Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Patterns of channel change on Chilliwack River, British Columbia Ham, Darren Gary 1996

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

Item Metadata


831-ubc_1996-0471.pdf [ 12.28MB ]
JSON: 831-1.0087677.json
JSON-LD: 831-1.0087677-ld.json
RDF/XML (Pretty): 831-1.0087677-rdf.xml
RDF/JSON: 831-1.0087677-rdf.json
Turtle: 831-1.0087677-turtle.txt
N-Triples: 831-1.0087677-rdf-ntriples.txt
Original Record: 831-1.0087677-source.json
Full Text

Full Text

PATTERNS OF C H A N N E L C H A N G E O N CHILLIWACK RIVER, BRITISH COLUMBIA by DARREN GARY H A M B.Sc. (Hon.), University of Victoria, 1990  A THESIS SUBMITTED IN PARTIAL F U L F I L L M E N T OF T H E REQUIREMENTS FOR T H E D E G R E E OF M A S T E R OF SCIENCE in T H E F A C U L T Y OF G R A D U A T E STUDIES (Department of Geography)  We accept this thesis as conforming to the required standard  T H E UNIVERSITY OF BRITISH COLUMBIA August 1996 © Darren Gary Ham, 1996  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 The University of British Columbia Vancouver, Canada  DE-6 (2/88)  ABSTRACT This study investigates changes to channel morphology along Chilliwack River in southwest British Columbia, and relates these changes to the transport of bed material. The channel was mapped using a stereoplotter from a sequence of historical aerial photography between 1952 and 1991. Maps for selected years were overlaid, then analyzed with a GIS. Erosion and deposition volumes of bed material were determined by multiplying measured plariimetric changes by the varying depth of bed material along the river. A sediment budget framework was used to estimate bed material transport rates from these net changes in stored sediment volumes. The transport rate of bed material for Chilliwack River is estimated as 55,000±10,000 irrVyr for the period 1983 to 1991. Instability along lower channel reaches has increased over the 40 year period of study due to an increase in both the magnitude andfrequencyof large floods. As a consequence, channel width, bank erosion rates and the volume of sediment transported past Vedder Crossing became increasingly large. Between 1952 and 1975, Chilliwack River was in a transient state of equilibrium as the channel continued to recoverfroma sequence of large floods in previous decades. Large floods in 1975 and 1980 caused significant bank erosion along lower reaches, which increased the amount of sediment available for transport. Extreme floods in 1989 and 1990 caused further erosion, and in fact, were large enough to alter the pre-flood channel regime. However, in the absence of further large floods, the channel should recover from these events in 10 to 20 years. Sigriificant morphologic change on Chilliwack River occurs roughly once every 5 years, when peak flows exceed 500 mVs. These floods are sufficient to erode channel banks, where the dominant supply of mobile bed material is stored. Aggradation occurs in the short term as more sediment is introduced to the active channel zone than can be removed by subsequent smaller flows. However, as flows as small as 250 m /s (which occur several times per year, on average) are capable of mobilizing bed material, there is net degradation over the longer term. As the length between survey dates used in this study averaged 10 years, some information on bed material transport between dates is lost. Nonetheless, this study has demonstrated that considerable information on historic channel change can be obtained form aerial photographs. 3  (  T A B L E OF CONTENTS  Abstract  ii  Table of Contents  iii  List of Figures  vi  List of Tables  viii  Acknowledgements  ix  Chapter 1: Study Objectives and Background 1.1 Introduction  1  1.2 Development of the 'morphologic approach'  5  1.3 Sediment Transfer Rates  12  1.3.1 Sediment budget approach  12  1.3.2 Depth of mobile sediment  16  Chapter 2: Chilliwack River Site Characteristics 2.1 Setting  20  2.2 Recent history of Chilliwack-Vedder River  22  2.3 Need for further study  25  2.3.1 General watershed resources  25  2.3.2 Timber harvesting and fisheries impacts  25  2.3.3 Fisheries resources  29  2.4 Hydrologic regime  30  2.4.1 Strearnflow measurements  30  2.4.2 Climatic influences  31  2.4.3 Flood frequency analysis  33  2.4.4 Frequency analysis of flows capable of transporting bed material  36  2.4.5 Reach strearnflow estimates  41  Chapter 3: Channel Change Data 3.1 Aerial photographs and maps  46 iii  Chapter 3: Channel Change Data 3.1 Aerial photographs and maps 3.2 Photo record selection 3.3 Methods of data collection and analysis 3.3.1 Photo overlays 3.3.2 Analytic stereoplotter 3.3.3 Geographic Information Systems 3.4 Creating and analysing Cliilliwack River spatial data  Chapter 4: Historic Channel Changes 4.1 Morphology of study reaches 4.2 Patterns of channel stability 4.3 Lateral stability 4.3.1 Water surface width 4.3.2 Bankfull width 4.3.3 Bank erosion 4.4 Prediction of bank erosion and channel migration  Chapter 5: Sediment Transfer and Yield 5.1 Channel change assessment 5.1.1 Estimation of bed material depth 5.1.2 Areal exposure corrections 5.1.3 Channel change analysis 5.2 Bed material sediment budget 5.2.1 Net storage changes 5.2.2 Sediment contribution from tributaries 5.3 Estimating the sediment transport rate in Chilliwack River 5.4 Discussion of transport estimates 5.4.1 Short-term (flood length) changes 5.4.2 Results from other studies 5.4.3 Error analysis  References  154  Appendices Appendix A : Summary of storage change error terms  161  Appendix B : Sediment budgets for Criilliwack River  163  v  LIST O F F I G U R E S  Figure 1: Definition diagram for bed material transfer  6  Figure 2: Location of study area  21  Figure 3a: Mean annual and maximum annual daily discharge, Vedder Crossing Figure 3b: Monthly flow regime for Chilliwack River at Vedder Crossing  32 32  Figure 4: Flood frequency analysis based on annual maximum series  35  Figure 5a: Bedload grain size on Vedder River Figure 5b: Relation between discharge and bedload transport on Vedder River  37 37  Figure 6a: Frequency count of mean daily discharge > 250 m /s Figure 6b: Flow-frequency analysis based on partial duration series  39 39  Figure 7a: Reach strearnflow estimates Figure 7b: Predicted vs. gauged strearnflow  45 45  Figure 8: Significant floods in relation to photo dates in study  50  Figure 9: Mean daily flood flow sequences  51  Figure 10: Channel maps for reach 1, 1966 and 1973  63  Figure 11: Overlaid channel maps for reach 1  64  Figure 12: Photo of boulders along channel banks, reach 15  68  Figure 13: Photo of mid-channel island in reach 8  69  Figure 14: Photo of erosion along glaciolacustrine deposits in reach 4  71  Figure 15: Photo of lag boulder deposits near Tamihi Creek  72  Figure 16: Photo of gravel berm along reach 2  74  Figure 17a: Variations in water surface width Figure 17b: Variations in bankfull width  77 77  Figure 18: Comparison of bankfull width changes  81  Figure 19a: Measured and predicted bankfull width 1952-1973 Figure 19b: Measured and predicted bankfull width, 1973-1991  83 83  3  vi  Figure 20: Bank erosion from Vedder Crossing to Liumchen Creek  88  Figure 21: Bank erosion from Liumchen Creek to Ryder Creek  89  Figure 22: Bank erosion from Tamihi Creek to Slesse Park  90  Figure 23: Bank erosion below Slesse Creek  91  Figure 24a: Reach averaged bank recession rates Figure 24b: Reach maximum bank erosion rates  95 95  Figure 25: Variability map of historic channel occupation  100  Figure 26a: Downstream variations in gravel bar depth Figure 26b: Reach averaged gravel bar and bank height estimates  104 104  Figure 27: Exposure of gravel bars to variations in discharge  110  Figure 28: Streamflow vs. water surface width along lower reaches  111  Figure 29: Net storage change of bed material (bulk volumes)  120  Figure 30: Sediment transport rate on Slesse Creek  129  Figure 31: Bed material transport rates for Chilliwack River  134  Figure 32a: Rate of vegetation growth on gravel bars Figure 32b: Rate of floodplain reconstruction  137 137  Figure 33: Bed material transport rates with estimated error ranges  148  vii  LIST O F T A B L E S  Table 1: Water survey of Canada gauging stations, Chilliwack River watershed  31  Table 2: Streamflows for different dates of photographic records  42  Table 3: Available stereo photographic coverage for Chilliwack River  47  Table 4: Percentage of total channel occupied area  99  Table 5: Exposure correction coefficients for reaches 1-6  112  Table 6: Summary of channel changes along reach 1, 1966-1973  116  Table 7: Annual volumetric erosion and deposition volumes  123  Table 8: Regression of mean monthly discharge for Slesse Creek and Vedder Crossing  125  Table 9: Bed material transport rates for Slesse Creek  130  Table 10: Morphologic changes and bed material transfer rate on reach 1  139  viii  ACKNOWLEDGEMENTS  My primary thanks go to Dr. Mike Church for all of his careful guidance and patience during the completion of this project. You inspire your students to do their best, for which we are all grateful. I also wish to thank the other members of my committee, Dr. Hans Schrier and Dr. Brian Klinkenberg, for providing assistance when it was needed.  