characteristics of surface D 5 0 in the photo reaches Runs test results (low frequency longitudinal survey data) Runs test results (high frequency photographic data) Recommended sample sizes List of Figures 1. Link type and textural variation 15 2. Study area 24 3. Longitudinal surveys 30 4. Sketch map of replicate sampling sites 36 5. Replicate bulk sample distributions GLS 1 90 37 6. Replicate bulk sample distributions REP 90 37 7. Photographic surveys and study jams 43 8. Photo calibration curve 47 9. Longitudinal survey data, grain size variations 58 a) Riley surface D 5 0 b) Riley subsurface D 5 0 c) Gregory surface D5Q 10. Photographic survey data, grain size variations 65 a) Lower Riley b) Lower Gregory c) Upper Gregory 11. Schematic of grain size changes in the active channel in the vicinity of log jams 70 12. Tracer particle displacements 75 13. Surface texture in the vicinity of log jams 79 14. Survey photographs of the bed surface at Jam U, Upper Riley 80 15. Sediments in the vicinity of Jam K K , Upper Riley 81 16. Subsurface texture in the vicinity of log jams 83 X 17. Surface grain size variations a) Riley Creek photo reaches 85 b) Gregory Creek photo reaches 86 18. Sediments associated with Jam GG, Upper Gregory 87 19. Sediments in the vicinity of Jam O, Lower Gregory 88 20. Grain size in the vicinity of jams classified by span and integrity 96 21. Grain size in the vicinity of jams classified by sediment storage 97 22. Grain size in the vicinity of jams classified by number of channels 98 23. Median surface grain size at similar downstream jam environments 103 xi A C K N O W L E D G E M E N T S I would like to thank my supervisory committee: Dr. Michael Church, for teaching me so much and helping me to learn so much more, and Dr. Olav Slaymaker, for his calmness in the face of my adversity. Very special thanks go to Dan Hogan, who introduced me to the joys of field work on the Queen Charlotte Islands, and whose advice, encouragement, and beer are much appreciated. None of this would have been possible without the Jambusters (David Ramsey, Craig Nistor, Steve Bird, and Alan Paige) who worked incredibly hard, and with great spirit in the field. Special thankyous go to Sue Fairburn and Phil Owens; the former for her patience, encouragement and for listening to me talk about gravel, the latter for suffering with me throughout the summer. Also to my parents, for their sacrafices and support over the years. This work was supported by the contract fund of the Federal and Provincial Fish Forestry Interaction Program, and thanks are due to the administrative supervisor, Stephen Chatwin, for his interest and support. 1 1. INTRODUCTION 1.1 Overview The size distribution of clastic material in the bed of a stream channel varies in space. Despite a long history of interest, ability to predict local bed material texture (grain size distribution) remains negligible. This is regrettable given the significance of bed material texture geomorphologically, hydraulically, and ecologically. This study investigates the nature of the variation in the headward part of the drainage basin, where in the absence of a buffering alluvial floodplain the stream channel is subject to direct inputs of colluvial material from adjacent hillslopes. In an attempt to improve our conception of how bed material texture varies throughout the drainage basin a model of textural variation is presented. The distinction between coupled geomorphic areas (strong hillslope-channel interaction) and uncoupled geomorphic areas (weak hillslope-channel interaction) is the basis of this model. In the Pacific Northwest strong coupling dominates basins which support valuable timber and fish harvesting industries. Understanding the variation of bed material texture in these areas is important because logging activities can affect texture, while anadromous fish habitat is partly dependent upon it. The formulation of a predictive model of textural variation in the strongly coupled basins of the region is of substantial significance in the context of integrated resource management. It would simplify habitat mapping, improve the validity of monitoring programmes and provide information on the diversity of spawning substrate. 2 The study utilises data collected on the Queen Charlotte Islands for two basic objectives. The first is to assess empirically the basic assumptions of that part of the model pertaining to strongly coupled areas. The second is to develop an effective operational model for predictive use by resource managers in the Pacific Northwest. This regional application of the general model is being undertaken as part of the Fish Forestry Interaction Programme (Poulin, 1984). 1.2 The significance of Bed Material Texture The justification of this study lies in the significance of bed material texture hydraulically, geomorphologically, and ecologically. 1.2.1. Hydraulics Water flowing down a channel at a uniform speed is opposed by an equal frictional resistance. A flow resistance equation describes the balance between the propelling force of the flow and the resistance at the boundary. The Chezy, Manning, and Darcy-Weisbach equations are examples. They can be used to predict flow velocity and, in turn, discharge. It has long been established that the mean flow velocity is some function of the channel slope and hydraulic radius, and that the constant of proportionality is a function of the flow resistance. The crux of flow velocity prediction is evaluating this coefficient. Subjective comparisons can be made with descriptions of channels for which coefficients have been determined empirically. Alternatively, simple relations between resistance coefficients and channel parameters can be used. Bed material characteristics are often encountered in the tabulated 3 observations, and several relations have been proposed which utilise grain size (cf review in Bray, 1979). Recently, semi-empirical evaluations based on fluid mechanical principles and boundary layer theory have been sought, and have focussed on the resistance offered by bed material (Bathurst, 1985). It is clear then that some idea of local bed material characteristics is needed in order to estimate flow velocity using a flow resistance equation, irrespective of how the resistance coefficient is determined. Such equations are useful when velocity measurements are not available, for example when working from air photos, during paleohydraulic reconstructions, or when the additional cost of a measurement program is simply not warranted. 1.2.2. Geomorphology Channel form is the product of sediment transport by flowing water. Our understanding of transport processes determines our ability to explain observed channel forms, and to predict morphological responses to natural or anthropogenic disturbances. Sediment transport equations attempt to relate sediment transport rates to flow conditions and the properties of the sediment stored in the bed. Church and Hassan (pers. comm.) identify three sets of sedimentological factors which influence the entrainment, transport and deposition of a particle: grain parameters; bed structure; and bed morphology. Grain size has long been the basic sediment property used in transport equations. In some cases a single representative size is used while in others data of sedimentation of gradation are required. Structure and bed morphology have been acknowledged as important sediment transport controls relatively recently (Einstein, 1950; Laronne and Carson, 1976; 4 Fenton and Abbott, 1977; Parker and Klingeman, 1982; Brayshaw, 1985; Church, 1985; Komar, 1987; Wolcott, 1990). They are difficult to quantify, and our understanding of their influence is incomplete. However, initial work such as that by Komar and Li (1986) suggests that grain heterogeneity is an important factor in this respect. Consequently sediment transport equations may eventually utilise measurements of grain sorting. It is apparent then that local grain size characteristics are key independent parameters of sediment transport equations. The estimates of rates which these provide are used in numerous management problems, such as forecasting rates of scour and fill around proposed engineering structures. 1.2.3. Fisheries Ecology The size distribution of stream bed sediments affects fish and other stream biota. Forest harvesting activities (road construction, yarding and slash disposal) can change texture detrimentally. In particular the accelerated production and mobilisation of fine material can increase the proportion of fine sediments in the channel (Platts and Megahan, 1975; Scrivener and Brownlee, 1989; Cederholm et al., 1981; Sheridan et al., 1984; Tripp and Poulin, 1986), although this is not always the case (Burns, 1972; Moring, 1975; Sheridan et al., 1984). The effects of increased fines on salmonid productivity can generally be regarded as detrimental, particularly if disturbances are prolonged and accompanied by other habitat disruption (Everest et al., 1987). Fine sediments can reduce productivity by reducing gravel permeability (McNeil and Ahnell,1964). This reduces water flow through the substrate and limits oxygen availability and waste removal (Cooper, 1965). In addition fines may 5 trap alevins during emergence (Bjornn,1968) and abrade incubating eggs (Philips, 1971). The study of the interactions between bed material texture and aquatic productivity has focussed on the fines issue, and the proportion of fines in the bed material of a creek has long been used as an index of its fisheries value. Everest et al. (1987) suggest that this is a somewhat limited approach, and that a more holistic view of "sediment" is required. The overall texture and the quantity of sediment are important, not just the fines content. It is a geometric fact that permeability is dependent upon the composition of the entire substrate, not just the fine fraction of it. The use of alternative indices which take this into account has been encouraged by several researchers (Platts et al. 1979; Lotspeich and Everest, 1981). In any case it is apparent that spawning and emerging salmon can mitigate the adverse effect which fines have on permeability (Burner, 1951; McNeil and Ahnell, 1964; Bams, 1969). Furthermore, habitat quality is related to bed material in a variety of ways, not just by permeability. The general hydraulic characteristics of a reach, and the diversity and abundance of the microinvertebrate benthic community which is the fishes' primary food source, are both dependent on sediment texture (Nuttal, 1972; NCASI, 1986). The volume of sediment is also important. Too much sediment can lead to channel dewatering whereas too little sediment limits spawning and feeding opportunities (Everest and Meehan, 1981). Thus, because of its importance as a spawning resource, a food source and a determinant of flow hydraulics, bed material texture is an important 6 aspect of fish habitat. Local texture is a central concern of fisheries biologists making habitat inventories. 1.3. The need for a predictive model Whether one is interested in the hydrology, geomorphology, or ecology of a river, the nature of the bed sediments is of major importance. Bed material texture is a key parameter governing sediment transport, fish habitat, and flow velocity. A knowledge of local bed material texture is therefore often required by resource managers and engineers. Due to the arduous nature of sampling bed material representatively, especially in gravel bed streams, obtaining such information is time-consuming and costly. Furthermore low flow conditions are required, and this restricts sampling operations to particular periods of the year. A predictive model of bed material texture at the drainage basin scale is therefore desirable. Routine management needs which cannot justify the necessary field effort would benefit significantly from such a model. 1.4 Previous work 1.4.1 Downstream Changes in Bed Material Size and Heterogeneity In general bed sediments tend to get finer and more uniform as one moves downstream. Downstream diminution and increasing homogeneity are associated with sorting and abrasion processes (Krumbein, 1937; Knighton, 1980). Sorting is the process whereby smaller grains are preferentially entrained and transported, travelling greater distances than larger material in a given time interval. According to classic sediment transport theory this is to be expected as a consequence of their lower 7 inertia (Shields, 1936). Abrasion is a summary term for a suite of processes, such as grinding and chipping, which reduce the size of an individual clast either during transport or in situ. Abrasion tank experiments (e.g. Daubree, 1879; Kuenen, 1956; Adams, 1978) simulate abrasion in transport. The rates of wear obtained are consistently less than rates of grain size diminution observed in the field. Some workers claim that such discrepancies are due to the dominance of sorting processes in real channels (Bradley, 1972; Brewer and Lewin, 1990). Others claim that the discrepancies are due to the unrealistic nature of the simulated abrasion mechanism, not the dominance of sorting processes (Barrell, 1925; Schumm and Stevens, 1973). The ongoing debate surrounding particle entrainment has a direct bearing on this issue. Gary Parker and collaborators (e.g. Parker and Klingeman 1982) claim that particle entrainment and distance of transport are virtually independent of particle inertia (size). If this holds then longitudinal sorting based on the classic assumption of size-dependent entrainment and transport must be of little significance. Other particle entrainment studies suggest that transport is size-dependent (Komar, 1987) and that in turn sorting is a major contributor of downstream fining. Additional evidence that sorting is important is found in the field observations of Brierly and Hicken (1985) and MacPherson (1971). 1.4.2. The "Sternberg" Model of Downstream Diminution As yet no experimental method has been developed to clarify the relative importance of the two processes. The exact mechanics of grain size diminution and improved sorting therefore remain unclear and any ? 8 functional relation of size to distance must be treated as a generalisation (Krumbein, 1937). The classic model of abrasion suggested by Sternberg (1875) supposes that reduction in grain size per unit distance is proportional to particle size [equation 1]. Thus, as a result of abrasion processes, grain size decreases exponentially with distance [equation 2]: -8D/6L = a-p [1] D = D 0 e " a l L [2] where, D is grain size L is distance is a coefficient of abrasion D Q is initial grain size at L = 0 Knighton (1984) pointed out that this result is similar to those which model downstream changes in grain size as a consequence of sorting by size-selective transport (Rana et al., 1973; Troutman, 1980). These suppose that the difference in distance travelled, for a unit reduction in grain size due to sorting, is inversely proportional to grain size [equation 3]. Hence, as a result of size-selective sorting, grain size decreases exponentially in a downstream direction [equation 4]: 5L/6D = l/(-a2D). [3] D = D 0 e " a 2 L . [4] where, a 2 is a coefficient of sorting. It is apparent from the similarity of equations 2 and 4 that abrasion and sorting are both expected to produce an exponential decline in grain size with distance downstream. In practice then, a single empirically determined 9 coefficient of diminution can be used to reflect the undifferentiated (and indistinguishable) effects of both abrasion and sorting (Knighton, 1980) : D = D 0 e-( al + a 2 ) L = Dne"^ [5] where a is an empirical coefficient of the combined effect of both sorting and abrasion. For the sake of convenience this general model of exponential decline [equation 5] shall be referred to as the Sternberg model. The Sternberg model has been applied to the longitudinal variations of bed material size observed in a variety of natural channels (Table 1). In general the model is consistent with the overall trend of the observations but scatter about the relation is high. It is apparent from the coefficients of determination given in the table that distance alone is insufficient to explain grain size behaviour in natural channels. Attempts to account for the observed scatter have identified several additional factors (Table 1). 1.4.3. Local Scale Variability One source of scatter which is not considered explicitly in many of the studies is that over the surface of bars and between bar, riffle and pool morphologies. This variability is a ubiquitous feature of all rivers and is qualitatively well understood (Bluck, 1976). When considering texture at larger scales it should be removed from the analysis. Church and Kellerhals (1978) suggest stratifying samples sedimentologically in order to eliminate this local scale variability. All samples along the study reach should be taken from similar sedimentological environments, such as bar heads. The consistent use of such a sampling unit TABLE 1 : PREVIOUS STUDIES OF DOWNSTREAM DIMINUTION Authoris) Sternberg (1875)* Berrell (1925) Plumley (1948) Yatsu (1955)+ Miller (1958) + Brush(1961)+ MacPherson (1971) Bradley etal (1972) Church and Kellerhals (1978) Knighton (1980) Shaw and Kellerhals (1982i + Brierly and Hickin (1985) Dawson (1988) Brewer and Lewin (1990) Ichim and Radoane (1990) Desloges (1990) Study Area Upper Rhine (Basel - Mannheim) Length of Channel Studied Rapid Creek, Black Hills. S. Dakota 30 miles Battle Creek, Bear Butte Creek, " Watarase River, Central Japan Kiso River, " " Pecos River, New Mexico Shaver Creek. Pennsylvania Standing Stone Creek, " Sixmile Creek, " Two O'Clock Creek, Alberta Knik River, Alaska Peace River, British Columbia River Noe, Cheshire. U.K. Bow-South Saskatchewan, Alberta North Saskatchewan, " " Squamish River, British Columbia Sunwapta River, Alberta River Severn, Wales. U.K. River Severn, " " River Dyfi, " Siret River, Romania Bella Coola River, British Columbia 18 miles 21 miles 60 km 5 7 km 34 miles 20 miles 30 miles 10 miles 5 miles 24 miles 20 km 1400 km 1100 km 60 km 12 km 66 km 66 km 37 km Tertural Parameter Surface DM Bulk DM Bulk DM Bulk DM ? D50 ? D50 Surface DM Surface DM Surface DM Surface DM Surface DM Surface D96 Surface DM Surface D50 Local Sampling Unit- Number of Data Sternberg Model: D = Doe'1  Points PQ (intercept) a (elope) aL D60 D60 Surface D50 Surface D50 Surface DM Bulk DM Surface DM Surface /bulk D50 Bulk D50 Terrace exposures 10 Terrace exposures 6 Terrace exposures 7 12 12 12 10 7 22 57 ? ? Straight Reaches Riffles Riffles Riffles Coarsest area(s) of bar Bar heads Bar heads (usually) Coarsest area of bar Bar heads Bar heads Bar heads Bar heads Exposure closest to thalweg Coarsest area of bar 16 34 34 10 10 15 33.67 20.1 17.0 35.3 2467.2 503.9 200.9 136.3 100.3 61.4 4.5 (inches) 197.2 601.8 68.6 57.8 170.0 194.4 85.0 43.8 46.5 -0.0026 -0.0394 -0.0860 -0.1074 -0.2634 -0.1788 -0.1205 -0.0609 + 0.0405 -0.4933 -0.0435 -0.0034 -0.0417 -0.0042 -0.0047 -0.0397 -0.2499 -0.0057 -0.0034 0.67 0.87 0.65 0.97 0.77 0.69 0.78 0.64 0.24 0.24 0.18 Additional Factors Identified None Local channel gradient Gravel-Sand discontinuity Tributary inputs Changing bedrock lithology Tributary inputs Bedrock Lithology None Tributary inputs Glacial drift outcrop Lateral sampling position on sandur 0.25 Tributary inputs Tributary inputs Changes in bedrock lithology 0.68 0.56 Tributary inputs Gravel-Sand discontinuity 0.79 Channel Planform Tributary inputs Tributary fan morphology Tributary inputs Reworking Quaternary deposits 0.36 Tributary inputs Gravel-sand discontinuity 0.16 Tributary inputs : contains additional, similar observations for other rivers in the same region D50 ; Median b-axis • DM : Mean b-axis # : Original data of Sternberg converted from number of clasts per cubic D foot to average grain weight, W (assuming 20% porosity and density i_, of 2.6 gem'3) by Berrell, and then to b-axis in mm, D, according to DQ W= 1.41M0-6D (Church, McLean and Wolcott, 1987). R2 Calculations of the cited authors (otherwise the values given are the calculations of the present author). Grain size (in mm unless otherwise stated) Distance downstream (units in column 3) Grain size at L=0 Coefficient of determination calculated for the transformed data (L against InD) 11 significantly reduces overall variation (Wolman, 1954;Church and Kellerhals 1978; Nordin et al. 1980). However, it cannot eliminate it altogether since within-unit variability may still be large (Church and Kellerhals, 1978; Dawson, 1982). 1.4.4. Longitudinal Grain Size Discontinuities In some of the work described in table 1 a sampling unit is identified. It is apparent from these studies that local scale variability is insufficient to account for the scatter around the Sternberg model. Non-random, large-scale structure is still apparent. In particular discontinuities in an underlying exponential decline are frequently observed. These vertical displacements of the Sternberg curve have been traced to sand-gravel discontinuities (Yatsu, 1954 et seq.) and tributary inputs (Miller, 1958 et seq.). The latter are the more commonly cited. For example, Dawson (1988) found a significant increase in grain size at the confluence of one of the four tributaries within his study reach. By considering the reaches upstream and downstream separately his ability to explain observed variations in grain size using the Sternberg model was greatly improved. Using his figure 4 (Dawson 1988, p605) as a source I obtained r^ values of 0.73 and 0.61 for the downstream and upstream reaches respectively, but a value of only 0.38 for the whole data set. This example indicates that as a model of sorting and abrasion processes the Sternberg model is relatively powerful and that, as Knighton (1984) suggests, inputs of distinct material at tributary junctions are superimposed on the systematic modification of main stem sediments by sorting and abrasion processes. 12 Miller (1958) found that tributary inputs resulted in coarsening in 7 of the 15 cases he studied, and fining in 4. Overall he found that mean particle size below the confluence is the average of the mean particle sizes in the two upstream channels. This work and subsequent observations suggest that the magnitude and direction of individual discontinuities is variable and in some cases negligible. Knighton (1980) suggested that the magnitude of any given change is related to the relative sizes of the mainstem and the tributary, and to their lithological differences. Beyond Knighton's identification of relative size and lithological differences as controls on the magnitude of discontinuities, little attempt has been made to investigate these relations more thoroughly. It is apparent that confluence discontinuites must ultimately be related to the relative bed material transport from the confluent channels, and the comparative texture of those materials. 1.4.5. Downstream Changes in Sediment Heterogeneity The heterogeneity of bed sediments is measured in terms of a sorting coefficient. This is a measure of dispersion which reflects the variability of sediment size in a sample. It is generally assumed that preferential sorting mechanisms lead to improved sorting; that is, to reduced sorting coefficients and increasing homogeneity in a downstream direction. Most of the studies in table 1 do not comment on changes in sorting. Of those which do only MacPherson (1971) found such an improvement. Others find no trend at all (Miller 1958 ; Brush, 1961) and Ichim and Radoane (1990) report a very slight increase in the sorting coefficient downstream. These differences probably reflect differences in the nature of the tributary inputs along the reaches studied. Several rivers enter the Siret 13 (Ichim and Radoane, 1990) toward the lower end of the study reach and are associated with substantial increases in the sorting coefficient. These dominate the overall trend but between confluences sorting coefficients decrease steadily. Dawson (1988) found steady downstream improvements in sorting upstream and downstream of the major tributary input, but a significant discontinuity at the confluence. Thus it is apparent that a general downstream trend may be disrupted by tributary inputs to produce a discontinuous change similar to that for particle size. Knighton (1980) has proposed that, where tributary entry follows some regular pattern, pseudo-periodic variations in sorting are observed. 1.5 Model Development It is apparent that tributaries have a systematic effect on particle size and sorting, a clear understanding of which will significantly improve our ability to predict local bed material texture (Knighton, 1980). The crux of this problem is the ability to explain and predict the magnitude and direction of the discontinuities. The implications of understanding the tributary effect go beyond the ability to predict textural changes along a main-stem channel. If tributaries are a source of variation then local texture is intimately linked to drainage network topology. In turn an understanding of the tributary effect opens up the possibility of understanding basin-wide variations largely on the basis of cartographically derived information. In the course of examining the issue a priori it became apparent that tributary inputs are just one type of sediment input which a successful 14 model must recognise. A successful model depends upon an appreciation of the true nature of sediment inputs to the fluvial system. With the exception of Dawson (1988) the applications of the Sternberg model in Table 1 represent a failure in this respect. The Sternberg model imposes a single notional point, LQ , from which sediment originates and is transported downstream in the absence of any additional sediment inputs. The real nature of sediment supply is ignored. In this context a new approach to the problem of basin-wide textural variation is outlined below. The model is only in its formative stages. The examination and application of one part of the model are the primary objectives of this thesis. 1.5.1. Link Subsystems The drainage basin consists of interdependent hillslope and channel components. The channel component can be subdivided into progressively smaller networks and ultimately described as a set of channel reach subsystems linked hierarchically to form the complete system. Considering the drainage basin in this way allows one to isolate within-reach variation and between-reach variation and so study downstream trends and tributary effects independently. A link is here defined as a portion of channel connecting adjacent network junctions (i.e. a channel link in network geometry). Each link can be conceptualised as a discrete subsystem which consists of various sediment and water inputs, storage elements, and sediment and water outputs. Storage of sediment can be either short-term or long-term. The short-term storage component consists of the active bed material and bar deposits. The Input • from confluence mixing lluvlal transport and channelised 'debits Hows Short Term Storage 4 • bed and bars lo depth ol mean annual scour Long Term Storage • bed below depth ol mean annual scour rapid mass movement end creep Output • lo confluence mixing lloodplaln development « N (0 .'.7 lloodplaln .-.-.'l •yW Transfer Subsystems ^conlluence -SOURCE-REACH -TRANSFER -REACH Distance Input Irom confluence mixing Hillsbpe Colluvium creep and rap/d mast movement Long Term Storage • lloodplaln deep scour lluvial tianspott Hlllslope Colluvium Short Term Storage • bed and bars lo depth ol mean annual scour Long Term Storage • bed below depth ol mean annual scour • vegetated islands bank erosion overbank deposition and point bar co/ortrsafJon Long Term Storage • floodplain Output • lo confluence mixing S3 c-t-< tt> f» S3 c < to p 5" S3 On 16 long-term component consists of the deep bed material, the floodplain adjacent to the channel, and any vegetated islands. The nature of sediment input to individual links varies as a function of the position of the link within the basin Church(1983). Two geomorphically distinct link types can be defined and it is suggested that the nature of the bed material variations in the two types of link is very different (Figure 1). 17 1.5.2. Source subsystems in head ward areas The headward areas of a drainage basin supply sediment to zones of transport and deposition downstream (Schumm, 1977). In these headward areas there is strong coupling between the hillslopes and the channel (Church, 1983). That is, in the absence of a buffering alluvial floodplain material is transferred directly from the adjacent hillslopes to the channel by creep and episodic mass-movement processes. Channel links in these areas therefore receive fresh colluvial material along their entire length. These overbank inputs consist of a wide range of grain sizes and are generally poorly sorted. In addition bars commonly develop in response to non-alluvial boundaries and large in-channel obstructions such as large organic debris (Hogan 1986, 1987; Lisle 1986, 1987). It is hypothesised that the combined effect is to preclude the development of any systematic downstream variation in texture which fluvial sorting and abrasion processes might otherwise produce. Tributary inputs may be indistinguishable as discontinuities within the resulting signal. 1.5.3. Transfer Subsystems downstream As one moves downstream the channel becomes increasingly uncoupled from the surrounding hillslopes (Church, 1983). A point is reached where floodplain deposits are sufficiently extensive to prevent colluvial material from reaching the channel. The only sediment entering the link is that transported fluvially from the upstream channels, and sediment storage is characterised by regularly spaced, hydraulically controlled bars.(For purposes of the model it is assumed that within-link 18 exchanges of sediment involve material of fluvial provenance only; that is, the material in the floodplain is predominantly alluvial and reflects the existing transport regime.) What is of interest is the within-link modification of a sediment input to produce a sediment output. In the absence of colluvial inputs processes of abrasion and sorting are relevant. There exists sufficient empirical evidence to show that the combined effect of abrasion and sorting processes is at least approximated by the Sternberg model. Hence it is fair to expect that within-link modification of mean grain size is approximated by that model. Furthermore sorting is expected to improve between the input and the output. 1.5.4. Confluence Mixing Where any two links meet, the sediment outputs of each link are mixed to form the sediment input to the downstream link. This mixing process results in the discontinuities discussed above. The texture of the mixed sediment depends on the texture of the two inputs and their relative volumes. Continuity dictates that the summation of the respective grain size fractions will determine the texture of the mix. Theoretically the texture of any output can in turn be determined. This is simply a function of the length of the link, the input texture at its top end, the nature of abrasion and sorting as indexed by the diminution coefficient in the Sternberg model, and a general function of sorting with distance, yet to be defined. The volume of any output can be determined as the sum of all the outputs from the source subsystems upstream of this point. If we know the texture of the material leaving the source subsystems (DQ) and we can estimate the volume of material leaving each one, we can 19 work through the network using the Sternberg modelwithin links and a mixing function at confluences to estimate texture at any point in the basin. While DQ will require field sampling, it may be possible to predict volume using sub-basin parameters. The model has been sketched conceptually. In order to make it operational several aspects require attention: determination of initial grain sizes, Sternberg coefficients and the sediment outputs from the source subsystems; quantification of the sorting-distance relation; and the development of a successful mixing function. The model as it stands is valuable since it conceptualises the problem and reveals the issues which need to be addressed. One of these is the basic question of whether the distinction between source and transfer systems can be made. To this end a study of textural variation in strongly-coupled basins is valuable. It will reveal whether or not systematic downstream structure can develop given the supposed dominance of overbank inputs and non-alluvial storage elements. 1.6. Source Subsystems in the Pacific Northwest Such a study is of interest to resource managers in the Pacific Northwest where strongly coupled basins support valuable fish and forest harvesting industries. Bed material texture is rated during habitat assessments, and monitored in association with logging activities and habitat enhancement projects (Everest et al., 1981). Several studies have noted the spatial variation of bed material texture at a variety of scales in the region (e.g. Adams and Beschta, 1980; Duncan and Ward, 1985; Tripp and Poulin, 1986). Between creek and within creek variations are generally found to be high. Between creek variation has been associated with a number of factors (e.g. land use, lithology) but no single factor is consistently found to explain textural differences. The reasons for within creek variation have not been thoroughly explored (Everest et al., 1987). Adams and Beschta (1980) report that stream gradient and the average size of material in the surface layer are weakly correlated (0.35 and 0.52) with fines content. Keller et al. (1981) suggest that channel roughness features such as large organic debris, boulders, and bed-rock outcrops can sort stream sediments hydraulically. This is the extent of research into this issue and in many cases the sampling criteria used are in any case suspect. Because of the limited information given in many of the papers it is difficult to make substantive generalisations, but it appears that sample size is often too small for representative sampling of the population. A systematic programme of within-creek bed material sampling, based on sound sedimentological criteria, has never been attempted in these basins. Consequently the nature and causes of textural variation remain largely unknown. 1.7 Study Objectives Any improvement of this situation would be desirable since it would furnish resource managers with some basic understanding of the phenomenon with which they are dealing. This would be useful in a number of ways : 1. While systematic variation with distance is not expected to exist in the bed material texture of strongly coupled creeks, it is possible that other 21 forms of structure are present. The identification of any such structure would introduce a certain predictive capability which would greatly simplify habitat mapping. 2. Identifying systematic variation at different scales would enable sampling for inventory or monitoring purposes to be conducted with greater validity. Everest et al. (1987) point out that high variability precludes attempts to set universal quality standards between creeks and to instigate monitoring programs within creeks. By improving appreciation of the system, unexplained variability would be reduced and in turn it would be easier to formulate sampling schemes which minimise confounding. 3. Spatial variability is an important aspect of fish habitat since it is associated with diversity. An appreciation of this in itself would be of benefit to managers. The small basins (<50 km )^ characteristic of the Queen Charlotte Islands are assumed to be representative of the Pacific Northwest's coastal ranges in terms of strong geomorphic coupling, hydrometeorology, vegetation, in-stream large organic debris, and fish/forestry value. Hogan (1986, 1989) has investigated the morphological adjustments in space and time which the creeks in these basins make in response to mass-wasting and logging. This study is intended to complement Hogan's work by documenting and explaining the textural variation of sediment in these creeks. In particular this study has two primary objectives : 1. To assess empirically that part of the preliminary basin-wide sediment texture model pertaining to strongly coupled areas. 2. To develop an effective operational model of textural variation for predictive use by resource managers on the Queen Charlotte Islands. The data collection and analysis programs necessary to meet these objectives are described in section 2.2., and the technical details of the field work are given in section 2.3. 2. STUDY A R E A AND METHODOLOGY 2.1 Queen Charlotte Islands The Queen Charlotte Islands lie approximately 130 km off the northwest coast of British Columbia (Figure 2). The islands are within the Coastal Western Hemlock biogeoclimatic zone, the most productive forest zone in the province. In addition they produce nearly thirteen percent of the total provincial escapement of salmon (Pearse, 1982), and support a prized sport fishery. The climate is Perhumid Marine. Annual precipitation varies from 1500 mm on the eastern plains to more than 5000 mm on some west coast slopes. The islands lie in a Pacific westerly storm track and are therefore subject to intense rainstorms and highly destructive winds (Alley and Thompson, 1978). The major geologic formations consist of volcanic, plutonic, and sedimentary rocks. Generally well bedded, fissured, and jointed, these are deeply weathered and highly susceptible to erosion (Sutherland Brown, 1968). The last glaciation produced steep valley hillslopes which are now mantled by unstable till and soils generally less than 1.0 m thick (Roberts, 1987). Seismic activity along the Queen Charlotte Fault is common and may be accompanied by widespread mass wasting (Alley and Thompson, 1978). These climatic, geologic, topographic, and tectonic conditions promote high rates of sediment production and rapid mass wasting (Roberts and Church, 1986). Gimbarzevsky (1986) estimated an average slope failure frequency of 1.0 per square kilometer over the entire land mass. The characteristically small, linear, steep basins exhibit strong coupling, and Rood (1984) estimated that of the total volume of sediment mobilised by mass wasting an average of 43 percent directly enters the stream system. Additional colluvial material is delivered to the streams by bank erosion and creep (Roberts, 1987). Estimates of total sediment input to four unlogged creeks, made by Roberts (1987), vary between 41.0 and 88.0 m^km'^ yr" .^ Valley flats may develop along the lower reaches of some creeks. Channelised debris torrents and dry debris slides not only introduce large amounts of sediment but also substantial amounts of large organic debris (logs, limbs and rootwads greater than 10 cm in diameter). Additional LOD may be supplied by bank erosion, windthrow, and snowloading. The volume of LOD per unit channel area in the eight reaches studied by Hogan (1986), varies from 0.028 to 0.058 m 3 m - 2 . 