Initial funding for this project was provided by the Ministry of Forests Research Branch. This project could never have been completed without that generous support. In particular, I wish to express my gratitude to Mr. Dan Hogan for finding the resources, and for his willingness to talk whenever I stopped by his office.  Several individuals assisted with the field component of this study, including Bertrand Groulx, Lars Uunila and especially Craig Nistor (who still owes me 7 days work and a hot meal).  Finally, I wish to thank my parents for supporting my decision to do this project, even when I wasn't certain it was correct.  ix  Chapter 1: Study Objectives and Background  1.1 Introduction In small watersheds, hillslopes are directly coupled to the stream channels and the transfer of sediment through the channel system occurs shortly after the landscape is disturbed. Many studies in the Pacific Northwest have examined the effects of different land-use practices (mainly timber harvesting) on sediment yields and impacts in these systems (see Roberts & Church, 1986; Caine & Swanson, 1989; Tripp & Poulin, 1992). Recently, more attention has been focussed on the sediment cascade in larger systems (see Jordan & Slaymaker, 1990; Benda, 1990) where land-use impacts over time and space occur in more complex ways. Larger watersheds are generally characterised by a well 1  developed floodplain in their lower reaches which isolates the channel from adjacent hillslopes and stores a considerable amount of alluvial material for comparatively long periods of time. As a result, the physical and ecologic effects of different land-use practices on slopes may not be introduced to the river for decades or even longer because the floodplain acts as a buffer. There is a long transit time for coarse sediments to travel from headward source areas to lower reaches, delaying channel response to these inputs. Consequently, these effects may also persist for similarly long time periods as sediment is slowly exchanged through the system.  Larger watersheds arefrequentlymultiple-use systems where the provision of sustainable aquatic habitats and recreational opportunity competes with timber harvesting, urban and industrial  1. As used here, large watersheds are characterized by channels with alluvial floodplains along lower channel reaches. In coastal regions, basins as small as 10 km may have a floodplain (e.g. Carnation Creek) but the minimum size is more commonly in the 50 - 100 km range. For interior regions of British Columbia, channel floodplains are seldom found within basins smaller than 100 - 200 km . Larger watersheds arefrequentlyassociated with channels haying a minimum bankfull width of20 - 30 m (Channel Assessment Procedure Guidelines, 1996). Church (1992) states that the morphology of large (20 - 30 m) channels is determined by fluvial and geologic constraints, rather than riparian effects. 2  2  2  -1-  development and agriculture. An understanding of the response of channels to changes in sediment input, and the processes responsible for supplying this sediment, is necessary to determine the effects of various land-use practices and to provide effective management strategies.  The staging of stored sediment along a river causes changes in channel morphology. Most natural alluvial channels are subject to regular shifts in position causing erosion and deposition of the riverbank (floodplain), channel bars and islands. This process leads to the transfer and distribution of eroded materials downstream, where they are redeposited. Therefore, observed changes in channel morphology can be used to estimate sediment transfer. This "morphologic approach" was first introduced by Popov (1962) and expanded by Neill (1971, 1987), who demonstrated that, by measuring erosion volumes over some time period(s), estimates of sediment transport rates could be made. Refinements of these techniques are presented in McLean (1990) and Martin (1992).  The morphologic approach has received little attention even though there are many problems associated with conventional approaches for measuring transport rates and volumes. Transport formulae which relate sediment transport to strearnflow hydraulics and sediment characteristics consistently fail to predict sediment transport accurately (see Gomez and Church, 1989), especially on channels for which they were not developed (Parker et al., 1982). In fact, a successful solution for predicting bedload transport has not been found despite more than a century of work on the problem (see discussion in Church, 1985; Carson & Griffiths, 1987; Reid and Frostick, 1994). Rivers have a complex, three-dimensional structure that is subject to temporal changes (eg. timing and magnitude of floods, sediment supply, geomorphologic history) and spatial changes (eg. variations in grain size, bedforms, geologic setting). This makes it exceedingly difficult to adequately  characterize channel boundary conditions, and ultimately results in the failure of hydraulic equations (Carling, 1992). Given these difficulties, and the time, cost and effort required to take direct bedload measurements, alternative methods of estimating sediment transport should be evaluated.  The morphologic approach does not require any description of channel mechanics or assumption of boundary conditions to estimate sediment transport. When applied over geomorphically relevant time periods, the method can account for variations in transport rates that may not be observed over shorter sampling periods. Church et al. (1987) note that the technique can be applied to increase our understanding of the long-term sediment cascade in alluvial river channels. This approach can incorporate a longer data record than studies in which direct field measurements of sediment flux are made while increasing the speed of data collection and analysis. This is because the technique takes advantage of existing maps, airphoto s or other sources of historical documentation for channel morphology. Finally, the morphologic approach provides additional data about channel changes, such as bank erosion rates, that other methods do not. All of these circumstances appear to make it a preferred technique for the study of alluvial rivers.  In this study, the bed material sediment budget will be estimated for Chilliwack River, a gravelbedded channel which drains a 1230 km basin in southwest British Columbia. The river is an ideal 2  site to further test the morphologic approach because it has extensive hydrologic and photo data records. As well, the basin is a multiple-use watershed with a well developed floodplain along the lower reaches of the channel. Patterns of erosion and deposition along the mainstem channel can be digitizedfromhistorical aerial photography over time scales from one year to several decades. This allows an examination of short-term process-related changes as a result of individual floods, and of  the resulting longer-term process-related trends which ultimately influence channel development. The net changes are calculated and analysed on a reach-by-reach basis using a Geographic Information System (GIS). These changes can be used in a sediment budget framework to provide lower bound estimates of sediment transport rates and to identify major sediment sources, storage zones and principal transfer pathways along the channel.  The use of an analytic stereoplotter and GIS for the morphology-based sediment budget is fairly new. McLean (1990) developed a digital terrain model (DTM) to relate changes in stored sediment volumes to transport rates, effectively building a GIS. Lane et al. (1994) discuss the application of analytic photogrammetry and DTM's for a study of changing bed topography for a short (50 m) section of a small proglacial stream. They suggest that the superimposition of the D T M with additional spatial data could be used to develop a GIS. The GIS-based analysis presented in this thesis is based upon spatial overlay analysis of mapped features, rather than of DTM's, so is not directly comparable to the methods used in these prior studies. This overlay approach offers a more detailed analysis of changing channel morphology characteristics than could otherwise be made. Some examples of these unique characteristics and types of analysis will be presented in following chapters to illustrate the value of GIS as a research tool in studies of fluvial geomorphology.  The main objectives of this study are: •  To map the patterns of channel changes along the river and determine the processes governing these changes. The equilibrium channel regime will be investigated;  •  to determine the role of frequent low-magnitude, and infrequent high-magnitude, flows as agents of morphologic change and to examine the nature of channel response in each case;  •  consequently, to estimate the rates of sediment transfer and yield along the channel and to examine how these rates may have changed over time; and  •  to demonstrate the capabilities and merits of analytical photogrammetry and GIS as research tools for quantitative studies of river channel change.  1.2 Development of the 'Morphologic Approach* There is practical significance in estimating bed material transport rates for river management purposes. This has led to the proliferation of bedload transport functions, which assume that a relation exists between hydraulic variables and the transport of bedload. By comparison, development of the alternative [morphologic] approach is based on the observation that morphologic changes along a river necessarily indicate that sediment has been transported. The earliest discussion relating morphologic changes along a river to sediment transport rates was given by Popov (1962) for the River Ob', Russia. Popov used historic maps and airphotos to examine erosion of floodplain and channel bed sediment along a 450 km section of channel within a sediment balance (budget) framework. Neill (1971) noted that Popov failed to demonstrate clearly how erosion volumes were related to bed material transport and attempted to quantify this relation for rivers exhibiting the systematic down valley migration of meander bends. Neill stated that material eroded along an outer concave bank is deposited downstream on an accreting point bar. In assuming that the average downstream travel distance of sediment was equal to one-half the meander wavelength (i.e. from cutbank to successive point bar; refer to Figure la), Neill estimated the volumetric transport rate as:  Q  = -fix 2  Ax— dt  [2.1]  where !„, is the meander arc length, h is the bank height and de/dt is the annual bank recession rate.  Meander wavelength (!„  Plan of Meander Loop Migrating Down-Valley HWM deposition  h erosion  Cross - Section Y - Y  Figure 1: Definition diagram for bed material transfer. Diagram A (top) shows a simplified meander sweep process in a natural river (from Neill, 1987). Diagram B (bottom) depicts erosional and depositional areas as measured from successive airphotos and calculated in the GIS (adapted from Martin and Church, 1995).  Neill's approach can be used to estimate transport rates at a number of sites along a river if estimates of the variables can be made.  Neill (1987) noted that the assumption of travel distance, /, was weak when a river had poorly developed meander bends or a substantial portion of the eroded volume moved as wash material, such that / was very long. In the former case, the actual transfer length may be longer than one-half meander, so the value of Q is underestimated. In the latter case, Q is overestimated as only a fraction s  s  of the eroded material actually constitutes bed material. Neill was able to account for this problem through stratigraphic assessment of the channel banks; eroded volumes were determined as eroded areas multiplied by the height of the bed material deposits rather than the total bank height. A similar approach is used in this thesis, though Neill's adoption of the meander wavelength as an appropriate tranfer length is avoided. Neill applied the method to the Tanana River, Alaska and found that the estimated average annual transport rates were close to estimates provided from a direct bedload sampling program. This suggested that the assumption of complete bed material exchange between eroded banks and downstream point bars was valid for the Tanana River.  Church et al. (1987) discussed the feasibility of using historical seqences of aerial photography to map channel changes and infer sediment transport rates along Mackenzie River. This paper presented the first generalized outline of the morphology-based method and re-introduced the idea of the sediment budget approach first discussed by Popov. The authors discussed how an estimate of sediment transport at a given section could be used in conjunction with changes in sediment storage volumes to extend transport calculations along adjacent reaches, although the study was along a single test reach. The bed material transport rate was estimated to comprise between 1.5% and 3%  of the total sediment load, which they considered reasonable. Although there are key differences in project scope, methods and analysis, several of the conventions presented by Church et al. (1987) have been adopted in this thesis.  Collins (1990) discussed the validity of using planimetric data from historic aerial photographs to study channel changes over a 44 year period along the Tanana River, Alaska. Areas of erosion and deposition for islands and floodplain sediments were calculated from overlay maps, but no estimate of bed material transport was made. The study was not able to show changes in bar erosion / deposition or areas of bed aggradation / degradation. Collins used channel overlay maps to measure total bank erosion rates and quantitatively demonstrated the relation between anthropogenic influences (dam construction in this study) and changes to channel morphology. Several of these ideas are further expanded in this thesis.  The most comprehensive study to date relating channel changes to sediment transport is presented by McLean (1990) for the Fraser River. McLean attempted to demonstrate the feasibility of relating patterns of channel instability to sediment transport rates along a wandering gravel bed reach. Previous studies had constrained application of the procedure to regularly meandering channels for which the transfer distance of bed material is more easily determined (i.e. Tanana River). A digital terrain model (DTM) was constructed to calculate the difference in stored bed material between bathymetric surveys completed in 1952 and 1984. A separate term in the sediment budget was determined for bank erosion and reconstruction. These data were measured from planimetric surveys of bankline position and multiplied by estimated or measured bank heights; a similar procedure is used in this thesis. A stratigraphic assessment of the channel banks and eroding islands was used to  distinguish gravelsfromoverbankfines(sand) in the sediment budget and a masking routine was used to limit calculations to a common active channel zone. By assuming that gravel transport past a downstream sand-bed section of channel was negligable, sediment budget calculations were extended upstream. Estimated gravel transport based on the sediment budget was 120,000 mVyr, which compares well to an estimate of 100,000 m /yr based on direct bedload samples taken at the upstream 3  end of the 45 km test reach.  McLean (1990) also estimated transport rates based on planimetric changes from aerial photographs and maps from 1890 to 1971 and compared these to measurements based on the D T M derived sediment budget. Erosional volumes from islands and floodplain deposits were estimated as eroded areas multiplied by bank heights; depositional volumes were not calculated as a representative deposit thickness could not be estimated from the maps. The transfer of material from gravel bars was ignored as these deposits are primarily derived from bank erosion (they were assumed to be constructional features that are eroded on sharply defined [bank] edges). The transport rate was estimated by applying a representative step length for sediment transfer downstream, taken as the average spacing between major deposition zones. The bed material transport rate based on planimetric changes is given as 80,000 m /yr, which is lower than the estimate based on the sediment 3  budget. Results based on the sediment budget and morphology changes were the most reliable and more generally applicable when compared to other techniques because they examined changes over a time scale of years to decades, similar to the time scale for processes governing transport on the Fraser River.  