2.2 Study Design 2.2.1. Operationalizing the objectives It is clear that the creeks on the Queen Charlotte Islands are strongly coupled to the adjacent hillslopes. If the preliminary model is correct then bed material texture should reflect the coupling such that distance alone does not account for observed variation of sediment texture. In accordance with the first objective the validity of the coupled/uncoupled distinction was assessed by evaluating the explanatory power of distance alone (the Sternberg model) in explaining grain size variation between tributaries. This was achieved by conducting a longitudinal survey of surface and subsurface materials in the two study creeks (section 2.3). The creeks were divided into between-confluence reaches and sites were selected within each reach strictly as a function of distance upstream. For each reach the explanatory power of the Sternberg model can the be assessed by considering the strength of the relation between distance and grain size. In order to meet the second objective it was expected that the field program would have to identify alternative structure and a more powerful explanatory factor of textural variation. This was confirmed by the surface samples of the longitudinal survey, which could be analysed in the field, and by field observations, which suggested that large organic debris dominates textural variability. Hogan (1986) has documented the morphological significance of debris pieces and jams, but their sedimentological significance has been considered only superficially. The longitudinal survey used standard sedimentological techniques which provide complete and accurate information but which are arduous and time-consuming. The average frequency of the longitudinal survey sites is two per kilometer. This resolution is too low to identify any structure caused by LOD, which has a much higher frequency of occurrence. In order to obtain an appreciation of textural variation at a higher resolution four sections (2 in each creek) were sampled using a less rigorous photographic technique. This provided limited information about the surface materials at a large number of sites in a relatively short period of time. In addition to this high resolution sampling the LOD effect was investigated by examining, in detail, the textural variations and sediment transport around several log-jams. While it was apparent that understanding the LOD effect would provide the best chance of producing an operational model of textural variation, one other factor was also considered. Land use history may affect textural variation by altering the magnitude and frequency of mass-movement events (Schwab, 1983; Rood, 1984) and the size and in-stream arrangement of LOD (Hogan, 1986). In order to assess the importance of this factor the longitudinal and photographic surveys were conducted in two basins which are similar except for their land use histories. A between-basin comparison then provides information on the effect, if any, of land use. 2.2.2 Study Creeks Riley Creek and Gregory Creek, adjacent streams which drain into Rennel Sound (figure 2), were selected for several reasons: 1. They have been examined during various FFIP projects, most notably in terms of stream morphology (Hogan, 1989a). 2. They are regionally typical in terms of biophysiography. 3. Road access is relatively good. Drainage basin characteristics are given in table 2, and in many respects the basins are similar, e.g. area, geology, precipitation. The nature of geomorphic activity and the textural composition of weathered materials are therefore expected to be similar. This provides the basis for assessing the effect of the contrasting land-use histories on textural variation. Fourteen per cent of Riley's basin area and two per cent of Gregory's is logged. The logged proportion of the actual study area is substantially greater and is adjacent to the creek in Riley (figure 2). Riley is relatively unconfined in its lower reaches and consequently exhibits no incidental hillslope activity close to the sea (figure 2). Of the eleven slope failures (> 1.0 ha) which impact directly on the main channel, none occur within approximately 6.0 km of the mouth (Hogan and Schwab, 28 Table 2. Study Basin Characteristics Characteristic Riley Gregory Geology § Major formation Yakoun Masset Minor formation Masset Biogeoclimatic zone C.W.Cdr. C.W.Cdr. ± Mean annual precipitation (mm) > 3600 > 3600 Drainage basin area (km2) 28.3 35.5 Relative Relief (m) 840 840 Percent Valley flat area (< 2° ) 23 11 Percent steepland area (> 2 0 ° ) 39 41 Density of slope failures (no. km 2 ) All failures > 0.02 ha t 3,6 2.9 Failures incident upon creek > 1 . 0 h a 0.6 0.4 Mean volume of incident failures (m^) 5250 3350 Logged area (%) 14 2 Year last logged 1978 1980 Data from Hogan (1989a) except for § From Sutherland Brown (1968): Yakoun soft volcanics (pyroclastics, volcanic sandstones); Masset soft volcanics (pyroclastics, basalts), t From Rood (1984) t Coastal Western Cedar in press). This is useful in that it provides the opportunity to assess the possible influence of a buffering valley flat on textural variation. The variations of bed material texture in the lower 9.6 km of Riley and 6.8 km of Gregory were studied. The technical details of the sampling strategy described above, are outlined below. 2.3. Longitudinal Survey 2.3.1. Site Selection The first samples were taken close to the sea. Thereafter samples were taken at an interval calculated on the basis of the total distance to be covered and the maximum number of samples that could be taken in the time available. The initial sampling density in Riley was high because of an underestimation of the time required to collect a single sample. An interval was calculated for each reach linking successive blue-line confluences (1:50 000 topographic map), and between three and ten sites were located in a reach. At each interval, as determined by hip chain, the actual site was designated as the closest riffle-pool break, irrespective of local LOD and overbank inputs. The consistent use of riffle-pool breaks throughout all of the sampling ensured that local scale variability was minimised. Riffle-pool breaks were chosen as the standard sedimentological unit because this is where prime spawning gravels are found. Samples were also taken in the riffle-pool breaks closest to the mouth of the four tributary channels considered in Riley. Site locations are shown in figure 3. The sample at 9608 m in Riley is being used as both the downstream sample of tributary four, and the upstream sample for one of the study jams (section 3.5). channel network (1=50000 topographic. 103 F/8/7 NTS ) subsurface and surface surface only 0 1 2 3 km Key Site Distance Upstream (m) t Riley IS i RLS 1 105 J O ' 2 RLS 2 3BS 2 3 RLS 3 910 4 RLS 4 1211 <x> 5 RLS 5 1490 6 RLS 6 1801 CO 7 RT 1 DS 2594 a RT 1 US 2746 F"1 9 RT 1 T R o 10 RLS 7 3514 C3 11 RLS 8 4186 S. 12 RT 2 DS 4954 13 RT 2 US 5089 C 14 RT 2 T R _ IS RLS 9 6089 >-< 16 RLS 10 7089 >—1 17 RT 3 DS 8969 18 RT 3 US 9034 19 RT 3 TR 20 RLS 11 9241 21 RT 4 DS / 9608 < J A M W US 22 RT 4 US 9656 < 23 RT 4 TR •-Gregory I G L S 1 60 2 REP 180 3 G L S 2 680 4 GLS 3 1230 5 GLS 4 1720 6 GLS 5 2330 7 G L S 6 2961 8 G L S 7 3942 9 G L S 8 4870 10 G L S 9 5455 11 GLS 10 5925 12 GLS 11 6395 13 GIs 12 6865 From Hogan's benchmark RBM 30 in Riley From Hogan's benchmark G B M 8 in Gregory C O O 31 2.3.2. Sampling Procedure At most sites a single bulk sample was taken using an oversize McNeil sampler. This enabled the substrate in the active channel to be sampled to a depth of 25 cm with minimal loss of fines. A sediment sample must contain sufficient material such that each size class present in the sample is represented in its true population proportion. Church, McLean and Wolcott (1987) found that in order to obtain a stable estimate of the proportion of a size class, approximately 100 particles are needed. Fewer than this and there is significant variation in the estimated proportions between replicate samples of the same material. Since large particles are least frequent a criterion based on the size of the largest particle present is sensible. In terms of percentage by weight any individual clast in the largest size class present in the sample should not exceed 0.1 per cent of the total sample weight (Church, McLean and Wolcott, 1977). In this study a more relaxed 0.5 per cent criterion was used. Truncation of the subsurface sample at the coarse end is inherent in both of the commonly employed sampling techniques (freeze-core and McNeil), and in any case is necessary if any of the spawning quality criteria currently in use are to be comparable between samples. The bulk samples were truncated at 64 mm, which is well within the physical limits imposed by the 205 mm opening of the McNeil. Given the 0.5 per cent criterion, a dry sample of 70 kg is required and wet samples of 80 to 90 kg were therefore obtained. This amount is feasible in terms of time limitations. In order to obtain such a sample upto six McNeil "pulls", of approximately 15 kg each, were pooled. These were made across the riffle top immediately below the upstream transition from riffle to pool. The 32 surface material was cleared to the depth of the largest clast present, and the McNeil sampler worked in to the bed. The material contained in the inner pipe was then excavated until the serrated edge of the pipe was felt. The water occupying the pipe, in which fine material was suspended, was recovered using a plunger fitted with a spring-loaded valve. Drying the samples in the field was not feasible given the high chance of precipitation on any given day. They were therefore wet sieved to < 16 mm on the assumption that water retention on the surface of a particle introduces a negligible weight increase. A set of rocker sieves was used to sort the material finer than 64 mm into half phi classes. Material greater than 64 mm was noted but not weighed or included in subsequent analysis. At the end of the sieving procedure the material finer than 16 mm, the water from the McNeil pulls, and any additional water used during sieving, was placed in 25 litre buckets and left to settle for a minimum of 16 hours. This was sufficient time for the suspended fines to settle. Free water was siphoned and carefully decanted, the damp material weighed, and in turn spread on a tarp where it could be thoroughly mixed. To ensure that a representative split of this material was obtained four subsplits each weighing approximately 3.0 kg were collected. These were then double bagged and returned to the laboratory for analysis. In addition to the bulk sample a surface sample of 100 stones was taken at each site. A fibre tape was used to delineate one or more transects across the riffle top. Stones were selected by moving along the tape at an interval determined by the largest stones present, and then reaching down with eyes averted to pick up the first clast touched with the index finger. Clasts smaller than 8 mm (b-axis) were noted but not included. This truncation is necessary since it is likely that a large proportion of surficial material less than 8 mm is hidden between larger clasts and is not detectable with an outstretched finger. For each stone the three principle axes were measured using a stainless steel carpenter's rule. The radius of curvature of the most acute corner of each stone was measured using a card with concentric circles drawn on it. In addition to the size of each stone, the roundness and shape can be determined. At the 23 sites in Riley both surface and subsurface samples were obtained, although a failure to record a tare weight at RLS 3 means that this sample is excluded from the analysis described in Chapter 3. Only at six of the thirteen sites in Gregory were subsurface samples taken (figure 3). Despite relatively good access and an efficient system of sampling and portage, moving the sampling and sieving equipment from site to site was difficult. Full coverage of one creek was the main priority. In Gregory greater longitudinal coverage was more important than taking bulk samples at every site, and consequently they were taken only at alternate sites. 2.3.3. Laboratory Analysis For each of the finer than 16 mm splits the four subsplits were weighed, oven dried at 120 degrees Centigrade, allowed to cool in the oven, and re-weighed individually. For each sample four estimates of water content were then available. The mean of these values was used to adjust the field weight of material finer than 16 mm to the true dry weight. The four subsplits were then combined and a sample splitter used to obtain approximately 5.0 kg for analysis (0.1% criterion for 16mm material). This split was dry sieved using standard procedures through half phi intervals to less than 2.0 mm. The pan fraction was then split using a sample splitter to obtain approximately 100 g of material which in turn was sieved in half phi intervals to 0.063 mm. Material finer than this was weighed as the pan fraction. 2.4.4. Within Site Variance The pooling procedure used during the collection of the bulk samples meant that a large proportion of the population (i.e. much of the designated pool-riffle break) was included in each sample, and in addition riffle-pool breaks were consistently used. It is therefore reasonable to assume that within site variance is minimal. However, a knowledge of within-site variance is crucial if between site comparisons are to be meaningful, that is, if between-site variance is to be confidently assigned to something other than sampling chance. Replicate bulk samples were therefore collected at two sites in Gregory creek, and are assumed to provide a measure of within-site variance which is typical of the entire set of samples. In order to investigate how and where sampling of bulk material should take place the individual samples were collected in a systematic manner within each riffle-pool break. This area is defined as the upstream portion of the riffle where the gradient begins to decrease and white water begins to disappear. At the first site, GLS 1 90 (Figure 4), five bulk samples were taken. The dewatering of this particular riffle meant that the samples could be shovelled from five separate locations without any loss of fines. Each of these samples was truncated at 64.0 mm as explained in section 2.3.2. They were positioned in order to assess the variance of the subsurface material laterally, and in a downstream direction, within a riffle-pool break (Figure 4). A full latin-square was not a practical possibility. The five cumulative distribution curves are shown in Figure 5 and statistical descriptors are given in Table 3. The samples taken at locations B and D are fairly similar to each other but differ significantly from that taken at A, which is generally much coarser. This may reflect the relatively elevated position of location A. The sample taken at C is also relatively coarse, while that at E is very similar to B. The dissimilarity of A, B, and D shows that bed material texture varies laterally within the riffle-pool break. This can be related to the topography of the channel cross section and differential shear at a given flow depth. In order to characterise the bed material in this area, a composite of laterally positioned samples should be collected. A single sample taken from some point is unlikely to be representative of the entire unit. The similarity of B and E suggests that as long as lateral topography is maintained, the downstream positioning of the sampling row within the riffle-pool break is not crucial. Figure 4. Sketch map of replicate sampling sites 36 Figure 5. Replicate bulk sample distributions GLS 1 90 Grain Size Figure 6. Replicate bulk sample distributions REP 90 Grail Size Table 3. Replicate Sample Characteristics for GLS 1 90 A B C D E D 5 (mm) 0.37 0.57 0.42 0.52 0.57 D 1 6 (mm) 1.46 1.64 1.93 1.18 1.70 D 25 (mm) 3.33 2.83 4.38 2.22 2.92 D 5 0 (mm) 16.82 9.65 17.20 10.17 10.21 D75 (mm) 39.51 26.73 32.91 26.52 28.49 Dg4 (mm) 49.27 36.36 41.37 35.89 38.37 D95 (mm) 58.98 53.45 55.69 53.74 54.70 ai (phi) 2.38 2.11 2.17 2.24 2.12 Notes: X percent finer than Inclusive standard deviation Table 4. Replicate Sample Characteristics for GLS REP A B C D D 5 (mm) 0.68 0.73 0.48 0.79 D 1 6 (mm) 2.45 2.81 2.18 3.55 D 2 5 (mm) 4.55 4.92 4.73 6.19 D 5 0 (mm) 14.00 15.82 15.44 16.35 D 7 5 (mm) 29.87 32.68 30.25 37.79 D § 4 (mm) 38.56 41.66 38.58 47.36 D 9 5 (mm) 54.33 56.11 54.13 58.26 o-j (phi) 1.95 1.92 2.07 1.87 Notes: CTI X percent finer than Inclusive standard deviation At the second site, REP 90 (Figure 4), four bulk samples were collected. Two of these were shovelled from adjacent positions at the water's edge: one in shallow water (< 5.0 cm), with little current (C); and one in slightly deeper, faster flowing water (D). The other two samples were collected using the pooled McNeil technique described in section 2.3.2., and employed throughout the bulk sampling program. Individual pulls were taken at alternate positions within a 1.0 m wide strip across the riffle-pool break (A and B). The four cumulative curves are shown in Figure 6, and statistical descriptors are given in Table 4. Samples A and B are very similar, and although sample C is somewhat finer in the lower part of the distribution, it is generally similar to A and B. This is consistent with the notably less variable topography at this site. Sample D is consistently coarser than the others as a consequence of shovelling the bulk material out of the inundated riffle. These results confirm the value of the sampling method used throughout the bulk sampling program. The pooled McNeil method overcomes lateral heterogeneity caused by cross-section topography, and ensures that losses of fine material are minimised. Samples GLS REP A, B, and C plus D (C and D pooled) have been used to estimate the within site variance of subsurface material that can be expected in riffle-pool breaks. These three samples are comparable in terms of lateral pooling. However, the variance terms which they yield should be regarded as conservative estimates because of the known shovelling bias introduced by C plus D. Table 5 summarizes the relevant statistics for the four descriptive parameters used in this study. A within-site variance for surface material estimators has also been estimated in order to provide some basis for between site comparisons. Bray 4 1 Table 5. Bulk Sample Within-Site Variance Surface DKQ Subsurface (sub 64.0 mm) mm Phi D 5 0 (mm) Fredle Mean 0.13 0.039 15.57 4.43 n 8 8 3 3 S.D. 4.20 0.149 1.47 0.37 Variance 17.6 0.022 2.15 0.14 Conf. Limits ± 3 . 5 1 ± 0 . 1 2 4 ± 3.64 ± 0.92 Notes : i) For the surface, values refer to the differences between eight pairs of replicate means, ii) Confidence limits are for a=0.05 (1972) recommends that a consistent estimate of mean size requires at least 50 stones. By randomly dividing the 100-stone sample from a single site in half, two representative replicate samples are obtained. This was done for eight randomly selected sites. For each 50 stone replicate the median grain size was determined, and for each pair the difference of the estimates calculated. These can then be used to provide an estimate of within-site variance of median estimates. The mean of the differences is 0.125, the standard deviation, 4.1982, and the standard error, 1.4843 mm (n = 8). With a t-value of 2.365 (a=0.05, v = n-l = 7), the 95 % confidence limits on the mean difference are ± 3.5104. Given that the mean difference is very close to 0.0, one can be confident that in 95 cases out of 100, two estimates of the surface D 5 0 at any site, will be within 3.5 mm of each other. This analysis has been repeated in phi units for use in chapter 3 and the details are presented in Table 5 Fifty stone samples yield less consistent estimates of a population characteristic than 100-stone samples. Consequently the confidence limits based on 50-stone replicates ought to be considered as conservative when applied to the 100-stone estimates of D5Q which have been determined for each of the surface samples. As field assessments of within-site homogeneity suggested, the variability of surface material within a single riffle-pool break is very small. 