Martin (1992) and Martin and Church (1995) estimated sediment transport along Vedder River by  analysing changes in cross section area between years. Martin's approach is conceptually similar to McLean's survey technique, but did not require the construction of a full digital model. Martin estimated the net of scour and fill for cross sections and multiplied these by the distance between surveys to estimate volumetric changes. This procedure makes the assumptions that there is no compensating scour and fill at survey locations and that measured changes at a cross section are representative of the channel to one-half the distance between adjacent surveys. The main Umitation of this approach is that it is constrained by the availability of repeated cross sections and can not provide additional data on morphologic change (eg. bank erosion rates) that the photography based methods can give (unless the cross-sections are known to be representative of bankline changes). As Martin's study is located in a downstream (renamed) reach of Chilliwack River, the results can be compared to estimates made in this thesis. Further, Martin's data can be used as a useful reference with which to check the magnitude of estimates of bed material transport along the Chilliwack River.  Two recent studies (Goff and Ashmore, 1994; Lane et al., 1994) measured short-term bed material transport rates by comparing rapidly repeated topographic surveys of the bed. Goff and Ashmore used closely spaced transects (cf. Martin, 1992) to compute mean bedload transport rates using a step-length approach (after Neill, 1971, 1987) and estimated daily transport rates within a sediment budget framework for a 60 metre reach of the proglacial Sunwapta River. Although there was no independent check to compare with these results, Goff and Ashmore claimed that the highly variable nature of bedload transport is related to the morphologic response of the channel. Lane et al. modeled river bed topography over a 50 metre length of a small actively braiding proglacial stream in Switzerland using analytic photogrammetry and ground surveys to construct DTM's. The net volume of scour and fill was determined as the difference between two bed surfaces. The authors  -10-  recognized the need to survey at an appropriate temporal scale (dependent on channel size and type) to ensure no between survey transfer of material if the technique is used to determine transport rates, but no estimate of this time scale was provided. Although the techniques described in both papers demonstrate the feasibility of relating transport rates to morphologic change, they are limited by intensive data requirements to studies of small spatial and temporal scales for gravel-bed proglacial streams. By comparison, McLean (1990) was able to study changes in the Fraser River using a temporal scale of years to decades, which is more appropriate for investigating channel stability and sedimentation on large rivers.  The approach used to estimate the bed material transport rate in this thesis is a hybrid of several techniques. In principle, the study design is most similar to Neill's (1971) regular meanders and McLean's (1990) planimetric airphoto/map study but offers the convenience of computer based data analysis (cf. McLean's D T M , 1990; Lane et al., 1994). This study differs from previous works, however, in the combination of study length (40 years), study scale (extends over 15 study reaches, rather than only 1 reach) and method of data analysis (overlay modeling in a GIS). Ideally, an approach based on bathymetric surveys makes the fewest assumptions about bed topographic changes, but data collection is costly, impractical on rivers which can neither be conveniently navigated by boat nor waded through, and comparable historic data may rarely be available. The morphology-based sediment budget is limited by assumptions that a representative bed material depth and an appropriate sediment transfer distance can be found. Further, the sediment budget allows errors to be propagated through the reaches and could significantly bias the results. However, even if reliable estimates of sediment transport rates can not be made, the technique provides useful information on bank erosion, channel width and changes in morphology that other techniques do not.  1.3 Sediment Transfer Rates 1.3.1 Sediment Budget Approach Sediment budgets have been used to quantify the sediment cascade in smaller coastal watersheds (see Dietrich and Dunne, 1978; Dietrich et al, 1982) but they may also be applied to larger systems (eg. Kelsey et al, 1981; Jordan and Slaymaker, 1991). In constructing a sediment budget for a particular drainage basin, it is important to identify and quantify the main processes responsible for the production and transport of sediment (Dietrich and Dunne, 1978). For alluvial channels, sediment inputs, outputs and storage changes in the fioodplain and active channel zone must be considered. The primary sediment transfers include erosion of island and fioodplain deposits into the active channel reservoir, reconstruction of island and fioodplain deposits by sediments from the active channel and net changes in the active channel reservoir (scour and fill of gravel).  In constructing a morphology based sediment budget, it is important to distinguish between the types of material that are present in the channel, and those that are actually important in terms of channel formation. Fluvial sediments in motion can be classified as bed material load or as wash material load. Wash material can be derived from catchment slopes, upper channel banks and overbank deposits on islands and floodplains. Once entrained, wash material travels through the channel in suspension without being deposited. Wash material has little impact on channel morphology except as a depositional feature on upper channel banks, although it does influence vegetation growth, hence bank erosional strength. The suspended load (which comprises all of the wash load, plus some fraction of the bed material load) for Chilliwack River averaged 132,200 tonnes/year between 1966 and 1975 (from Water Survey Canada records). Bed material is found on the bed and lower channel banks of a river. Particles which are entrained may move in suspension if they are small enough.  -12-  Larger, heavier particles will roll or slide along the bed by traction or hop along the bed (saltation) as they impact other grains (Carling, 1992). These particles are usually referred to as bedload sediments and travel only relatively short distances during entrainment. Although bed material may comprise only afractionof the total sediment load of many rivers, it is of practical significance as it governs the morphology and stability of alluvial channels. Changes in channel morphology therefore reflect the transport of bed material in gravel-bedded channels. No estimate of bedload or total sediment yield is available for Chilliwack River.  For a defined channel reach over some arbitrary timeframe,a sediment budget quantifies the primary transfer of sediments. Reaches are segments of a river having similar morphologic characteristics and are used to better understand the character of erosion and deposition (Church et al., 1987). The sediment budget can be expressed by the following equation:  V =V AV 0  [1.2]  r  Where V is sediment output volume from a reach, V is sediment input and A V represents change 0  ;  in storage volume. The sediment output from a given reach is equal to the sediment input for the reach immediately downstream. By this approach, only two terms must be known as the third can be derived as the difference. Calculations can be extended downstream (or upstream) on a per reach basis if the sediment transport rate is known at at least one location along the channel (such as a reach boundary). However, any errors in measurement will be similarly propagated and cumulate along the channel. The sediment budget must also account for tributaries which contribute significant volumes of sediment, or operations which remove sediment such as gravel dredging, or the terms will not balance. Transport rates can be determined from direct measurement (ie. using bedload samplers) or  -13-  estimated from certain boundary conditions. For example, the Chilliwack River drains a large lake, so sediment output from the lake is assumed to be zero because the lake acts as a storage sink for bed material sediment.  The storage term, AV, can be obtained as the difference between erosion and deposition of bar material and islands in the active channel zone and fioodplain sediments (see Figure lb). This is summarily expressed as: av=v -v 4  [1.3]  t  Planimetric changes as measured directlyfromsequential aerial photography can be used to establish locations of erosion and deposition along channel reaches. These areas can be converted to estimated erosion and deposition rates if an assumption is made about the depth of mobile bed material. These procedures are discussed in greater detail in Section 4.3. These results can be reduced to a mean flux (transport) rate for an arbitrary period, At, which is expressed as V = V - A V I At. o  t  The transport rate can alternatively be estimated for different sections of channel if the average travel distance of the bed material between survey times is known. This is useful for studies of bed material transfer along individual sections of channel or where the transport rate at a reference section is not known. The transport rate (after Neill, 1987) is given as:  [1.4]  If the length of sediment transfer between sites of erosion and deposition (L,) is equal to the reach length (L ) then the equation can be simplified as the volume eroded per unit time. Neill noted that r  -14-  if meanders are weakly developed or part of the eroded material passes into suspension, the transfer distance may be longer than the reach length (or bed material may simply bypass the first deownstream deposition zone) so only lower bound estimates may be made. McLean (1990) added that if morphologic methods are to be used for predicting the bedload transfer rate, the volume of sediment behaving as throughput will have to be small relative to the total sediment volume. Church et al. (1987) used the following expression to estimate throughput for each reach:  The sediment throughput is analogous to a measure of wash material load. As Martin (1992) noted, reach lengths must be short enough to ensure no within-reach transfer of bed material. Material that is redeposited within the same reach from which it was eroded will deflate values of V or V , e  d  resulting in negative values for V . Conversely, reach lengths that are too short will show inflated w  values of V as some bed material is mistakenly counted as washload. The approach based on Neill w  (Eq. 1.4) avoids this constraint when applied to estimates of bed material transfer.  It is important to note some additional limitations of the reach-based sediment budget approach for estimating bed material transport. For channels which are in equilibrium, A V may be very small between survey dates as sediment input and output for a reach are roughly equal. Church et al. (1987) noted that the usefulness of the approach to detect aggradation or degradation is based on the notion that the time scale for changes in storage may be long in some rivers. However, McLean (1990) added that if the time scale for channel adjustments is very long, these may be difficult to detect in short-term studies. A channel will also appear to be in equilibrium if no flow event capable of modifying channel form has occurred between surveys. Conversely, if the temporal resolution of  -15-  the study is too coarse, material may be stored and eroded within the time frame of the study and only a lower bound estimate of actual sediment transported may be made. It is also important that sites of erosion and deposition within a reach be unique. Martin (1992) notes that this limits application of the approach in sand bed rivers which experience compensating scour and fill within individual floods (or survey intervals).  1.3.2 Depth of Mobile Sediment Several studies have examined planform changes using historical aerial photography to qualitatively assess natural and land-use impacts on morphology (see Beschta, 1984; Collins, 1990). This type of analysis is affected by variations of discharge on different photo dates and is best suited for studies of channel shifting, channel widening and bank erosion rates. Quantification of the sediment budget and bed material transport rates requires knowlege of the depth of bed material deposits throughout the study region. Planimetered areas of eroded and deposited sediments (from bars, islands and channel banks) are multiplied by the estimated depth of the mobile sediment layer in each study reach to determine volumetric channel changes. Net differences in stored sediment volumes between periods are used to estimate bed material tranport rates.  In small channels, the depth of scour of bed material determines the depth of the active sediment layer. Bed material is commonly found as an irregular surface fill (facies) above an inactive boundary (eg. bedrock) surface. Scour depth is estimated as the maximum vertical mobilization of the bed material as determined from scour chains or repeated cross section surveys (Madej, 1984). Below this depth, the bed is considered inactive. Scour and fill measurements can also be used along larger gravel bed channels. Martin (1992) used repeated cross sections along the dyked Vedder River. She  -16-  assumed that there was no compensating scour and fill at survey locations. The depth of mobile sediment may also be approximated as the height difference between the average bed surface (thalweg) and the top of the channel banks. Laronne and Duncan (1989) demonstrate this for the North Branch Ashburton River. They found that the depth of vertical bed activity is dependent on the depth of scour plus deposition and that this total depth increased with discharge. As this difference will be reduced within aggrading sections of channel, the most reliable estimates probably can be made following high magnitude flows.  Eschner et al. (1983) commented onfloodplainreconstruction along the Platte River. They described a process whereby bars become vegetated during low water periods, stimulating further vertical growth as these forms decrease flow velocity and trap fine sediments at higher discharge. Infilling between islands or, alternately, channel migration allow these islands to attach to the floodplain. Along actively migrating or wandering reaches, this 'lateral accretion' process is thought to be the dominant depositional process, while 'vertical accretion' tends to dominate along constricted reaches (Reid & Frostick, 1994). The height of the  gravel banks above the bed surface then, should provide  a reasonable estimate of the thickness of mobile  bed material sediment for island and floodplain  deposits. However, erosion (and deposition) of bar material may be more precisely estimated by separate measures.  Neill (1987) distinguished bed materialfromthe total sediment load simply by measuring depth from the channel bottom to the top of the bed material deposits. In effect, the bed material forms a basal gravel platform which is overlain by afinersandy matrix. As islands and floodplains are formed from similar depositional processes (vertical accretion offineson the basal gravel layer), they are assumed  -17-  .  to have equivalent depths. The depth of gravel bar deposits is lower overall, but is assumed to be similar to the depth of the island and fioodplain deposits excluding the overlying fine sediments. Therefore, the difference between the depth of bars and island/floodplain deposits is simply the depth of the overlying finer material. Based upon these stratigraphic differences, the sediment budget can be used to estimate the transfer of gravels separately from the fine sediments, which actually comprise part of the wash material load.  The depth assumption associated with the planimetric mapping technique may make it a less reliable estimate of bed material transport than other approaches. Estimates made using digital terrain models (see McLean, 1990; Lane et al., 1994) are the most reliable technique. This procedure gives the net change over some time period between two surface volumes, thereby eliminating the need to extrapolate depth over a distance (cross section technique) or estimate depth over an area (planimetric technique).  