2.4 High resolution photographic surveys The surface sediments of two reaches in each creek (figure 7) were studied using a photographic technique. This method is based on the CO establishment of an empirical relation between median grain size and the number of particles in a given area on the bed surface. A standard area, delineated by a quadrat placed on the surface, is photographed at each site. The number of particles within the quadrat can be counted on the photograph. At a selection of the photographed sites the median grain size of the material within the quadrat is determined using standard sedimentological techniques. These "calibration" samples are in turn used to define the relation between grain size and number of clasts. By counting the number of stones in the respective photo this curve can be used to estimate the grain size at any site. 2.4.1. Field Technique Within each reach every riffle-pool break was photographed. A 1.0 m z quadrat was placed on the bar adjacent to the riffle, as close to the water as possible. With a 35 mm lens the photograph must be taken from a height of approximately 1.7 m in order to include the whole quadrat. This is possible standing on the bar surface with the camera held above one's head over the centre of the quadrat. In order to avoid camera shake a shutter speed of 1/250^ of a second was used where possible. A bullseye spirit level mounted on the back of the camera, and fitted with a small mirror, enables one to keep the camera parallel to the bed surface while the photo is taken. A 1.0 m quadrat was used on the basis of the range of grain sizes evident and the commonly accepted view that with 100 stones a very good measure of the surface mean size is obtained. In practice this area was more than adequate and therefore only the stones in the central 0.25 m 2 of the quadrat were counted on the photographs. In order to construct the calibration curve median grain size was determined for fourteen of the 130 sites photographed. These were distributed between the four reaches and were chosen to provide a wide range of grain sizes. A 10 by 10 cm grid built into the quadrat allowed the random selection of 100 stones. The b-axis of each was measured. If a stone lay beneath n intersections this was noted and the measurement included n times in the analysis. The sample was truncated at 8.0 mm and clasts smaller than this, as indicated by scale bars drawn on the quadrat, were not included in the photo count. 2.4.2. Calibration of Particle Size and Number The untransformed data from the fourteen sites (Table 6) produce a significant relation, but it was found that higher predictive power is attained if both variables are log transformed. When fewer than approximately 150 stones are present in the 0.25 m 2 quadrat, the relation approaches that expected theoretically, i.e. grain size (D) is inversely proportional to the square root of the number of stones per unit area (N): log D logN" 0 - 5 For greater numbers of stones the relation is more sensitive, perhaps because of greater counting errors and resolution problems as grain size decreases. Over the full domain a curve is therefore most appropriate. The curvilinear relation shown in Figure 8, fitted by eye, has the highest coefficient of determination (r 2 = 0.924) and lowest standard error of any of several relations tried. The calibration was used to estimate the median grain size for each of the 116 photographed sites. The remaining scatter reflects the complexity of natural gravel surfaces, which contain particles of many different shapes and sizes packed Table 6. Photographic Survey Calibration Data Site Number of stones Surface material > 8 mm (0.25 m-2) median (mm) LR 292 209 50 LR 845 284 34 LR 975 246 50 LR 1331 176 61 LR 1658 188 44 L G 0 207 71 L G 189 260 31 L G 160 257 52 L G 466 456 21 L G 259 574 19 U G 15 135 107 U G 472 26 188 U G 1469 58 160 UR 194 84 115 Figure 8. Photo calibration curve 47 Number of particles per 0.25m 2 (log base 2) together such that grains overlap to varying degrees. Some error is also introduced at the stone-counting stage when, in spite of scale bars, a degree of subjectivity is needed in order to determine whether a clast has a b-axis of less than 8 mm. Photographs of wet and rain spotted surfaces are particularly prone to error. 2.5 Log Jam Studies Field observations suggest that log jams and pieces of LOD can cause significant local sorting by altering channel gradient and presenting a physical barrier to mobile sediment. These observations are supported by Hogan's data on the variation of D95 in his surveyed creeks (Hogan, 1989a). In some cases surface texture is obviously very different upstream and downstream of log jams and the surface is homogeneous. In other cases a number of distinct facies are evident and the distinction between upstream and downstream is less apparent. It is supposed that this reflects differences in sediment transport rate through the jam. This permeability is in turn a function of various jam characteristics related to age and morphology. In order to investigate sorting, homogeneity, and permeability, three jams in Riley Creek were considered in some detail (sections 2.5.1 and 2.5.2). In addition the distances of all of the photo samples and many of the longitudinal samples from upstream and downstream jams were determined. Jams were classified according to the scheme of Hogan (1989a). 2.5.1. Bed Material Sampling Three jams in Riley were selected on the basis of their different ages and morphologies (figure 7), and surface and subsurface samples were taken upstream and downstream of each. The characteristics of the entire unit of sediment upstream and downstream is of interest and the sampled unit was therefore defined as the riffle closest to the jam plus the adjacent bar deposits back to the bankfull width. Within the unit several distinct facies were sometimes present. While an appreciation of these is important it was not possible to sample the facies individually because of time constraints. Instead they were carefully mapped using qualitative descriptions based on the relative prevalence of fines, gravel, cobbles and boulders. The general characteristics of the entire unit were then determined using a method suggested by Wolcott and Church (in press). Working on a gravel bar in the Quesnel River they compared a single sample which was an aggregation of 46 small grabs, with the results from 46 individual bulk samples. The pooled sample falls well within the 95 % confidence limits of the median based on the 46 samples. A single pooled sample can thus be representative of the whole area from which the grabs are taken. Sixteen sites were located within each unit using a randomly positioned grid which maximises areal coverage and avoids bias with respect to a single facies. At each site approximately 10 kg of subsurface material and seven clasts from the surface were collected. Where the site was not under flowing water the bulk samples were shovelled out. The resulting pooled sample gives a surface sample of 112 stones and approximately 160 kg of subsurface material. Analysis of these samples in the field and the plaboratory was the same as for the longitudinal survey samples, except that the bulk samples were truncated at 90 mm rather than 64 mm. This was convenient, and gives a greater overlap between the surface and subsurface sampled ranges, making surface/subsurface comparisons more meaningful. 2.5.2. Permeability Measurements In order to assess the relative sediment transport rates through the three jams (jam permeability) two complementary methods were used. At each jam the creek was seeded with magnetic tracer particles both above and below the jam. These fiberglass "stones" contain a small ceramic magnet and can be traced, even when buried, using a magnetic field locator. In the summer of 1989 thirty of these stones, painted fluorescent orange, were placed in a line perpendicular to the channel banks immediately upstream of each jam. The stones were regularly spaced across the full width of the channel and were dropped to the surface from a height of approximately 50 cm. The line was marked in the banks at each end, and minor displacements around the line noted. By monitoring the transfer of these stones from the upstream to the downstream side of the jam a measure of each jam's permeability can be obtained. However in order to make comparisons between the three jams some attempt must be made to normalise each rate relative to the ambient transport rate for the local reach. A line of thirty green stones was therefore deployed, as above, across a line located downstream of the jam beyond any morphological evidence of jam influence. Upon return in the summer of 1990,the magnetic field locator was used to search for the stones. Access problems meant that this was only 51 done at the two most different jams (W and 3L). The jam and bankfull channel were searched from the upstream deployment line to a point approximately 100 m below the downstream line. The colour, distance moved, burial depth, and in-channel position, were recorded for each stone found. As an adjunct to the stone placements, scour chains were positioned around jams W and 3L, to assess relative aggradation and degradation. In each case four chains were placed at sites chosen because they appeared to be susceptible to net erosion or deposition. Each scour chain consists of sixteen perforated practice golf balls, 4 cm in diameter, threaded onto 180 cm of heavy duty fishing line. A weight was tied to one end, and a numbered tag to the other. A scour chain inserter similar to that used by Tripp and Poulin (1986) was used to plant the chains. An additional brightly painted ball was left above the surface as an aid to recovery, and the position of each chain was marked on the adjacent banks.The length of the line from the top of the uppermost ball to the base of the identification tag was measured before planting, and the number of balls buried was noted afterwards. Upon return in the summer of 1990 the burial positions were located. For those chains which were found on the surface the number of balls visible was noted, and the length of line from the bed surface to the base of the tag was measured. A note was also made of the depth below the surface of the uppermost ball in its original position. For those found beneath the surface the depth of burial, and the length of line from the tag base to the uppermost ball in its original position, were measured. The number of balls detached from their original position was also noted. From these 52 observations the maximum scour and fill and the net change in bed elevation at each site can be calculated. CHAPTER 3 EVALUATION OF THE STERNBERG MODEL This chapter aims to determine whether or not the Sternberg model can be applied to explain the variation of grain size within between-confluence links in Riley and Gregory Creeks. Consequently the applicability of the Sternberg model in strongly coupled basins is assessed. Discontinuities are expected at the confluences as a result of fresh sediment inputs, but according to the general Sternberg model, grain size should decrease exponentially as one moves downstream within a link. The four test links in Riley, and the three in Gregory are shown in Figure 3. Systematic changes in texture are most likely to be seen in the surface materials, since these are exposed to the sorting affect of the flow. It therefore seems prudent to assess the Sternberg model using these data; a consideration apparent in previous work (Table 1). However, variations in the subsurface material are of interest in the context of spawning habitat, and the bulk material in Riley Creek is therefore considered too (the paucity of bulk samples within individual links in Gregory makes it difficult to make any reasonable inferences here). 3.1. Longitudinal Survey Data 3.1.1. Construction of Synthetic Bulk Samples The subsurface samples collected during the longitudinal surveys were truncated at 64.0 mm. In the context of testing the Sternberg model it is necessary construct the coarse portion of each of these samples. This is because fining in the bed material, predicted by the model, is synonymous with the successive downstream removal of coarse material from the bed load. It is the decline in the proportion of coarse bed material that causes the predicted decline in grain size as one moves downstream. In turn, by extending the bulk sample coverage to the full range of sizes present, one is maximising the chances of detecting the Sternberg phenomenon. Construction of the full bulk distributions is based on the classic view of the relation between surface and subsurface materials in gravel bed rivers. A typical feature of such rivers is a surface layer which is coarse relative to the underlying bulk material. This is assumed to be a consequence of the preferential removal of fine material from the surface layer. The lag deposit which results is therefore a truncated sample of the subsurface material, such that at any site the grain sizes present in the surface layer are found, in the same proportions, in the bulk material below. A single un-truncated bulk sample and a regular surface sample were collected at a site in Gregory (GLS 1 90). The fractional percentage of the samples in each half phi class (normalised over the common range, i.e. + 8.00 mm) were compared in order to identify the point at which truncation occurs. Since the bulk material was analysed "sieve by weight", and the surface material "grid by number", the class proportions are directly comparable (Kellerhals and Bray, 1971). In Table 7 the differences between the two distributions are given, and it is evident that relative to the subsurface, the surface material is deficient in material as large as 44.0 mm. In the 45.0 to 64.0 mm range the two proportions are however very similar, which suggests that the amount of material in this class, and those above it, is the same in the surface and the subsurface. There is a problem in that the 90.0 to 127.0 mm class looks suspiciously large. This suggests that the surface distributions are not well characterised with a sample of 100 stones. However, one has no choice but Table 7. Surface and Subsurface at GLS 1 90 Class Fractional Percent Finer Subsurface Comments Subsurface Surface - Surface (Surface) 128.0 + 0.08 0.07 + 0.01 Poor data 90.5 + 0.05 0.26 - 0.21 Poor data 64.0 + 0.17 0.21 - 0.04 Enriched 45.3 + 0.16 0.18 - 0.02 Similar 32.0 + 0.13 0.10 + 0.03 Deficient 22.6 + 0.14 0.11 + 0.03 Deficient 16.0 + 0.13 0.05 + 0.08 Deficient 11.3 + 0.08 0.01 + 0.07 Deficient 8.0 + 0.08 0.00 + 0.08 Deficient to assume that the proportions given by the surface sampling approximate those present in the bed. For each of the longitudinal survey samples which are truncated at 64.0 mm, the distribution of the overlying surface materials is available. From the bulk sample one knows the weight of the material in the 44.0 to 64.0 mm class, and from the surface sampling one can calculate this class as a proportion of the untruncated part of the surface distribution (+ 44.0 mm). In turn one can calculate the total weight of subsurface material in this range and, using the relative proportions given by the surface sample (normalised for the untruncated range), one can calculate the weight of each class greater than 64.0 mm in the subsurface material. These weights are then added to the sub-64.0 mm distribution (common class only once) and the whole set renormalised to give a full synthetic subsurface distribution. 3.1.2. Results and Analysis For each of the longitudinal survey sites the median grain sizes of the surface and synthetic subsurface samples are presented in Table 8. These data are plotted in phi units against distance upstream in Figure 9. Phi units are used in order to simplify the identification of the Sternberg phenomenon (an exponential trend will plot as a straight line). The within -site variance estimate of the surface material, determined in section 2.3.4 , has been used to assign 95 % confidence limits to the median values of the surface samples. No such variance estimate is available for the full subsurface samples, and confidence limits cannot therefore be determined, although within-site variance is not thought to be significantly large given the nature of the sampling procedures (section-2.3.2.). Table 8. Longitudinal Survey Data Site Distance Surface Surface Bulk Bulk Ustr(m) D5 0(mm) D 5 0 (Phi) D 5 0 (mm) D 5 0 (Phi) RLS 1 105 44.3 -5.46 17.27 -4.11 RLS 2 385 40.8 -5.35 10.20 -3.35 RLS 4 1211 31.8 -4.95 15.24 -3.93 RLS 5 1490 52.0 -5.70 21.56 -4.43 RLS 6 1801 32.0 -5.00 21.56 -4.43 TI DS 2594 64.9 -6.02 29.86 -4.90 TI US 2746 65.3 -6.02 22.32 -4.48 RLS 7 3514 45.3 -5.50 16.11 -4.01 RLS 8 4186 53.1 -5.73 31.13 -4.96 T2 DS 4954 58.5 -5.87 14.62 -3.87 T2 US 5089 24.6 -4.62 9.00 -3.17 RLS 9 6089 61.8 -5.95 19.97 -4.32 RLS 10 7089 122.8 -6.94 106.15 -6.73 T3 DS 8969 128.0 -7.00 72.00 -6.17 T3 US . 9034 37.5 -5.23 20.82 -4.38 RLS 11 9241 37.8 -5.24 16.80 -4.07 JW US 9608 32.4 -5.02 14.32 -3.84 T4 US 9656 38.1 -5.25 20.68 -4.37 TI TR 2650 33.6 -5.07 16.68 -4.06 T2 TR 5000 75.1 -6.23 42.22 -5.40 T3 TR 9000 57.3 -5.84 20.25 -4.34 T4 TR 9656 34.5 -5.11 24.08 -4.59 GLS 1 60 58.1 -5.86 22.63 -4.50 GLS 2 680 102.5 -6.67 GLS 3 1230 49.9 -5.64 19.97 -4.32 GLS 4 1720 128.0 -7.00 GLS 5 2330 52.0 -5.70 25.46 -4.67 GLS 6 2961 157.6 -7.30 GLS 7 3942 39.7 -5.31 19.97 -4.32 GLS 8 4870 68.6 -6.10 GLS 9 5455 87.4 -6.45 49.18 -5.62 GLS 10 5925 187.4 -7.55 GLS 11 6395 152.2 -7.25 GLS 12 6865 28.6 -4.84 Figure 9. Longitudinal survey data, grain size variations a)Rilev surface D50; b) Riley subsurface D50; c) Gregory surface D50 -ao •7.0 h Link t S. &0 •5.0 y -4.0 -ao -ao -7.0 i Link 2 i Lit* 3 ll I I I I <jj L. fl//«y ,*> _l 1 I L _ 2 3 4 5 6 7 Distance upstream (Km) Link 1 ao 3 -5.0 OT -4.0 -ao 4, Link 2 Link 3 -J l_ -ao , 4 s 8 7 Otstance upstream (km) 4/ Link 2 4' Link 3 I I I I I OT -5.0 4^4 4/ U- Riley 9 10 I Link I 4 / 4 ^ 10 58 -4.0 <j3 L. G r e a cO U. Greg cj, "3.0 Distance upstream (km) As an adjunct to the graphical evidence, a formal analysis of surface material variations is presented in Table 9. For each link, in each creek, between-site variance was calculated and compared with the within-site variance term. By forming the ratio of between-site to within-site variance, an F-value is obtained which can be compared with the critical value for the given degrees of freedom. In turn one can make a statistical statement about the significance of the between-site differences within a link. The within-site variance term is considered to be accurate, and in this respect the tests are good. However, the small number of samples in most of the links means that the critical F-values are high. While this is of little consequence in Gregory Creek, and in links one, three, and four in Riley Creek, it may be important in reach two of Riley where the difference between the critical and calculated values is relatively small. The within-link variations revealed in Figure 9, are very inconsistent, and each link warrants attention. First, consider the surface materials, where important considerations are the statistical significance of between-site variations, whether these are erratic or systematic, and whether a Sternberg-type trend is apparent. In link four in Riley there is little overall change, which is in any case expected given the short distance involved (ca. 700m). In link one in Riley overall downstream fining is apparent and the variation between sites is significant. However, the decline is not systematic, and the between-site significance reflects the strongly erratic behaviour of the data rather than a strong downstream fining trend. Link one in Gregory is similar in these respects, but the median grain sizes exhibit greater fluctuations. In link two in Riley there is no significant overall change in grain size, although a fairly systematic decline is evident upstream of the relatively Table 9. Within-Link Variance of Surface Materials n l a2W n 2 F Fcrit Signif Surface Rfley 1 2 0.17. 0.05 6 4 0.02 8 7.62 2.21 3.97 4.35 Yes No 3 1.24 4 56.19 4.35 Yes 4 0.02 3 0.70 4.74 No Surface Gregory 1 0.52 6 0.02 8 23.47 3.97 Yes 2 0.34 3 15.43 4.74 Yes 3 2.21 3 99.86 4.74 Yes Notes : i) F is the ratio of between-site variance (cr^ g) to within-site variance (CT^^T ) ii) Fcrit is for a=0.05, (n^-l,n2-l) degrees of freedom iii) When F > Fcrit, between site variance is significantly greater than within site variance 61 coarse sample taken above tributary one. In link three between-site differences are definitely significant, and strong downstream fining is apparent. Although the phi data do not plot as a straight line, it is possible that they reflect a strongly perturbed Sternberg signal. The same interpretation could be attached to the data of link two. Link two in Gregory provides convincing evidence of a Sternberg-type decline: between-site variations are significant; the overall trend is downstream fining; and the phi data plot very close to a straight line. In contrast, in link three in Gregory, between site variations are significant, but the surface materials coarsen downstream. The subsurface material in link one of Riley exhibits an overall fining trend which would be fairly systematic were it not for a relatively coarse sample at RLS 1. Overall coarsening is evident in reach two, but the data are very erratic. In link three overall fining is apparent and, as with the surface material, may reflect a perturbed systematic decline. It is especially difficult to judge without a clear appreciation of sampling error. In summary, in only one case is a significant, systematic, clearly exponential downstream trend observed. In some instances it is possible that the data reflect a strongly perturbed Sternberg signal. In several others the data behave very erratically, and exhibit no clear trend, and in one case material coarsens significantly downstream. Because of the ambiguity of these results, and the limited number of data within any given link, it is difficult to make any general comments. The higher resolution photographic survey data has therefore been examined. 3.2. Photographic Surveys Surface median grain size estimates for every riffle-pool break in the four photographed reaches, are given in Table 10. The location of these reaches within the longitudinal surveys is indicated on Figure 9, and in Figure 10 the D50S are plotted against distance upstream, again in phi units in order to simplify recognition of exponential trends. The 95 % confidence limits based on the standard error of the calibration model (section 2.4.2) are also shown. The upper reach in Riley is not considered here because of its short length. In addition to this graphical evidence the data have been analysed statistically in order to determine whether any significant downstream trends are apparent. The data presented in Table 10 do not yield a within-site component of variance since no replicate samples (photos) were taken at any of the sites. In order to overcome this, adjacent D 5 0 estimates were pooled into small groups, each of which is considered to be a set of observations under the treatment J , where J is the range of distances over which the n grouped samples occurred. In most cases samples within consecutive 200 m sub-sections were grouped, yielding up to nine treatment groups of between three and seven observations each. The analysis of variance model is then : D i j = D - + Tj + eij where Djj is an individual observation; D.. is the overall mean median grain size; T is the between site variance due to location (3 < j ^ 9); and ejj is the within site variance or experimental error term (3 < i < 7). Taking into account that within any reach the number of observations in different treatment groups varies, a one-way analysis of variance was 63 Table 10 a). Surface Grain Size Estimates in Riley Creek Dist. (m) Number D50 (mm) Dist. (m) Number D50 (r Lower section. Upper section 0 398 27.1 0 138 82.7 20 352 31.6 30 394 27.5 40 346 32.5 62 315 36.3 83 363 30.5 83 139 82.1 140 334 33.8 109 502 19.2 170 399 26.7 118 84 117.0 228 324 35.3 137 229 51.3 272 295 39.4 176 230 51.3 292 209 56.5 203 470 21.4 335 303 38.1 235 452 22.0 409 279 41.9 283 328 34.8 436 495 19.6 293 846 7.1 469 379 28.6 307 508 18.8 512 347 32.2 560 488 20.1 605 309 37.3 689 248 47.5 757 305 37.8 845 284 40.8 868 327 34.8 906 406 26.0 975 246 47.8 992 523 18.0 1034 501 19.2 1070 224 53.1 1151 262 44.6 1201 306 37.5 1278 587 14.9 1302 252 46.5 1331 176 67.2 1392 214 55.3 1407 263 44.3 1487 263 44.3 1492 559 16.1 1537 327 34.8 1588 315 36.3 1629 153 76.1 1658 188 62.7 1710 449 22.8 1735 367 33.1 1768 556 16.3 1788 401 26.5 64 Table 10 b). Surface Grain Size Estimates in Gregory Greek Dist. (m) Number D50 (mm) Dist. (m) Number D50 (mm) Lower section Upper section 0 207 57.3 15 135 84.5 29 248 47.5 107 293 39.7 84 187 62.7 176 225 52.7 111 177 66.7 223 265 44.0 160 257 45.9 303 178 66.3 189 260 45.6 327 186 63.1 216 309 37.0 366 153 76.1 230 231 51.3 423 116 94.3 259 547 15.6 472 26 245.6 263 226 52.0 507 441 23.4 281 184 64.0 579 145 79.9 315 336 33.6 605 235 49.9 371 275 42.2 658 155 75.1 410 655 12.2 700 251 46.9 448 354 31.3 818 429 24.3 466 456 22.3 940 216 55.0 487 219 54.1 1172 217 54.8 522 394 27.5 1211 213 55.7 613 245 47.5 1391 73 127.1 709 295 39.4 1469 58 144.0 747 63 137.2 1660 453 22.7 823 638 13.0 1702 303 38.3 857 745 9.5 1757 409 25.8 909 288 40.2 1815 261 44.6 998 146 79.3 1845 347 32.2 1037 154 75.6 1869 278 41.9 1099 101 104.0 1893 229 51.3 1120 173 68.6 1198 164 71.5 1258 712 10.3 1315 276 42.2 1355 241 48.8 1387 344 32.5 1429 187 62.7 Figure 10. Photographic survey data, grain size variations a)Lower Riley; b)Lower Gregory; c)Upper Gregory s a 1 1 !' , i ! i "I I 200 «0O 600 800 5000 1200 1400 1600 Oisianc* Uoslraam im) 3 -4.0 I I I I I 200 400 600 600 1000 1200 1400 65 c 8 0 O -5 .0 L i I I 200 400 600 600 tOOO 1200 1400 D ' S l s n c * Upstream <m> « 0 0 1800 conducted for each section. The results are presented in Table 11. In all cases the calculated F value is less than the critical value. This indicates that in each reach the grain size difference between any two groups is not significantly greater than the average within group difference. In turn this suggests that distance alone is not a significant cause of surface grain size variation within these reaches. However the possibility of a strongly perturbed downstream decline is evident in Lower Riley, and perhaps in Upper Gregory too. In Lower Riley, below the tributary (Figure 10 a) the majority of the data show a very erratic overall decline in median grain size. A number of anomalously fine samples are largely responsible for the lack of any trend. Of the three reaches this is the one which, according to the theory presented in Chapter 1 has the greatest potential for exhibiting a Sternberg-type decline, because it is within the relatively unconfined portion of Riley Creek which is free of incidental mass movement events. However, the strong perturbation is not wholly unexpected given that special storage mechanisms (particularly large organic debris jams) are as prevalent here as in the confined parts of the creek. In Upper Gregory there is evidence of relatively systematic grain size declines over Limited distances. Indeed, several discontinuities are evident but unlike the data of previous studies these are not associated with tributary inputs. In Lower Gregory the data are extremely erratic and show no trends at all. Table 11. Within-Reach Analysis of Variance for Surface D50 Reach Source df SS MS F Fcrit Lower Distance 8 1172.5 146.6 0.70 2.27 Riley Error 33 6904.2 209.2 Total 41 8076.7 Lower Distance 6 6077.4 1012.9 1.56 2.46 Gregory Error 27 17513.8 648.7 Total 33 23591.1 Upper Distance 7 21690.5 3098.6 1.71 2.71 Gregory Error 19 34458.7 1813.6 Total 26 56149.1 Notes : Fcrit Critical value of F for a = 0.05 SS Sum of squares MS Mean sum of squares df Degrees of freedom 3.3. Summary In general, the Sternberg model alone does not explain the within-link variations of median grain size in these creeks.In general then, distance appears not to be a useful predictor of grain size in strongly coupled creeks. However, a perturbed Sternberg signal might be interpreted in several cases, and what could be viewed as very strongly perturbed signals are evident in some others. Thus, while it is clear that bed material texture does not respond to distance as the Sternberg model predicts, it may be that the model is useful in explaining some of the variability observed. Distance and the Sternberg model, might therefore be a useful component in a model which aims to explain textural variations in strongly coupled systems. It remains to be determined what the causes of the perturbations are, and whether they can explain sufficient of the variation observed to yield a more useful predictive model of grain size than one based solely on distance. 69 4. T H E E F F E C T OF L O G JAMS ON BED M A T E R I A L T E X T U R E It was suggested in Chapter 1 that the input of colluvial materials to strongly coupled creeks would lead to erratic fluctuations in grain size . It was also pointed out that storage elements such as large organic debris jams could add to this variability. In addition to these effects the input of fresh material at tributary junctions was identified as a general cause of grain-size discontinuities. On the Queen Charlotte Islands the erratic fluctuations which characterise grain size variation are primarily caused by large organic debris jams. These produce highly variable local sediment transport rates, and in turn cause significant variations in bed material texture. Evidence in support of this conjecture is presented in this chapter using the results of the detailed jam studies, some other examples, and the photographic survey data. 4.1. A Priori Ideas It is possible to suggest a priori how bed material texture is affected in the vicinity of a log jam. In the most simple case a log jam acts as a dam which is impermeable to sediment . This physical barrier leads to aggradation upstream and in turn local channel gradient is reduced. The entrapment of material prevents the resupply of the mobile fractions to the bed downstream, and supposing that size-selective transport is in operation, relative coarsening will occur. Upstream the most mobile fractions will constitute the largest portion of the material being trapped, and consequently, relative fining will occur here (Figure 11). The ambient rate of 70 Figure 11. Schematic of grain size changes in the active channel in the vicinity of log jams Grain Size •A. IMPERMEABLE Downstream Distance from Jam Upstream Grain Size fining INCREASINGLY PERMEABLE s coarsening < Downstream Distance from Jam > Upstream 71 sediment transport and the stability and longevity of the log jam are likely to be important determinants of the absolute magnitude of the upstream/ downstream contrast in grain size. Log jams are temporally dynamic, commonly forming and disintegrating on timescales of up to 60 years in the Queen Charlotte Islands. (Hogan, pers. comm.). An initially impermeable jam gradually becomes more permeable as it gets older. The integrity (strength) of the jam and the proportion of the channel-width which it occupies (span) decrease, while the number of channels which flow through it increase. In turn the upstream sediment accumulation is gradually released downstream and the upstream/downstream grain size disparity is moderated until it is no longer evident (Figure 11). Logjams are therefore expected to have a variable effect depending on their integrity, span and, in general, their age in so far as this affects permeability. 4.2. Study J a m s The relations between log jam characteristics, sediment permeability, and grain size characteristics were examined at three log jams in Riley Creek. The jams were described in the field in terms of five characteristics which are thought to reflect jam permeability (Table 12). For each characteristic semi-quantitative criterion were used to assign categorical values to the jam (see Appendix 1). In general the higher the number the greater is the supposed permeability. Magnetic tracer particles and scour chains were used to confirm the relative permeability of the jams directly, and sediment samples were collected upstream and downstream as described in section 2.5. 72 Table 12. Study Jam Characteristics Span Integrity Age Channels Sed.Stor. J a m N 2 3 5 3 3 Jam K K 2 2 2 2 2 Jam U 1 1 2 2 1 Notes : i) See Appendix 1 "Hogan's Jam Classification Scheme" for explanations of the parameters and categories used, ii) * Jam is severely undercut. The three jams represent three levels of permeability. Jam U is an example of a highly impermeable jam. It spans the full channel width, is very strong, is breached by two relatively small channels, and is less than five years old. There is significant sediment storage upstream such that the channel is filled to the height of the jam (approximately 3.0 m) across the full bankfull width (approximately 10.0 m). This wedge extends some 50.0 m upstream before grading back into the ambient channel gradient. In contrast jam N is highly permeable. Although it spans three-quarters of the channel width it is undercut by three channels, is of low integrity, and is between thirty and fifty years old. Upstream, the active channel is approximately 2.5 m below the uppermost bar surfaces close to the jam, suggesting active incision of a previous sediment accumulation. Jam K K is slightly older than U (although it belongs to the same age class), is weaker, and has lost more of its upstream sediment accumulation. It represents a point somewhere between the high permeability of N and the low permeability of U , and is probably closer to U. 4.3. Confirmation of Differential Permeability Although tracer particles were deployed at all three jams, for logistical reasons they were recovered only at U and N. In both cases the movement of the green stones placed downstream of the jam indicates the ambient transport conditions. The movement of the orange stones placed upstream of the jam indicates the sediment transport through the jam. Distances of movement and burial depths are given in Table 13, and shown graphically in Figure 12. 74 4.3.1. Tracer Recovery Of the green stones deployed in the summer of 1989, 63% were recovered at jam N and 60% were recovered at jam U, in the summer of 1990. The remaining green stones are undiscovered either because they were transported more than 100.0 m (which is how far downstream of the deployment line the creek bed was searched) or because they were buried more than 60.0 cm below the surface. Although the magnetic field detector used can locate particles upto a depth of 150.0 cm, background noise caused by magnetic minerals in the bed material reduce the effective capability. The deepest stones found were 60.0 cm below the surface, and one can therefore be confident that the detector was effective to at least this depth. Of the orange stones deployed, 40% were recovered at jam U, and 47% were recovered at jam N. Those stones which remain undiscovered may have been transported through the length of channel between the two deployment lines and then beyond the 100.0 m search limit, may have been buried more than 60.0 cm below the surface, or may have been trapped in the jam where only a very superficial search is possible. 4.3.2. Downstream Tracers At both jams green stones were recovered from the reach downstream of the deployment line (Figure 12). At jam N a large proportion were found in a bar approximately 90.0 m downstream of the line, and two of these were buried beneath 40.0 cm of gravel. This suggests a relatively active sediment transport regime at this site. At jam U a greater proportion of the green stones remained close to their original positions, suggesting a less active bed, but stones were still found 90.0 m downstream. JAM N Distance (m) o o » o JAM U Distance (m) i , 1 6 0 ~ T — 160 Green line 120 8 0 Green line 120 i — . ^ . 6 8 0 o Stones originally at Orange line • " Green 4 0 i— 4 0 Orange line JAM o 2 -ho 2 0 3 0 4 0 Depth (cm) Orange line JAM oo4 o o - ^ 2 0 ^ 3 0 0 10 r—• • 0) to H p CD ••»• o p a P o CD CD E3 ct-C/2 Depth (cm) 4 0 5 0 6 0 Table 13. Tracer Stone Movements (1989-1990) Jam U Jam N Green Orange Green Orange Dist. Depth Dist. Depth Dist. Depth Dist. Depth 0.0 0.02 0.1 0.15 0.9 0.00 0.0 0.02 0.2 0.00 0.5 0.60 0.3 0.00 0.0 0.02 0.1 0.00 0.0 0.55 0.2 0.00 0.7 0.10 0.5 0.02 0.3 0.25 1.4 0.02 16.5 0.10 0.4 0.00 0.2 0.15 1.7 0.02 36.6 0.10 0.9 0.00 0.3 0.15 3.5 0.05 97.0 0.00 1.2 0.00 0.3 0.12 5.2 0.05 126.1 0.00 2.2 0.00 0.4 0.16 6.9 0.03 119.5 0.15 2.9 0.00 0.3 0.15 44.0 0.00 120.5 0.00 3.1 0.05 1.7 0.20 76.9 0.00 176.9 0.15 0.6 0.01 2.7 0.15 76.9 0.00 181.4 0.10 0.3 0.05 0.2 0.02 81.9 0.40 184.4 0.10 5.3 0.00 84.1 0.40 190.7 0.10 14.0 0.00 87.0 0.10 196.7 0.15 27.4 0.12 91.3 0.15 54.2 0.02 92.4 0.01 61.1 0.10 92.6 0.04 89.4 0.10 90.0 0.05 61.9 0.07 Notes : All measurements are in metres Jam U : Jam is 6-12 m d/s of orange line Green line is 80 m d/s of orange line Jam N : Jam is 5-15 m d/s of orange line Green line is 96 m d/s of orange line 77 4.3.3. Upstream Tracers At jam N all but two of the orange stones had moved from their original positions. The two stationary stones were high on the bar at one end of the deployment line. The fine sands which buried them to a depth of 2.0 cm reflect the fact that at high flow their elevated and peripheral location, placed them in slack water. Of the recovered orange stones 36 % (5) were found in the same bar as the majority of the green stones, some 190.0 m downstream of their original position. One stone was found on the upstream side of the jam, and one close to the downstream side, while the others were found in the reach between the jam and the bar. In contrast, none of the orange stones located upstream of jam U were found downstream of the jam. All of the stones which were recovered were found in their original positions buried by between 2.0 and 60.0 cm of gravel and sand. It is possible that the unrecovered stones passed beyond the reach searched but, given that many of the green stones which moved were found within this reach, this seems unlikely. It is more plausible that the missing stones are in the log jam. 4.3.4. Scour Chains Additional evidence of differential permeability comes from the scour chains which were inserted. Of the original sixteen chains, five were found broken, and seven were not recovered at all. At jam U a chain was recovered intact from its position 15.0 m upstream of the jam in the centre of the wedge. It recorded 8.0 cm of scour, followed by 30.0 cm of fill. This and the in situ burial of many of the tracer particles reflect the aggradation at this site and in turn the low permeability of the jam. At jam N three chains were inserted within approximately 5.0 m of the jam but were not found in 1990. Morphological changes suggest that these were scoured out completely. The fourth chain was recovered intact from its position 10.0 m upstream of the jam in the centre of the bar. It recorded 26.0 cm of scour and 36.0 cm of fill (net fill of 10.0 cm). It appears that this sediment wedge is degrading in an upstream direction as flow is forced laterally across the front of the jam and into a channel close to the right bank. This upstream degradation reflects the increased permeability of the aging jam. The final two chains were recovered from the downstream side of N. One recorded a net scour of 5 cm (which is only one or two particle diameters) and the other net filling of 25 cm (47 cm of erosion and 72 cm of fill). This downstream aggradation is in keeping with the transfer of stored material through the jam and its deposition in less active parts of the channel complex. In spite of uncertainty about the whereabouts of the unrecovered stones, the data which are available strongly suggest that jam N is relatively more permeable than jam U. The contrast of relative displacements upstream and downstream is striking and suggests that the transport rate through jam N is in keeping with local rates, but that the transport rate through U is significantly less than normal. 