There are several available methods for deterrnining the value of bed material (or total) depth. The best method is directfieldsurvey as this makes the fewest assumptions and can be applied in the most relevent locations. These methods are described below: 1) Repeated cross-sectional surveys. Surveys of the Chilliwack River have been completed by the Water Management Branch, Ministry of Environment and are available for 1976, 1979, 1990 and 1991. In total there are 35 sites between Vedder Crossing and Slesse Creek where cross sections are measured, though a complete survey is available for 1991 only (for which cross-sections extend upstream to reach 4 only). Furthermore, cross section locations were changed after 1976 so these data can not be directly compared with surveys for other dates.  -18-  2) Field checks can be used to verify (1) above and (3) below, though preferably should be used instead where possible. This technique is more precise than (3) and can directly exclude measurement of the overbank fines. Given an absence of surveyed cross sectional data to estimate scour along upper reaches of the Chilliwack, spot field measurements of the channel bottom to bar top height were made. Along lower reaches, channel bottom to bank top measurements were taken. The thickness of the overbank fine deposits was measured concurrently 3) Airphoto estimates can be made of bar and bank heights above water level. A set of 1:5000 aerial photographs was taken after the November, 1989 flood between Vedder Crossing and Slesse Creek. Bank and bar heights are clearly visible along parts of the channel and can be distinguished from the water level where shadows are not present. The scale of photography must be large enough to discern topographic variations on gravel bars if this method is used. It can not be used reliably along reaches where the banks are low or not easily distinguished (eg. masked by vegetation or shadows). The procedure is given below: i) Measure height from water surface to bank top at many locations ii) Estimate reach-averaged water depth based on the Manning equation, provided that Q and S are known. Width can be measured directlyfromthe photos.  [1.6]  The bed material thickness can be estimated as the sum of water depth and reach-averaged bank to water surface height (cf. Laronne & Duncan, 1989). Both the field and cross-section techniques are used to estimate bed material thickness along Chilliwack River. For further discussion on the procedures followed and presentation of results, refer to Section 5.1.  -19-  Chapter 2: Chilliwack River Site Characteristics  2.1 Setting Chilliwack River watershed has a total catchment area of about 1230 km and straddles the 2  Canada/United States border in the lower Fraser Valley (see Figure 2). The upper reaches of the river extend along the Cascade Range into Washington State and drain into Chilliwack Lake on the Canadian side of the border; several major tributaries also have their headwaters in Washington. From the lake, the river extends 49 km westward to Vedder Crossing which is about 100 km east of Vancouver. Below the crossing, Chilliwack River continues as Vedder River for 8 km and an additional 6 km as Vedder Canal before joining Fraser River at an elevation of roughly 3 m asl.  Chilliwack River has a glaciated U-shaped valley with a fairly broad valley flat, rising from about 25 m above sea level at Vedder Crossing to 620 m at Chilliwack Lake; channel slopes range from 0.62% (0.36°) above Vedder Crossing to 3.1% (1.75°) below the lake. The surrounding topography is rugged and mountainous, with ridges ranging from 700 m to 2400 m near Chipmunk and Foley Creeks, where small glaciers are found. Roughly 50% of the entire basin lies at elevations greater than 1100 m (Martin, 1992).  Much of the valley has been shaped by Pleistocene glaciation and many current valley landforms and processes reflect the continuing influence of glacial sedimentation. At the peak of Fraser glaciation, glaciersfromthe Cordilleran ice sheet and the Cascade glaciers coalesced in this valley (Saunders et al, 1987). Following the earliest glacial advances, the valley was steepened and widened and proglacial outwash was deposited in upland regions. During the main Fraser advance, the valley was  -20-  Figure 2: Location of study area  -21-  dammed by ice near Vedder Crossing to form a large lake, allowing thick deposits of glaciolacustrine sediments to accumulate (Saunders et al., 1987). Subsequent advances of ice smeared a thin mantle of till on valley walls and a large outwash plain of glaciofluvial sand and gravels developed at the margin of the retreating glaciers. Eventually, an extensive sandur was formed between Cliilliwack Lake and Larsons Bench above the Slesse Creek confluence (Thompson & VanDine, 1995, unpub.).  During the subsequent Sumas stade, many of the previously deposited unconsolidated formations were removed by ice and meltwater, but large volumes of sediment were re-deposited throughout the valley following the final glacial retreat. During this period, a 100 metre thick body of sandy till and drift was deposited on Ryder uplands and overflow from an ice-dammed lake formed large terraces in the lower part of the valley. The present Chilliwack Lake may also have formed during this period as a result of a jokulhlaup deposit from Post Creek damming the valley (McLean, 1980; Thompson & VanDine, 1995, unpub.). The present river channel is found in mainly unconsolidated deposits of outwash and glaciolacustrine sediments. The fioodplain itself is underlain by mostly cobbly gravel (Thurber Consultants, 1988) but, locally, the channel has scoured to bedrock (Hay & Co., 1992). The combination of steep slopes and the depositional history of the surficial materials leads to a large number of potential geomorphic hazards in the region, including fioodplain erosion, landsliding, and debris flows. Many of these processes deliver sediment directly to the tributaries and main channel. Included in this activity are a number of clay and silt bank failures (slumps) in exposed glaciolacustrine sediments along the mainstem channel which are initiated by seepage erosion and the undercutting of channel banks.  2.2 Recent History of Chilliwack-Vedder River Chilliwack River passes through a bedrock confined gap at Vedder Crossing and has formed a large -22-  alluvial fan downstream. The river from this point is known as Vedder River. Several different channels have served to carry the flow of the CWlliwack north to the Fraser including Chilliwack Creek, Luckakuck Creek and Atchelitz Creek (Bowman, 1992). Prior to the arrival of European settlers, the Chilliwack sometimes drained west into Sumas Lake, forming a network of small channels in the boggy Sumas plain (Orchard, 1983). The Sto:lo Indian word for the river at this time was Tswelmuh, the term Tswel meaning to "go away" and was used to define "a river that changes its course" (Orchard, 1983). The habits of the shifting Chilliwack River are also remembered by some of the place names given in the region; Soowahlihl, a place along the lower Vedder fan was referred to as "a large stream that disappeared" and Atchelitz described "a place where two rivers meet" (Peters, 1978).  Contemporary written records do not exist prior to the arrival of European settlers in the 1850s. At this time, CWlliwack River drained into both Chilliwack and Luckakuck Creeks. Vedder River was a small, shallow creek at the time and did not carry any of the Chilliwack River's flow. However, this changed in 1873 when another channel shift occurred as a result of increased flow caused by the removal of a beaver dam (Bowman, 1992). The most significant known change occurred on November 22, 1875 as the result of a significant rain on snow generated flood. A large logjam diverted water down both Vedder Creek and Luckakuck Creek (the width of which expanded eight times). Flood control measures were initiated the following year in an attempt to mitigate flood damages. The matter was settled following the freshet of 1894 when a judicial decision asserted that Vedder should be the sole outlet of Chilliwack River and a rock-filled crib was built across the mouth of Luckakuck Creek (Bowman, 1992). However, floods continued to hamper residents of Sumas until 1919, with the acceptance of the Sinclair plan to channelize Vedder River from the Fraser River  -23-  up to Yarrow and drain Sumas Lake. The plan was completed by 1924 at a total cost of $3.4 million and reclaimed 13,350 ha of farmland (McLean, 1980). Vedder River is now known as Vedder Canal along the dyked section. Flood protection was expanded upstream during the 1960s and consists of bank protection, dykes and gravel removal (Martin, 1992).  