4.4. Bed Material Characteristics 4.4.1. Surface and Subsurface Samples According to the a priori ideas above, the differences in sediment transport in the vicinity of these two jams should be reflected in the texture of the bed materials. The low permeability at U is expected to produce 79 80 Figure 14. Survey photographs of the bed surface at Jam Upper Riley a) Upstream b) Downstream Figure 15. Sediments in the vicinity of Jam K K , Upper Rilev 81 b) From the jam looking downstream (buckets indicate sampling positions) contrasting textures upstream and downstream, while the higher permeability at N should be associated with less textural differentiation. Surface and subsurface samples collected upstream and downstream of the three study jams support this trend, although strict comparisons of the distributions are not possible because no replications were made. Figure 13 shows the surface material texture at the three jams. At jam U (and at jam K K ) the downstream sediments are considerably coarser than those upstream. This is also apparent in survey photographs of the surface upstream and downstream of U (Figure 14), and in general views of jam K K upstream and downstream (Figure 15). In contrast the surface distributions at jam N are very similar. The subsurface material downstream of a jam is not expected to show great variability as a jam forms and disintegrates, since the progressive development of armour at the surface isolates it from the flow. However, upstream fining is expected, since the aggrading wedge is predominantly formed of the most mobile fractions of the bed load. Consequently while an upstream and downstream difference in subsurface material is expected at impermeable jams, its magnitude is expected to be less than that of the surface difference. As with the surface material the downstream transfer of the upstream sediment will moderate this difference as the jam ages. Full subsurface distributions have been constructed using the technique described in section 3.1, and are shown in Figure 16. These distributions confirm the above expectations. At jam U the upstream material is finer than that downstream but the magnitude of the difference is not as large as at the surface. In contrast, at jam N, the sample distributions are very similar and the subsurface materials are unlikely to be significantly different from each other. 83 4.4.2. Photographic Survey Data Additional evidence in support of the argument that grain size can be significantly affected by LOD is found in the photographic survey data. During the course of the survey all the log jams encountered were characterized and their positions recorded. These data are used in more detail in Chapter 5 but can be drawn upon here to indicate the abundance and significance of the impact of LOD on sediment texture. In Figure 17, the surface D 5 0 estimates (in mm) are plotted against distance for the four photographed reaches. Along the top of each plot arrows indicate the location of debris jams. Different sized dots above the arrows indicate the permeability of each jam. Consider the highly impermeable jam (jam GG) at 1640 m in Upper Gregory Creek. This jam is approximately 7 m high and crosses the full channel width (15m), a wedge approximately 200 m long has developed upstream and shows few signs of degradation (Figure 18a). Sediments in this wedge are fine and show some tendency toward coarsening as one moves upstream. In contrast, on the downstream side of the jam, surface material is very coarse for some 200 to 300 m. The surface materials remain relatively coarse for at least another 800 m, except for a couple of finer sites at approximately 510 and 820 m. In addition to the coarse texture, the volume of sediment in this reach is very low and sampling sites were difficult to find (Figure 18b). Large stretches are in bedrock and in general the reach is degraded as far as the tributary which enters the photo reach at 250 m. This is an extreme example of how a major jam can starve the downstream reach of sediment and lead to severe changes in grain size and Figure 17. Surface grain size variations a) Riley Creek photo reaches LOWER RILEY 85 120 100 E E o m Q 01 o SB 1 40 w 20 80 60 j i ii n i «± JL i i i i i i f i * _ i _ _ i _ 200 400 600 800 1000 1200 Distance upstream (m) 1400 1600 1800 UPPER RILEY 140 120 ~ 100 i 8°L I in • 6 0 o <o 5 40 20 H i i _j L J 200 400 Distance upstream (m) Surface Dsn with 95% confidence limits | Log Jam, span* integrity 8-10 5 - 7 5 4 86 b) Gregory Creek photo reaches CO 160 140 120 100 80 60 40 20 LOWER GREGORY i Ul 11 0 m u i i i x i u i i n 200 400 600 800 1000 Distance upstream (m) 1200 1400 UPPER GREGORY 280 260 240 220 200 180 | 160 S 140 Q y 120 w 100 80 60 40 20 i n If i i u I f 11 ff GG 1 f 200 400 600 800 1000 1200 Distance upstream (m) 1400 1600 1800 2000 87 Figure 18. Sediments associated with Jam GG, Upper Gregory a) Looking across the upstream sediment wedge toward the jam as indicated by prolific alder growth in middle distance (note person for scale). b) Coarse surface texture approximately 900 m downstream of jam (note people for scale). 88 Figure 19. Sediments in the vicinity of Jam 0, Lower Gregory. a) Looking from left bank across sediment accumulation upstream of the jam which is just off left of photo (open wedge is approximately 10.0 m wide). b) Looking from left bank toward degraded bed immediately downstream of the jam (indicated by overlying windthrow at bottom right). sediment storage. No other jams of this magnitude were found in either Riley or Gregory Creek. Jam G G represents an end member of a range of impacts which LOD can have. A more moderate effect occurs at jam 0 in Lower Gregory Creek where downstream coarsening over a much shorter distance is apparent. Two photos taken from a high bankside overlooking the jam (Figure 19) show clearly the fine upstream aggradation and coarse downstream degradation evident in Figure 15. 4.5 Summary Individual log jams can, depending on their permeability, have a significant impact on local sediment texture. In addition log jams are very common, occurring at a frequency of approximately 1 per 90 m in the photo reaches. The significant impact and abundance of log jams suggest that they are the dominant cause of the perturbations reported in Chapter 3. In turn, a knowledge of jam location and characteristics may be the key to a more powerful model of textural variation than one based on distance alone. 5. ALTERNATIVE PREDICTIVE MODELS The operational model of grain size variability which is being sought cannot be based on distance alone even within individual links (Chapter 3). However, significant structure is apparent in the vicinity of some log jams. In this chapter attempts are made to incorporate this local structure into a more powerful operational model of textural variation than the Sternberg model. Two possibilities are examined: general relations between jam type, jam proximity, and grain size; and the stratification of samples on the basis of position relative to log jams, in an attempt to smooth perturbed Sternberg signals. 5.1. Grain Size and Jam Proximity This section examines whether any general relations between grain size and jam proximity can be formulated. If so then a knowledge of the positions and characteristics of the log jams within a creek would be the basis of a predictive model of grain size variation. In accordance with the explanation of jam behaviour in section 4.1, one expects to see the relations shown in figure 11 as permeability varies. The existence of such relations was investigated in the following manner. For each of the sixty log jams encountered in the photographically surveyed reaches, the distances from the upstream side of the jam to the closest sample upstream, and from the downstream side of the jam to the closest sample downstream have been calculated. These distances (Table 14) were then plotted against the respective grain size estimate as in Figure 20. Table 14 a). Grain Size and Proximity to Log Jams in Riley Creek Jam Upstream sample Downstream sample Distance Surface D50 Distance Surface D50 Lower reach RA 14 31.6 4 27.1 RB 37 30.5 0 31.6 RC 18 35.3 6 33.8 RE/F 7 38.1 0 39.4 RG 24 28.6 6 19.6 R H 25 32.2 14 28.6 RI 19 37.7 6 20.1 R K / L 14 37.8 26 47.5 R M 1 40.8 8 37.8 RM' 7 47.8 7 34.8 RM1 0 19.2 41 18.0 RN 30 14.9 28 37.5 RO 37 55.3 15 67.2 RP 2 16.1 1 44.3 RQ 0 36.3 17 34.8 RR 17 76.1 13 36.3 Upper reach R A A 8 27.5 7 82.7 RZ 18 36.3 3 27.5 RY 16 82.1 2 36.3 RX 3 19.2 6 82.1 RW/V 2 51.3 3 117.0 RU 16 21.4 0 51.3 RT 1 7.1 5 34.8 RS - - 7 7.1 continued... Notes : i) Distances measured along thalweg from upstream / downstream edge of jam. ii) Surface D50 estimated from photographs. 92 Table 14 b). Grain Size and Proximity to Log Jams in Gregory Creek Jam Upstream sample Downstream sample Distance Surface D50 Distance Surface D50 Lower reach GC 20 47.5 7 57.3 GD 20 62.7 34 47.5 GE 0 51.3 9 37.0 GF 4 15.6 15 51.3 GG 1 64.0 11 52.0 GH 0 33.6 24 64.0 Gl 15 42.2 20 33.6 GJ 1 27.7 0 22.3 GK 23 47.5 24 27.5 GL/M 23 39.4 13 47.5 GO 45 13.0 16 137.5 GP 57 79.3 27 40.2 GQ 24 71.5 44 68.6 GR 17 42.2 13 71.5 GS 23 48.8 12 42.2 GV 1 62.7 41 32.5 Upper reach GW 0 84.5 . -GX/Y 5 52.7 28 39.7 GZ 72 66.3 14 52.7 GAA 62 79.9 7 23.4 GBB 7 49.9 11 79.9 GCC 2 75.1 47 49.9 GDD 2 46.9 38 75.1 GEE 11 55.7 13 54.6 GFF 126 127.1 34 55.7 GGG 15 22.5 141 144.0 GHH 16 32.2 13 44.6 Notes : i) Distances measured along thalweg from upstream / downstream edge of jam. ii) Surface D50 estimated from photographs. When all log jams are plotted together no relation is evident. Given that the effect on sediment texture varies between jams, this is not unexpected. The data were therefore stratified into permeability categories. Permeability measurements are not available for the sixty jams encountered but section 4.3 confirms that permeability is inversely related to jam integrity and span, and positively related to jam age and number of channels. Each of these surrogate variables was assessed semi-quantitatively in the field using the classification scheme developed by Hogan (1989) and presented in Appendix 1. The logjam characteristics for the four reaches are presented in Table 15, where higher values generally reflect higher permeability. Age does not vary very widely, but number of channels, sediment storage, and a composite span and integrity index were used to stratify the data into permeability categories. The grain-size versus jam-proximity plots were then constructed using different symbols to represent the integrity group of the jam with which each sample D 5 0 is associated (Figures 20, 21, 22). One should point out here that similar analysis was done using data normalised relative to local grain size, aswell as jam-type. The results were similar to those disussed below and are not therefore presented. The expected outcomes are only apparent in a few cases. In Figure 22 (number of channels) those jams broken by only one channel do show a tendency toward coarsening as one moves away from the jam in an upstream direction. In Figures 20 and 21 (span + integrity and sediment storage) the lowest permeability classes do show relatively fine material close to the upstream side of the jam and relatively coarse material close to the downstream side. Other than this, no trends are apparent. There are several reasons for this: 94 Table 15 a). Log Jam Characteristics of the Photo Reaches in Riley Creek Jam Dstr. Ustr. Span Intg. Age Chan. Sed. Lower reach RA 4 6 5 3 3 3 3 RB 20 46 3 3 4 2 3 RC 146 210 2 4 5 + 3 3 RD 160 210 3 4 2 2 3 RE 272 300 5 5 5 + 4 4 RF 316 328 4 4 5 + 2 4 RG 442 445 5 4 5 + 1 4 RH 483 487 4 4 5 + 2 4 RI 566 586 2 5 5 + 3 3 RK 715 717 3 5 5 + 1 5 RL 737 743 2 3 5 + . 1 2 RM 765 844 5 5 5 + 3 5 RM' 875 968 2 4 5 + 4 3 RM1 1033 1034 5 5 5 + 2 4 RN 1229 1248 2 3 5 3 3 RO 1346 1355 2 4 5 + 2 4 RP 1488 1490 3 4 3 2 4 RQ 1554 1558 3 4 5 + 2 4 RR 1601 1612 3 3 5 + 2 3 Upper reach RS 300 308 4 5 5 + 1 4 RT 278 292 5 5 1 3 4 RU 176 187 1 1 2 2 1 RV 126 135 5 5 5 + 1 5 RW 121 132 5 5 2 1 5 RX 89 106 3 4 2 2 4 RY 64 67 5 5 5 2 4 RZ 33 44 2 3 2 2 3 RAA 7 22 2 3 2 2 2 Notes : i) See Appendix 1 "Hogan's Jam Classification Scheme' for explanations of the parameters and categories used. ii) Dstr. :distance from starting point to downstream edge of jam (m) Ustr. :distance from starting point to upstream edge ofjam (m) Intg. :integrity Chan. :number of channels Sed. : sediment storage upstream 95 Table 15 b). Log Jam Characteristics of Photo Reaches in Gregory Creek Jam Dstr. Ustr. Span Intg. Age Chan. Sed. Lower reach GC 7 9 4 4 5 3 5 GD 63 64 3 4 5 + 2 4 GE 225 230 5 5 5 + 3 5 GF 245 255 3 4 5 + 4 3 GG 274 280 4 4 5 + 4 4 GH 305 315 4 4 5 + 5 4 Gl 335 356 4 4 5 + 5 3 GJ 466 521 3 4 5 + 4 3 GK 546 590 2 3 5 + 5 3 GL 626 636 5 4 5 + 5 5 GM 669 686 4 4 5 + 5 2 GN 708 716 5 4 5 + 3 4 GO 763 778 3 3 5 + 5 2 GP 936 941 4 4 5 + 1 5 GQ 1164 1174 4 5 5 + 3 5 GR 1211 1298 3 4 5 + 5 2 GS 1327 1332 5 4 5 + 1 5 GT 1353 1360 5 5 5 + 1 5 GU 1367 1370 2 3 5 + 3 3 GV 1428 1446 3 4 5 + 4 2 Upper reach GW - 0 3 4 5 + 2 3 GX 135 150 3 4 5 + 2 3 GY 165 171 4 4 5 + 3 3 GZ 209 231 5 5 4 1 4 GAA 514 517 3 4 3 1 4 GBB 590 598 2 4 5 + 2 4 GCC 652 656 4 4 5 2 4 GDD 696 698 3 2 5 + 2 3 GEE 1185 1200 3 4 5 + 3 4 GFF 1245 1265 5 4 5 + 1 5 GGG 1610 1645 1 1 2 1 1 GHH 1828 1829 5 5 5 1 5 Figure 20. Grain size in the vicinity of jams classified by span and integrity c o E a? o c E E o o H 1 1 1 h o <>%. • m o lO Q a) $ CO • • f © o • • + c <o 1 OO 1 o 1 <0 Q. <0 V • §1 8 o CO o o Figure 21. Grain size in the vicinity of jams classified by 97 sediment storage o CM E co O O E CO o o co o •3-c cu E co CJ c o CN I 9 y-H 1 h / • • o o , • O CO Q co & co o • o - •» • O CM GO a o co CO o c CO o CD 6 CO f i 8> 5 0 o • o • CO c CD 6 "6 CO co CM co in o oo o o I CM E CO CO w oo c 5 o Q co o c CO 98 c CD E b c 6 E 9 r-O m O 3 CO f i Figure 22. Grain size in the vicinity of jams classified by number of channels © © © © © o o '•.<> <* o o-© ® . 0 -£.©© o o ® ^ © ® o c c -c O « •O E © o • 7- CM n T i n o CM o o o CO o co o o o CM o o co o co 8 o O J e CO E CO CO k. co a CO o c « b E CO E CO cu w c 5 o Q 0) o c CO 1. There are very few young, impermeable jams, such as U and GG, in the two study creeks, and these have the strongest impact on bed material texture. 2. Understanding of log jam-behaviour is incomplete, and it is possible that the surrogate variables used do not reflect sediment transport through log jams well enough to make distinctions among the jams with relatively high permeability. 3. The data are not of a sufficiently high frequency to be able to distinguish the upstream effect of one jam from the downstream effect of another. If one is to maintain a systematic sampling scheme which avoids variation at the pool-riffle scale, then the number of possible sampling sites is of the same order of magnitude as the number of log jams (see Table 17). Consequently many of the data plotted on the downstream side of Figures 20, 21, and 22, is confounded by the upstream effect of log jams a short distance downstream, and vice versa. One can identify several log jams which are isolated and therefore show clearly the relative grain size variability expected, jam GG for example (section 4.4.2.), but there are insufficient of these to permit the development of any general grain-size proximity relations for these creeks. One cannot therefore use information on log jams alone to predict grain size in these creeks. 100 5.2. Grain Size, Sample Position in Relation to Log Jams, and  Distance Downstream If log jams have a strong and consistent effect on grain size then it should be possible to remove the variation associated with log jams, just as the variation associated with pool-riffle variability was removed by the systematic sampling of riffle-pool breaks. It may be that between the like depositional units downstream of jams a systematic change in grain size is evident with distance. In particular the latent Sternberg trend suggested by the longitudinal and photo survey data, may be revealed. According to Sternberg, grain size changes with distance because particles of different size vary in their mobility. At any given point the abundance of the most mobile (fine) particles tends not to be limited, and it is therefore the least mobile (coarse) particles that determine the average grain size. The lag deposits downstream of local log jams in theory represent the coarsest part of the bed material present in the vicinity of the jam. In as much they consist of the sediments which determine local grain size and are therefore most likely to reveal any downstream changes. The downstream jam environment is used here just as bar heads were in many previous studies of the Sternberg phenomenon. The photographic survey, median surface estimates, have been stratified according to their location in relation to local jams. The data were carefully filtered in order to maximise the chances of identifying any trends. First, only those samples which are separated from the next jam downstream by at least one other sample were considered. This was done order to try and ensure that only samples clearly associated with the 101 downstream effect of an upstream jam (and not the upstream effect of a downstream jam) are included in the analysis. In addition the permeability of the jams was considered, since the depositional environment downstream of a relatively impermeable jam is very different from that downstream of a relatively permeable jam. For each jam the sum of the span and integrity classifications was used to indicate relative permeability. Samples adjacent to jams of high permeability (span + integrity 8-10) were excluded; those adjacent to jams of moderate permeability (span + integrity 5-7) were included if they were located within 20 m of the jam; and samples adjacent to jams of low permeability (span + integrity 2-4) were included if they were located within 50 m of the jam. This filtering severely limits the number of data, but is necessary given that our aim is to compare sedimentary units which differ only in terms of longitudinal position within the creek. Lower Riley is the only reach which retained sufficient data to enable reasonable inferences to be made, and the median surface grain sizes of the similar downstream sites are plotted against distance in Figure 23. Although a significant amount of scatter is removed the data show no systematic trend and the overall change in grain size is not significant. The removal of the two upstream samples would produce a moderately consistent downstream decline. However there are no more grounds for doing this than there are for removing the relatively coarse sample at approximately 1320 m which is largely responsible for the resultant fining trend. This plot represents an attempt to remove all of the sources of variation which have been identified, yet the data deviate from the 102 Figure 23. Median surface grain size at similar downstream jam  environments. Smaller dots without error bars are the other medians. o 1 § o _ o CO CM O 00 CO CN (LULU) OSQ eoe;jng exponential decline predicted by Sternberg. This suggests that non-structured sources of variation remain dominant. Non-alluvial sources of sediment and the reworking of relict jam sediments may be important in this respect. In addition the data do not define the coarsest materials present as one moves downstream (this was also apparent in both of the reaches in Gregory) despite explicit attempts to ensure that this was the case. One must therefore question our understanding of the controls on grain size in these creeks, or at least our ability to isolate them. In particular, the nature of sediment storage and transfer, as it is affected by highly frequent and functionally distinct log jams, is very complex and is unstructured at the basin scale. 5.3. Summary Both unidentified factors and our limited understanding of the system are responsible for the inability to reduce the variation any further. It is reasonable to conclude that a simple deterministic model of grain size variation is not yet a possibility, and that sediment texture may behave stochastically in these creeks. 6. OPERATIONAL RECOMMENDATIONS In the previous three chapters, the variation of grain size at the drainage basin scale in headward creeks on the Queen Charlotte Islands has been examined . Attempts to identify structure, and relate it deterministically to causative factors (distance, log jams) have been unsuccessful. I remains possible that grain size variations in these creeks is a stochastic phenomenon. If this is the case one can develop suitable sampling recommendations which will ensure the accurate assessment of basin-wide bed material characteristics. However, it is possible that unexplained structure is present in the data, that is, systematic variation which has not been covered in the sedimentation models previously investigated. It is necessary to investigate this possibility before the data can be classified as stochastic. This is done using a general test of randomness for two commonly used bed material characteristics: the median surface grain size, and the Fredle index. Prior to this one must examine the variability of these characteristics in relation to land use and valley-flat buffering, because these factors may introduce significant high-level structure with implications for the generality of results and the development of sampling criteria. 6.1 The Effect of Land Use and Confinement 6.1.1 The Fredle Index The Fredle index (Lotspeich and Everest, 1982) is used by fisheries biologists to appraise the spawning quality of streambed gravels. It is regarded as a better index of substrate quality than percent fines measures because, by incorporating measures of both central tendency and spread, it relates more directly to permeability and porosity (Lotspeich and Everest, 1982). It is the ratio formed by the geometric mean and the standard deviation of the truncated distribution. In the original formulation the geometric mean is calculated by the method of moments, and the Trask sorting coefficient is used to approximate the standard deviation: FI = ( d x w l * d 2 w 2 d n w n) / (D 7 5 /D 2 5) 0 - 5 where d is the midpoint diameter of particles retained by a given sieve (mm), w is the fraction by weight of particles retained by a given sieve, D75 is the seventy-fifth percentile (mm), and D25 is the twenty-fifth percentile (mm). As with all fisheries indices it must be applied to samples truncated at a common limit if between-sample comparisons are to be meaningful, and was therefore calculated using the original bulk samples which were truncated at 64.0 mm in the field. The Fredle indices of the bulk samples collected in Riley and Gregory Creeks are presented in Table 16 along with various within-group statistics. The eighteen bulk samples collected along the main stem of Riley Creek are drawn from contrasting sedimentary environments. In the lower portion of Riley, below RLS 9 (Figure 3), the channel is relatively unconfined by steep hill-slopes and no incidental mass movements are apparent. In contrast, the upper portion of the creek, above RLS 9, is flanked by steep slopes and several debris torrent egresses can be identified (Figure 2). Thus the upper portion is intermittently supplied with colluvial material, while the lower portion is largely dependent on fluvial transport 1 0 6 Table 16. Fredle Indices for sub-64.0 m m Bulk Samples Site Freddie Index n Mean S.D. Variance RLS 1 2.49 RLS 2 2.57 RLS 4 3.25 RLS 5 3.88 RLS 6 3.44 TI DS 4.44 TI US 3.00 RLS 7 2.67 RLS 8 5.05 T2 DS 2.50 T2 US 1.91 RLS 9 2.68 RLS 10 3.39 T3 DS 3.33 T3 US 4.48 RLS 11 4.38 T4 DS 3.36 T4 US 4.96 GLS 1 4.75 GLS 3 3.81 GLS 5 4.91 GLS 7 3.97 GLS 9 4.34 RLS 1 -R4 US 11 3.20 0.9412 0.8859 3.79 0.8099 0.6599 5 4.36 0.4768 0.2274 18 3.43 0.9200 0.8464 107 for its sediment (numerous non-fluvial bank exposures are apparent along the lower portion). There are therefore grounds for supposing that the textural variability will differ, and that the lower and upper data should be analysed separately. This was checked by comparing the variance of the Fredle indices upstream of RLS 9 (inclusive) with that of the indices downstream of RLS 9. The resulting F value is 1.35 which is not significant at a = 0.05 with 10 and 6 degrees of freedom (F c rjT J = 4.00). Consequently one can be sure that the relative confinement has no effect on the variation of the Fredle indices, and that in turn the analysis can be applied to the whole set of data. Although the two means are not significantly different either (t-test, a = 0.05, 16 degrees of freedom), these findings do not contradict the argument presented in chapter 1 because, despite its "buffered" situation, lower Riley exhibits a frequency of special storage mechanisms (log jams) similar to the upper portion. This is in itself an indication of the significance of log jams in causing variability. It was pointed out in section 2.2.1 that logging practices may affect the nature of bed material variation by affecting mass-movement frequency and the in-stream arrangement of large organic debris. A comparison of the variances of the eighteen Fredle values in Riley (logged) and the five Fredle values from Gregory (unlogged) suggests that the variability of the indices is not significantly different (F-test, a = 0.05, 17 and 4 degrees of freedom). However, the mean Fredle in Gregory is significantly greater than in Riley (t -test, a = 0.05, 21 degrees of freedom). This suggests that Riley has poorer spawning substrate than Gregory, which is consistent with some studies of the effect which logging practices have on subsurface quality. Some caution should be maintained since the number of observations in Gregory is relatively small, and the absolute difference in the mean may not exceed a critical distinction in terms of fish survival rates. In the context of this study this finding is important, since differences in either mean or variance indicate that sampling criteria must be specific for the distinctive groups. 6.1.2 Median surface grain size. The surface DgQS of the longitudinal and photographic surveys are presented in Tables 8 and 10. In the following analysis a single, exceptionally coarse outlier was removed from the Upper Gregory photogrphic data set. Summary statistics for the four photographic reaches, and various groupings of them are given in Table 17. The variance of the median surface grain size in the Upper photo reach in Riley, is significantly greater than in the lower reach. (F-test, a = 0.05, 12 and 41 degrees of freedom). Therefore , although the means are not significantly different (t-test, a = 0.05, 53 degrees of freedom), relative confinement has an affect on surface texture, and the two reaches must be considered separately. As expected, no such effect is apparent in Gregory, where neither variance (F-test, ct = 0.05, 33 and 25 degrees of freedom), nor mean (t-test, a = 0.05, 60 degrees of freedom), is significantly different between the two reaches. Furthermore the data from Upper Riley are consistent with those in Gregory (F-test, a = 0.05, 12 and 59 degrees of freedom; t-test, a = 0.05, 71 degrees of freedom), and so one can conclude that logging has no effect on surface median grain size. The surface data for Lower Gregory, Upper Gregory, and Upper Riley can then be regarded as a homogeneous T a b l e 17. C h a r a c t e r i s t i c s o f s u r f a c e D50 i n the photo reaches Reach n mean s.d. var. A B U. Riley 13 43.9 32.16 1033.96 23.1 33.0 L. Riley 42 36.6 14.03 196.98 38.4 84.8 U. Gregory 26 58.2 29.08 892.90 70.3 152.3 L. Gregory 34 48.6 26.74 714.85 42.5 72.3 Gregory (U.G. + L.G.) 60 52.7 28.31 801.23 Gregory + U. Riley 73 51.2 28.99 840.34 Notes A B Average distance (m) between samples Average distance (m) between log jams set of data, which is distinguished from the surface material in Lower Riley by its greater variability. 6.2 A Test of Randomness : Runs Up and Down 6.2.1. Rationale of the test Individual values within a sequence of data can be classified as either "greater than the preceding value" or "less than the preceding value". In turn one can determine the number of runs in the data, where a run is defined as an unbroken sequence of increasing or deceasing values. If the process responsible for the sequence is non-random, then the arrangement of the n observations in the sequence is predisposed to follow a particular pattern. The number of runs which occur is then likely to be significantly greater than, or significantly less than, the number of runs which would occur if the n! possible arrangements are equally likely, i.e. if the generating process is random. Consequently the number of runs in a sequence of data can be used to test its randomness. A random process will tend to produce a number which is neither very large nor very small relative to the number of runs possible (n-1). A small number of runs indicates very little fluctuation, and systematic behaviour in the variate. Conversely, a very high number of runs reflects oscillatory behaviour of a frequency similar to that of the sampling, and is similarly non-random. 6.2.2 Analysis Tables in Bradley (1968) provide the probability that, for a given number of observations, the total number of runs will be r or fewer (where r 111 is the observed number of runs). For the longitudinal survey data of Gregory, and the upper part (above and including RLS 9) and lower part of Riley, the surface median grain sizes were classified into runs, and the number of runs counted. This was also done for the Fredle indices of Riley and Gregory. The results are summarised in Table 18. For the surface materials the Lower Riley set is definitely non-structured. In Gregory, and Upper Riley, however, the numbers of runs are relatively high, and the probabilities of getting more runs (as indicated by 1-p) are relatively low. Although this value does not approach a statistically significant limit (0.025, at. a = 0.05) , in both cases the possibility that some high frequency structure is present should be investigated. This is possible using the photographic data sequences which are nested within the longitudinal surveys (Figures 3 and 7) and have a much higher sampling frequency. The photo survey reaches consist of between 13 and 42 samples, and cover the length of channel from which between 2 and 5 longitudinal samples were drawn. For the two channel lengths in question (Gregory and Upper Riley) one sequence of twenty-five observations from each of the two photographic reaches in Gregory, and one of thirteen observations from Upper Riley, were analysed (Table 19). In Upper Riley and Lower Gregory the data show no structure, but in Upper Gregory the sequence again shows a high frequency oscillation (6% chance of more runs than the number observed). It is therefore certain that there is no low-frequency structure in the median surface grain sizes. However, there is evidence in Upper Gregory of high-frequency structure close to the scale of bar to bar spacing. Bar spacing is significantly shorter in this reach than elsewhere because of the degradation associated with jam GG. This is reflected in the relatively low 112 Table 18. Runs tests (low frequency) Surface Median Fredle Index U. Riley L.Riley Gregory Riley Gregory 7 11 12 18 5 5 7 9 11 4 0.8921 0.6460 0.9179 0.4568 1.0000 0.1079 0.3540 0.0821 0.5432 0.0000 Notes : n number of observations r number fo runs p probability of getting r or fewer runs U.Riley Riley above and including RLS 9 L.Riley Riley below RLS 9 Table 19. Runs tests (high frequency) L. Riley U. Riley L. Gregory U.Gregory n 25 13 25 25 r 18 8 18 19 P 0.8577 0.5413 0.8577 0.9436 1-p 0.1423 0.4587 0.1423 0.0564 Notes : n number of observations r number fo runs p probability of getting r or fewer runs sampling density of this reach where one sample was collected every 70.0 m (on average), compared with 42 m in Lower Gregory, 23 m in Upper Riley, and 38 m in Lower Riley. It is also notable that jam spacing is higher than elsewhere (152 m, compared with 72 m, 33 m, and 85 m). One possible explanation is that in the absence of much sediment, accumulations are largely restricted to log jams, which in turn produce the oscillatory signal if, on average, one sample is collected upstream and one downstream of each jam. This is consistent with the frequency of jams being approximately half that of the sediment accumulations, and to a limited extent is evident in Figure 15. Whatever the exact cause it is perhaps wise to treat reaches such as Upper Gregory as special cases, given the impact of these very large jams. Although the jam does not significantly affect the texture of the material present over the entire reach, it does seem to affect its arrangement within the reach. The Fredle indices in Riley certainly exhibit no structure. Although the results from Gregory indicate a highly non-random oscillating structure, one cannot investigate this further without more bulk samples. The small number of data is a limiting factor, and because of this one cannot conclude that oscillatory structure is present. There is more credibility, however, in an assertion that there is no low-frequency structure in the signal. 6.3 Sampling Recommendations 6.3.1 Normality In establishing any sampling criteria one must assess the distribution of the variable being sampled, since the sampling strategy needed to characterise the data set will vary depending on whether the data are normally distributed. After grouping data according to standard deviation ranges, a Chi squared test can be used to compare the observed distribution with an expected normal distribution for the given number of observations. This was done for the Fredle indices from Riley, Lower Riley photo reach surface DQQS , and the combined surface D5QS from Lower Gregory, Upper Gregory, and Upper Riley photo reaches. It was necessary to lump together some standard deviation groups in order to meet the requirements of the test. All three sets of data are approximately normally distributed; Riley Fredle and the three like-photo reaches at a = 0.10, and Lower Riley surface at a = 0.05. 6.3.2 Sample Size Given the approximate normality of the data, the number of observations needed to obtain an estimate of the general population characteristics can be calculated. One assumes that the data in hand provide best estimates of the true population standard deviation (s) and mean (x) so that, given a specified level of precision (f) and a given confidence level (tn) , the number of observations needed (n) is given by, Vn = (tn.s) / (f.x) that is f = (tn.s) / (\/n.x) With a = 0.05, precision levels of 0.10 and 0.20 indicate that an estimate of the mean will be within 10 and 20% of the true population mean in 95 cases out of 100. The number of observations required to obtain these levels are given in Table 20 for the distinct groups of data identified in section 6.1; namely Fredle index in Riley (logged), Fredle index in Gregory(unlogged), 115 Surface DgQ in Lower Riley (buffered), and surface D5Q in Upper Riley, Upper Gregory and Lower Gregory (confined). These levels have been chosen arbitrarily. For the Fredle indices a critical change in value will exist which relates to a significant change in fish survival rates, as defined by a fisheries biologist. This critical distinction should be used to determine more meaningful levels and in turn calculate the number of samples one needs in relation to fishery applications. Confidence in the result for Gregory is severely limited because of the small number of observations upon which the estimates of mean and standard deviation are based. It is not therefore wise to claim that fewer samples are required to characteristic the Fredle index in unlogged creeks than in logged creeks, without further investigation. The greater variability of the surface materials in the confined section is reflected by the larger number of samples needed to characterise them relative to the unconfined sections. 6.3.3 Sampling Strategy The surface median grain sizes and Fredle indices within these creeks vary in a stochastic manner as indicated by the runs tests and the deterministic analysis. They also tend to be normally distributed. Consequently, once such headward creeks on the Queen Charlotte Islands have been classified according to their land use history and confinement, sampling sites can be arbitrarily positioned without fear of sampling bias. That is, no stratification or particular spacing, other than that associated with pool-riffle spacing, is necessary in order to avoid biasing the sample. Such a conclusion is of significant benefit to resource managers since the implication is that samples can be collected from any accessible riffle-116 Table 20. Recommended sample sizes Precision 20 % 10 % Surface D50 a) Confined 30 120 b) Buffered 16 60 Fredle Index a) Logged 9 30 b) Unlogged 3 7 Notes : Recommended number of observations calculated using formula in text with a = 0.05. 117 pool breaks along the creek. This is the case irrespective of land use history, and the situation of the channel with respect to hillslopes. In the latter case individual log jams continue to cause significant variation despite a lack of incidental mass-movements. One should qualify this general conclusion by pointing out that major jams, such as GG in Upper Gregory, affect sediment texture over greater distances than more common jams, and may cause alterations in the organization of texture for many channel widths downstream. While one or two samples could be legitimately collected, it would be unwise to collect all of one's samples in such severely degraded reaches. Consequently they should be identified either by air photo analysis beforehand, or on the ground by recognition of a lack of sediment, prevalence of bedrock, and relative absence of LOD. 6.4 Summary This chapter has established the stochastic sequence and normal distribution of median surface grain sizes and Fredle indices in headward creeks on the Queen Charlotte Islands. The influence of land use history and valley flat buffering has been established, and the number of observations necessary to accurately determine average characteristics within particular land use/confinement categories, have been estimated. In all cases accessibility can determine the location of the sampling sites, with the provision that riffle pool breaks are consistently used, and that severely degraded reaches downstream of major jams are not the only reaches sampled. For the Fredle index the creek should first be classified as "logged" or "unlogged". In the former case nine commonly truncated bulk samples are required to obtain a mean value within 20% of the true mean, 95% of the time. In the latter case it is possible that only three are needed, but this is not recommended given that the result is based on a data set of only five samples. For surface D5Q, the creek should be separated into confined and unconfined reaches (in the majority of cases all of the creek will be confined). One then needs to obtain twenty and sixteen samples respectively in order to achieve the 20% precision, 95% of the time. These sample sizes are larger than those for the Fredle indices and reflect the fact that jams affect surface sediments much more than subsurface material. The jams continue to cause unsystematic variability despite the lack of incidental mass-movements in the unconfined reaches. 