The flooding problem has been largely resolved on Vedder River but continues to plague the lower reaches of Chilliwack River. McLean (1980) estimated that damaging floods would result when the mean daily flow exceeded 500 m /s; recent flows exceeding this threshold occurred in 1951, 1975, 3  1980,1984,1989 and 1990. Bowman (1992) notes that the floods of 1951, 1975, 1989 and 1990 all caused extensive property damage due to bank erosion. Private property losses from the 1989/90 floods were estimated at nearly $1.3 million (Chilliwack Valley Ratepayers, 1993,  unpub.) while  losses to governments and utilities between 1984 and 1990 were estimated at $4.9 million (Hay & Co., 1992). Large flow events are also sometimes responsible for initiating mass movement processes (landslides). The 1975 flood, for example, eroded glaciolacustrine deposits between Tamihi and Slesse Creeks, resulting in a suspended sediment yield higher than in any previous year for which measurements were made. These events are potentially hazardous to instream fish populations (see Section 2.4). Floods may also introduce substantial quantities of large woody debris into the channel which act to store sediment behind them or to divert flows, causing further bank erosion. In an attempt to reduce further property damages and protect the fisheries, remedial measures including bank protection and dyking have been completed in recent years. Gravel dredging may also be introduced along the lower Chilliwack River to reduce the risk of flooding and prevent further aggradation on the Vedder River fan (Vic Galay, 1995,  pers comm.).  -24-  2.3 Need for Further Study 2.3.1 General watershed resources Chilliwack River is one of the major tributaries of Lower Fraser River along its floodplain between Hope and Vancouver. Larger watersheds like the Chilliwack are becoming increasingly important in terms of land use management because of their multiple, and sometimes conflicting uses. Loss of property from erosion and damage to structures from flooding are major concerns to residents and landholders. Roughly 1300 residents (1991 est., Regional District Fraser-Cheam) occupy four small sites along the river including several small farms and a Native Reserve located just upstream of Vedder Crossing. Publically held lands are used for military training areas and three security institutions (prisons). There is also considerable recreational opportunity throughout the watershed, of which the most important is steelhead and salmon fishing. Further recreational opportunities include camping (2 Provincial Parks, 14 Forest Service campsites), Whitewater kayaking and commercial rafting, hiking along established trails and wudlife viewing. The need to protect properties from flood damage through mitigative works like berms, removal of large organic debris and gravel dredging is generally in conflict with the preservation of fish habitat and therefore a source of contention in this region. Numbers of users, types of activities and potential conflicts will probably continue to increase as the population of greater Vancouver further expands.  2.3.2 Timber Harvesting and Fisheries Impacts The production of sediment from hillslopes and sedimentation in low order streams are important land use concerns throughout the Pacific Northwest. Although increased sedimentation can be linked to many anthropogenic activities, timber harvesting is the most prominent of these. Sediment produced from mass wasting events and transferred through these systems may have important downstream  -25-  impacts on aquatic habitats and fluvial morphology in higher order channels. Although this thesis is not specifically concerned with issues surrounding forestry practices and associated land-use impacts on fish populations, it is relevant to provide some background discussion to demonstrate the significance of sediment budget and channel morphology studies.  Logging has been ongoing since the 1910s in association with early settlement and development in the valley. By the 1930s, much of the easily accessible timber along the main river valley flat had been harvested, so operations were extended into steeper tributary watersheds, where it continues today but at lower rates than in earlier periods (Hay & Co., 1992). A 1982 review of the watershed forest supply estimated 81,000 ha (810 km ) of forested land, including 37,000 ha identified as potentially 2  productive and 17,800 ha identified as mature forest available for harvest (Hay & Co., 1992). Jordan (1990) estimated about 15% of the forested land base had been logged by the late 1980s, with about one-third that area replanted or naturally revegetated and having reached "hydrologic recovery." Ministry of Forests data (as compiled by Hay & Co., 1992) show 50% of logged areas had been replanted and 75% had received silvicultural treatment by 1990. Only the forested area above Chilliwack Lake is protected and has had no logging, while half the forest has been removed in the Chipmunk Creek sub-basin, where numerous landslides can be observed.  It is generally accepted that rates of sediment production from hillslopes increase following timber harvesting (see Kelsey et al, 1981; Lyons & Beschta, 1983; Swanson et al., 1987), though it is not necessarily true for each individual watershed. In general, harvesting and road building practices degrade slope stability by undercutting hillslopes, loading of fill slopes, altering drainage and reducing the anchoring effect oftree roots (Clratwiner al, 1991). Many studies (e.g. Lyons & Beschta, 1983;  -26-  Rood, 1984) have shown that the overall effect of logging is to increase landslide rates and magnitudes over those in unlogged areas. These studies also report substantial increases in the amount of material that enters stream channels. To resolve sedimentation issues, natural rates of sediment production need to be quantified so that the effects of different land use strategies can be evaluated and the most benign (including no use) management strategy can be applied.  Forestry activities may have an impact on both the fine and coarse sediment cascades in coastal watersheds. Habitat quality in gravel-bed rivers is adversely affected by an increase in sand- and siltsized sediments. Numerous studies (see Phillips, 1971; Reid et al, 1981; Tripp and Poulin, 1986) show that the intrusion of fine grained sediments into spawning gravels degrades the quality of spawning habitat and ultimately reduces stocks. Common impacts include decreased permeability in redds, which affects the amount of oxygen reaching the eggs and removal of metabolites (Coble, 1961; Scrivener and Brownlee, 1989). Fines can also restrict the movement and emergence of alevins (pre-emerged young) by creating a physical barrier in pore spaces (Philips, 1971; Scrivener and Brownlee, 1989). Fines have also been linked to a reduction in populations of benthic organisms, thereby reducing the food supply for allfishspecies.  Although logging in the watershed has probably increased the rate of mobilization of fines from hillslopes, there are no data to quantify an increase (Jordan, 1990; Hay & Co., 1992). Available data records (WSC) for suspended sediment load between 1966 and 1975 indicate that the mean annual load remained relatively stable over the period of record, except during 1975 when there was a large (530 m /s) flood. However, given the relatively short data record and buffering effects of the 3  floodplain (which may delay channel response to sediment inputs for years to decades), it is difficult  -27-  to directly relate recorded suspended sediment yield to forestry activities on Chilliwack River. As well, fine sediment input and transport is controlled by events which occur episodically, including landslides, gully and tributary inputs, bank erosion and mobilization of bed sediments.  Most studies examining the impacts of forestry on channel morphology have dealt with relatively small channels. However, the morphologic (and ecologic) impacts of forestry on larger channels like Chilliwack River may be quite distinct and require a different study approach. It is appropriate (and often the only method available) to study channel morphology changes on large rivers from historical aerial photography. Some parameters of interest that may be related to timber harvesting include changes in active channel width, relative pool/riffle areas, gains or losses to off-channel habitats, changes in channel stability (pattern) and changes in sediment storage.  