119 7. CONCLUSION This study investigated the spatial variation of the bed material texture in two strongly coupled creeks on the Queen Charlotte Islands. The objectives of the study were to assess empirically that part of the proposed basin-wide sediment texture model pertaining to strongly coupled reaches, and to develop an operational model of textural variation for predictive use by resource managers on the Queen Charlotte Islands. Surface and subsurface materials at a number of sites within each creek were sampled using several methods. Both median surface and subsurface grain size exhibit no systematic relation with distance downstream, and in general grain sizes fluctuate widely over very short distances. This is true even when the potential tributary effect is excluded from the analysis by considering the variations within between-confluence links. Although there is limited evidence of a strongly perturbed downstream trend, one must conclude that the general Sternberg model, which describes textural variation in larger, uncoupled systems with varying degrees of success, does not explain textural changes in strongly coupled systems such as these. Thus, empirical evidence from these two strongly coupled creeks on the Queen Charlotte Islands confirm the proposals concerning headward areas incorporated in the basin-wide model of textural variability presented in chapter 1. Detailed consideration of relative sediment transport rates and local bed material texture confirmed initial impressions that, depending on permeability, individual log jams can cause significant variations in bed material textural. In relation to the second objective, an attempt was made to incorporate the local structure associated with some jams into a more powerful model of textural variation than that provided by distance alone. Unfortunately, the high frequency and functional variability of log jams means that at the drainage basin scale, the nature of sediment storage and transfer is very complex. Even when grain size is normalized relative to jam permeability, jam position and ambient texture, variations remain non-structured, and unsystematic. Consequently, at the drainage basin scale, a deterministic model of local grain size remains elusive. Further analysis has revealed that the variations in surface median grain size and Fredle index are stochastic. This is of significant practical consequence since it implies that sampling sites can be located arbitrarily without biasing the sample. Furthermore, the data are normally distributed and the number of observations needed to characterise the population within a given degree of precision can therefore be calculated. These sampling criteria vary depending on land-use history and the position of the channel relative to hillslopes, both of which introduce some high level structure. In particular, confinement has an effect on the variance of surface texture, and land use may affect both the mean and the variance of subsurface texture, as indicated by the Fredle index. Thus, while no basis for a predictive model of textural variation has been found, it has been possible to develop sampling guidelines for use by resource managers in headward creeks on the Queen Charlotte Islands. The generality of these findings in relation to the details of logging, and the local susceptibility of hillslopes to mass movement events, is a topic worthy of future consideration. 121 REFERENCES Adams, J.M., and Beschta, R.L. (1980) Gravel bed composition in Oregon coastal streams. Can. J. Fish. Aquat. Sci., 37(10), 1514-1521. Adams, J. (1978) Data for New Zealand pebble abrasion studies. New Zealand J. Sci. 21,607-610. Alley, N.F. and Thompson, B. (1978) Aspects of environmental geology, parts of Graham Island, Queen Charlotte Islands, British Columbia. B.C. Min. Environ., Resource Anal. Branch Bull. 2, 65 p. Bams, R.A. (1969) Adaptations of sockeye salmon associated with incubation in stream gravels. In T.G. Northcote (ed.) Symposium on Salmon and Trout in Streams, 71-87. H.R. MacMillan Lectures in Fisheries. University of British Columbia, Vancouver. Barrell, J. (1925) Marine and terrestrial conglomerates. Geol. Soc. Am. Bull, 36, 324-337. Bathhurst, J.C. (1985) Literature review on some aspects of gravel-bed rivers; Tucannon River Pilot Study. Report prepared for USDA, Agric. Res. Ser., Soil Cons. Ser., 139p. Bjornn, T.C. (1968) Survival and emergence of trout and salmon in various gravel-sand mixtures. In Proceedings, Forum on the relation between logging and salmon, 80-88. American Institute of Fishery Research Biologists and Alaska Department of Fish and Game, Juneau. Bluck, B.J._(1976) Sedimentation in some Scottish rivers of low sinuosity. Transaction of the Royal Society of Edinburgh, 69, 425-56. Bradley, J.V. (1968) Distribution Free Statistical Tests, Prentice Hall Inc., New Jersey, p 277-281 Bradley, W . C , Fahnestock, R.K., and Rowekamp, E.T. (1972) Coarse sediment transport by flood flow on Knik River, Alaska. Geol. Soc. Am. Bull., 83, 1261-1284. Bray, D.I. (1979) Estimating average velocity in gravel-bed rivers. In R.D. Hey, J.C. Bathurst, and CR. Thorne (eds.) Gravel-Bed Rivers, John Wiley and Sons, Chichester, UK, 517-542. Brayshaw, A.C. (1985) Bed microtopography and entrainment thresholds in gravel-bed rivers. Geol. Soc. Am. Bull., 96, 218-223. Brewer, P.A. and Lewin, J. (1990) Transport modification of alluvial sediment: field evidence and laboratory experiments. 4th International Fluvial Sedimentology Conference, Barcelona, Spain, 1989 Brierly, G.J. and Hickin, E.J. (1985) The downstream gradation of particle sizes in the Squamish River, British Columbia. Earth Surface Processes and Landforms, 10, 597-606. Brush, L.M., Jr. (1961) Drainage basins, channels and flow characteristics of selected streams in central Pennsylvania. United States Geological Survey Professional Paper, 282-F, 145-181. Burner, C.J. (1951) Characteristics of spawning nests of Columbia River salmon. U.S. Fish and Wildlife Service Fish. Bull, 52(61), 97-110. Burns, J.W. (1972) Some effects of logging and associated road construction on northern California streams. Trans. Am. Fish. Soc, 101, 1-17. Cederholm, C.J., Reid, L.M., and Saio, E.O. (1981) Cumulative effects of logging road sediment on salmonid populations in Clearwater River, Jefferson County, Washington. In Proceedings from the Conference Salmon-Spawning Gravel: A Renewable Resource in the Pacific Northwest? 38-74. Rep. 39. State of Washington Water Research Center, Pullman. Church, M. (1983) Concepts of sediment transfer and transport on the Queen Charlotte Islands. Fish/Forestry Interaction Program, Working Paper 2/83, Vancouver, B.C. 30 p. Church, M. (1985) Bed load in gravel bed rivers: observed phenomena and implications for computation. Canadian Society for Civil Engineering Annual Conference, Saskatoon, Proceedings Vol. 18, 17-37 Church, M. and Kellerhals, R. (1978) On the statistics of grain size variation along a gravel river. Can. J. Earth Sci., 15, 1151-1160. Church, M., Mclean, D., and Wolcott, J . (1987) River-bed gravels: sampling and analysis. In Thorne, C.R., Bathurst, J . C , and Hey, R.D. (eds.) Sediment transport in Gravel-bed Rivers, John Wiley and Sons Ltd., Chichester, 43-88. Cooper, A.C. (1965) The effect of transported stream sediments on the survival of sockeye and pink salmon eggs and alevins. Int. Pac. salmon Fish. Comm. Bull., 18, 71pp. Daubree, G.A. (1879) Geologie Experimentale 1, 250-53 Dawson, M. (1982) Sediment variation in a braided reach of the Sunwapta River, Alberta. Unpublished M.Sc Thesis, University of Alberta , 186 pp. Dawson, M. (1988) Sediment size variation in a braided reach of the Sunwapta River, Alberta, Canada. Earth Surface Processes and Landforms, 13, 599-618. Desloges, J.R. (1990) Sources for valley-fill alluvial sediments: Bella Coola River Basin, British Columbia. Can. Geog., 34 (2), 98-109. Duncan, S.H. and Ward, J.W. (1985) The influence of watershed geology and forest roads on the composition of salmon spawning gravel. Northwest Sci., 3, 204-212. Einstein, H.A. (1950) The bedload function for sediment transportation in channel flows. U.S. Dep. Agric. Tech. Bull. 1026, 1-78. Everest, F.H. and Meehan, W.R. (1981) Forest management and i fish habitat productivity. Transactions of the 46th North American Wildlife and Natural Resources Conference, 521-530. Everest, F.H., Beschta, R.L., Scrivener, J . C , Koski, K.V., Sedell, J.R., and Cederholm, C.J. (1987) Fine sediment and Salmonid production: A paradox. In E.O. Saio and Cundy Streamside management; forestry and fishery interaction. Fenton, J.D. and Abbott, J .E. (1977) Initial movement of grains on a stream bed: The effect of relative protrusion. Proc. R. Soc. London Ser. A 352, 523-527. Gimbarzevsky, P. (1986) Regional inventory of mass wasting on the Queen Charlotte Islands. B.C. Min. For., Land Manage. Rep. Hogan, D.J. (1986) Channel morphology of unlogged, logged, and debris i streams in the Queen Charlotte Islands, British Columbia. B. C. Min. For., Land Manage. Rep. 49, 94p. Hogan, D.L. (1987) The influence oflarge organic debris on channel recovery in the Queen Charlotte Islands, British Columbia, Canada. Erosion and Sedimentation in the Pacific Rim (Proceeding sof the Corvallis Symposium), IAHS Publ. no. 165, 343-353. Hogan, D.L. (1989) Channel responses to mass wasting in the Queen Charlotte Islands: A temporal and spatial evaluation of stream morphology. Progress Report prepared for the Fish/Forestry Interaction Program, Geomorphology Sub-Committee, 43 p. Hogan, D.L. (1989a) Channel responses to mass wasting inputs, Queen Charlotte Islands, British Columbia. Fish/Forestry Interaction Program Phase II; Watershed 89; A conference on the Stewardship of Soil, Air, and Water Resourses, Juneau, Alaska. 22 p. Ichim, I, and Radoane, M. (1990) Channel sediment variability along a river: A case study of the Siret River (Romania). Earth Surface Processes and Landforms, 15, 211-225. 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. Keller, E.A., MacDonald, A., and Tally, T. (1981) Streams in the coastal redwood environment: The role oflarge organic debris. In R.N. Coates (ed.) Proceedings of a Symposium on Watershed Rehabilitation in Redwood National Park and Other pacific Coastal Areas, 167-176. Center for Natural Resource Studies, John Muir Institute, Inc. Knighton, A.D. (1980) Longitudinal changes in size and sorting of stream-bed material in four English rivers. Geol. Soc. Am. Bull, 91, 55-62. Knighton, A.D. (1984) Fluvial Forms and Processes, 78-80. London: Edward Arnold Ltd. Komar, P.D. (1987) Selective grain entrainment by a current from a bed of mixed sizes: A re-analysis. J. Sediment. Petrol. 57, 203-211. Komar, P.D. and Li, Z. (1986) Pivoting analysis of the selective entrainment of sediments by shape and size with application to gravel threshold. 33, 425-436. Krumbein, W.C. (1937) Sediments and exponential curves. J. of Geol. Vol 50, No. 6, 577-601. Kuenen, Ph. H. (1956) Experimental abrasion of pebbles: 2 Rolling by current. Journal of Geology, 64,336-368. Laronne, J.B. and Carson, M.A. (1976) Interrelationships between bed morphology and bed material transport for a small gravel bed channel. , 23, 67-86. Lisle, T.F. (1986) Stabilization of a gravel channel by large streamside obstructions and bedrock bends, Jacoby Creek, northwestern California. Geol. Soc. Am. Bull, 97, 999-1011. Lisle, T.F. (1987) Overview: Channel morphology and sediment transport in steepland streams. Erosion and Sedimentiation in the Pacific Rim (Proceedings of the Corvallis Symposium), IAHS Publ. no. 165, 287-297. Lotspeich, F.B. and Everest, F.H. (1981) A new method for reporting and interpreting textural composition of spawning gravel. USDA Forest Service Research Note PNW-369, 1-11. MacPherson, H.J. (1971) Downstream changes in a sediment character in a high energy mountain stream channel. Arctic and Alpine Research, 3(1), 65-79. McNeil, W.J. and Ahnell, W.H. (1964) Success of pink salmon spawning relative to size of spawning bed materials. U.S. Fish and Wildlife Service Spec. Sci. Rep., Fish. 469, 15p. Miller, J.P. (1958) High mountain streams: Effects of geology on channel characteristics and bed material. New Mexico State Bureau of Mines and Mineral Resources, Memoir, 4, 23-76. Moring, J.R. (1975) The Alsea watershed study: Effects of logging on the aquatic resources of three headwater streams of the Alsea River, Oregon. Part 2: Changes in environmental conditions. Fish. Res. Rep. 9. Oregon Department of Fish and Wildlife, Corvallis. 39 p. NCASI (National Council of the Paper Industry for Air and Stream Improvement), (1986) A comparison of 4 procedure for determining streambed substrate composition. NCASI Technical Bulletin, 481, 40p. Nordin, C.F., Meade, R.H., Curtis, W.F., Bosio, N.J., and Landim, P.M.B. (1980) Size distribution of Amazon River bed sediment. Nature, 286, 52-53. Nuttal, P.M. (1972) The effects of sand depotiion upon the macroinvertebrate fauna o fht eriver Camel, Cornwall. Freshwater Biol. 2, 181-186. Parker, G. and Klingeman, P.C. (1982) On why gravel bed streams are paved. Water Resources Research 18, 1395-1408. Pearse, P.H. (1982) Turning the tide: a new policy for Canada's Pacific fisheries. The Commmision on Pacific Fisheries policy. Final Report. Can. Dep. Fish, and Oceans, Vancouver, B.C. Philips, R.W. (1971) Effects of sediment on the gravel environment and fish production. In J.T. Krygier and J.D. Hall (eds.) Forest land uses and the stream enviornment. Oreg. State Univ., Corvallis, Or. USA, 64-74. Platts, W.S., and Megahan, W.F. (1975) Time trends in channel sediment size composition in salmon and steelhead spawning areas: South Fork Salmon River, Idaho. USDA For. Serv. Gen. Rep. Intermountain For. and Range Exp. Stn., Ogden Utah. 21 p. Platts, W.S., Shiraze, M.A., and Lewis, D.H. (1979) Sediment particle sizes used by salmon for spawning with methods for evaluation. U.S. Environmental Protection Agency - 6--13-79-043. 32 p. Corvallis Environmental Research Laboratory, Office of Research and Development, Corvallis, Oregon. Plumley, W.J. (1948) Black Hills terrace gravels: a study in sediment transport. J. Geology. 56, 526-577. Poulin, V.A. (1984) A research approach to solving fish/forestry interactions in relation to mass wasting on the Queen Charlotte Islands, British Columbia. B. C. Min. For., Land Manage. Rep. 27, 16p. Rana, S.A., Simons, D.B., and Mahmood, K. (1973) Analysis of sediment sorting in alluvial channels. American Society of Civil Engineers, Proceedings, 99, HY11, 1969-1980. Roberts, R.G. and Church, M. (1986) The sediment budget in severely disturbed watersheds, Queen Charlotte Ranges, British Columbia. Can. J. Forest Res., 16(5), 1092-1106. Roberts, R.G. (1987) Stream channel morphology: Major fluvial disturbances in logged watersheds on the Queen Charlotte Islands, British Columbia. B. C. Min. For., Land Manage. Rep. 48. Rood, K.M. (1984) An aerial photograph inventory of the frequency and yield of mass wasting on the Queen Charlotte Islands, British Columbia. B. C. Min. For., Land Manage. Rep._ 34. Schumm, S.A. and Stevens, M.A. (1973) Abrasion in place: a mechanism for rounding and size reduction of coarse sediments in rivers. Geology 1, 37-40. Schwab, J.W. (1983) Mass wasting: October-November 1978 storm, Rennell Sound, Queen Charlotte Islands, British Columbia. B.C. Min. For., Res. Note 91, 23 p. Schwab, J. pers. comm. via Hogan, D. (1990) Scrivener, J.C. and Brownlee, M.J. (1989) An analysis of Carnation Creek gravel-quality data 1973 to 1981. In. G.F. Hartman (ed.) Proceedings of the Carnation Creek Workshop: A ten-year review, 154-173. Pacific Biological Station, Nanaimo, B.C. Shaw, J. and Kellerhals, R. (1982) The composition of recent alluvial gravels in Alberta. Alberta Research Council Memoir, 43, 151 pp. Sheridan, W.L., Perensovish, M.P., Faris, T., and Koski, K. (1984) Sediment content of streambed gravels in some pink salmon spawning streams in Alaska. In W.R. Meehan, T.R. Merrell, Jr., and T.A. Hanley (eds.) Fish and wildlife relationships in old-growth forests: Proceedings of a symposium, 153-165. American Institute of Fishery Research Biologists. Shields, A. (1936) Application of similarity principles and turbulence research to bedload movement. Translation by Ott, -W.P. and Uchelen, J.C. California Institute of Technology, Pasadena. Sternberg, H. (1875) Untersuchungen uber Langen - und Querprofil geschiebe-fuhrender Flusse. Zeit. Bauwesen. 25, 483-506. Sutherland Brown, A. (1968) Geology of the Queen Charlotte Islands, British Columbia. B.C. Dep. Mines Pet. Resources Bull. 54, 226 p. Victoria, B.C. Tripp, D.B. and Poulin, V.A. (1986) The effects of logging and mass wasting on Salmonid spawning habitat in streams on the Queen Charlotte Islands, British Columbia. B. C. Min. For., Land Manage. Rep. 50, 29p. Troutman, B.M. (1980) A stochastic model for particle sorting and related phenomena. Water Resources Research, 16(1), 65-76. Wolcott, J.F and Church, M. (in press) Stategies for sampling river gravels. J. Sedimentary Petrology. Wolcott, J.F. (1990) Flume studies of gravel bed surface response to flowing water. Unpublished Ph.D. Thesis, University of British Columbia. Wolman, M.G. (1954) A method of sampling coarse river bed material. Amer. Geophysical Union Transactions, 35, 951-956 Yatsu, E. (1954) On the formation of slope discontinuity at fan margins. Institute of Natural Resources (Japan) Miscellaneous Report 36, 57-64. Yatsu, E. (1955) On the longitudinal profile of the graded River. Trans. AGU., 36(4), 655-663. Appendix 1. Hogan's Log Jam Classification Scheme This appendix is an abridged version of the classification scheme developed by Dan Hogan during his work on the Queen Charlotte Islands (Hogan, 1989, 1989a). A logjam is defined as a major accumulation of debris that alters, or has altered recently, sediment transport downstream and channel stability. Several log jam features which relate to the geomorphological functioning of log jams are included in the classification. In this study log jams have been described in terms of five of the eight parameters identified in Hogan's (1989) classification. These are described below. For each parameter a ranking system from one to five is used to classify the jam in question. In general, lower rankings indicate impermeability and strength, while higher numbers are associated with the deterioration of the jam and increasing permeability. Jam Feature Rank Characteristics Jam Age 1 Very young; new trees (bark, branches), no nursed trees, apparently formed during last major storm event (less than 2 years old). 2 Recent; LOD has some bark and branches, nurse trees (usually red alder) are less than 5 m high, less than 10 years old 3 Moderate; nursed trees between 10 and 20 years old (aged by increment bores) 4 Moderate to old; nurse trees between 20 and 30 years old, LOD has few branches and no bark 5 Old ; debris is smooth except for moss, nursed trees are 30 to 50 years old 5 + Very old; nursed trees are over 50 years Jam Integrity 1 Very solid; large, well anchored LOD pieces, no rot, compact structure 2 Solid; smaller pieces, anchored, fairly compact 3 Moderate; less compact, open spaces between LOD, some rot 4 Weak; poorly structured individual pieces, poor anchoring, rot 5 Very weak; small pieces, no anchoring Jam Span (lateral extent) 1 Complete; jam completely crosses the channel and forms a dam over which water flows. 2 Incomplete ; 3/4 < span < 1 bankfull width 3 1/2 < span < 3/4 bankfull width 4 1/4 < span < 1/2 bankfull width 5 span < 1/4 bankfull width Sediment Storage (upstream) 1 Full; The channel zone is completely full of sediment (i.e. sediment is level with the top of the jam, fills the bankfull width, and extends upstream as a function of channel gradient) 2 Not full; < 1/4 of sediment evacuated (cf 1) 3 1/4 to 1/2 removed 4 1/2 to 3/4 removed 5 Empty; all sediment removed Number of Channels 1 Single channel 2 Two channels 3 Three channels 4 Four channels 5 Five or more channels 


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



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


Related Items