The primary factors determining local river morphology are the supply of water and sediment from upstream sources. As the morphology of a river provides the physical framework for fish habitat (see Kellerhals and Church, 1989), forestry activities can directly influence large rivers by affecting either the supply (or timing) of either water or sediment. Jordan (1990) stated that the overall effect of logging activities on the runoff regime in Chilliwack River is probably negligible given the size of the watershed. However, he found that logging-related sediment sources could have had a significant impact on the sediment supply of the river following the 1989 flooding. Numerous road-related failures were noted along Foley, Chipmunk and Liumchen Creeks while numerous debris slides and torrents were noted along Centre, Nesakwatch, Slesse and Tamihi Creeks (though many are natural). Beschta (1984) used historical aerial photos to examine the effects of harvesting on channel morphology along Middle Fork Willamette River, Oregon. He found that increases in landsliding  -28-  caused channel widening and aggradation in downstream reaches. The consequences of an increase in the width/depth ratio include increased water temperatures and filling of pools which are important rearing and feeding areas. Aggradation may also lead to lateral instability (eg. avulsions) thereby reducing access to important overwinter habitats in the riparian zone. Changes in sediment stability and storage can be demonstrated by the aerial photo / GIS approach presented in this thesis and will be discussed further in following chapters.  2.3.3 Fisheries resources Increasing sediment yield to streams is an important issue in Chilliwack River. The fisheries resource in the system is extensive, supplying anadromous salmonid fish stocks to Native, commercial, and recreational users both within the system and in the Fraser River and Pacific Ocean. The river, and associated tributaries and backchannels, supports large populations of pink, chum, chinook and coho salmon and both steelhead and cutthroat trout. In a pink salmon spawning year, total stock in the river is estimated at just over 1,000,000 (Hay & Co., 1992). It is one of the most important chum and coho streams on the Fraser River and is estimated to sustain 15% of its pink salmon run (Bonharn, 1980). This makes the Chilliwack the most productive tributary system in the lower Fraser (Farrell, 1987). Though it is difficult to put an economic value on this resource, the recreational revenue alone is estimated at $2.5 million annually (Chilliwack Valley Ratepayers, 1993,  unpub.) excluding those  fish caught outside the watershed. The pink and chum species are most at risk from increased sedimentation as they spawn in the lower, shallower reaches which are downstream of most logged tributary streams and the glaciolacustrine bank failures. These locations are also most susceptible to extremes of flow (causing local bedload scour and fill). Chinook, coho, sockeye and steelhead use pool and sidechannel sites further upstream which are more stable, so these species are at lower risk  -29-  overall.  The critical migration and spawning periods for salmonids in this system range between September and December (Farrell, 1987). However, when incubation and spawning periods for all species are included, only the time between July 15 and September 15 can be considered as non-critical (Hay & Co., 1992). This brief summer window is the period when activities such as gravel dredging can occur without harming the fish. However, sedimentation does not follow the same rules; inputs (both natural and forestry related) are typically highest in winter because of higher flows and precipitation. This is true, in general, throughout coastal British Columbia. An understanding of channel patterns and adjustments may lead to a greater understanding of the impacts of various land use activities on salmonid populations and may provide insight into restoring channel habitat.  2.4 Hydrologic Regime 2.4.1 Streamflow measurements There are four Water Survey of Canada (WSC) streamflow gauges operating within the Chilliwack River Valley; three are on the mainstem channel, while the fourth is located on the largest tributary, Slesse Creek (see Figure 2). There are partial records for several other gauges, but these were not included in the analysis because record lengths are insufficient to be of value, or there are many missing data. A summary of gauge information is given in Table 1.  -30-  Table 1: Water Survey of Canada gauging stations, Chilliwack River watershed Station  Name  Drainage Area  Period of record  08MH001  Chilliwack River at Vedder Crossing  1230 km  1911-1931 1951-present  08MH055  Chilliwack River below Slesse Creek  860  1956-1962  08MH103  Chilliwack River above Slesse Creek  645  1963-present  08MH016  ChilUwack River at outlet of Chilhwack Lake  329  1923-1951 1957-present  08MH056  Slesse Creek  162  1957-present  08MH033  Sweltzer River at Cultus Lake  65  1947-1964  08MH157  Liumchen Creek near the mouth  54.4  1985-present  08MH037  Tamihi Creek near Vedder Crossing  130  1950-1951  2  Note: Stations indicated in bold are used in thesis.  Historical mean and maximum annual daily flows at the Vedder Crossing gauging station (WSC gauge 08MH001) are shown in Figure 3a. Records are available for Vedder Crossing since 1911, though the data are based on manual stage records prior to 1968. As rating curves are seldom calibrated for high flow events, changing channel characteristics during floods can affect stagedischarge relations, so some error in the extreme data may be expected (Dunne & Leopold, 1978). There are also missing data for many larger flood events, and there is a gap in the data record from 1932-1951 when the gauge was not maintained. Estimates are given for the large unrecorded floods of 1951,1984, 1989 and 1990 as reported in Hay & Co. (1992); these are based on stage relations and flood hygrographs from various authors. Martin (1992) estimated the same floods using regression, but noted the weakness of that technique (large error) when applied to larger events.  2.4.2 Climatic Influences The watershed is characterized by a fairly typical west coast climate, with warm dry summers and  -31-  800  1911  1921  1931  1941  1951  1961  1971  1981  1991  Year Note: All Flow datafromWater Survey of Canada records as compiled on CD-Rom  Figure 3a: Historic record of mean annual and maximum annual daily discharge for Cfiilliwack River at Vedder Crossing. Note that many large floods are estimated (see text).  160.0  20.0 17.3  140.0 4120.0  15.0  100.0 01  CO  E 5 o  LL  80.0  -  60.0 4-6*- * -  10.0 t A  8.0  •8 4.4  5.0  40.0 20.0  "-0.0 0.0  0)  o  e  5.3  c  Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec  Figure 3b: Monthly flow regime for Chilliwack River at Vedder Crossing. Graph is based on 60 years of record. Numbers over bars are percent of annual flow.  comparatively cool, wet winters. Two generalized runoff peaks are associated with this climate: an autumn/winter peak caused by heavy rain or rain on snow events, and the early summer peak caused from melting of the winter snowpack (see Figure 3b). During the autumn/winter period, most precipitation falls as snow above 1000 metres (Martin, 1992). If accompanied by rain and warming temperatures, significant floods may result. Jordan (1990) notes that peak rainfall generated floods are consistently larger than snowmelt floods in the Chilliwack and many other coastal rivers in British Columbia. The largest recorded flood of 765 m /s in December, 1917 was the result of heavy rains 3  following a major snowstorm, as were the exceptional floods of 1989 and 1990. Summer snowmelt floods generally have a lower mean daily and peak discharge but are sustained for longer periods (eg. 5-7 days as opposed to 1-3 days) resulting in higher average monthly runoff totals. Although many summer floods are capable of transporting bed material, the autumn/winter floods are responsible for most of the historic flood damage along lower reaches of the Cliilliwack River. Between 1952 and 1991, the largest winter flood had an estimated mean daily discharge of 715 m /s, compared to 363 3  m /s for summer snowmelt floods. Martin (1992) adds that the magnitude of snowmelt floods was 3  greater in the period 1952-74 than 1975-90, though the reverse was true of rainstorm generated floods, possibly due to regionally warmer autumns and winters.  2.4.3 Flood Frequency Analysis Historically, large floods such as those that occurred in 1989 and 1990 have caused extensive bank erosion, so it is relevant to determine the probability of occurrence for floods of a given magnitude. Martin (1992) used historical strearnflow summary data to calculate cumulative percentage departures from the mean for the annual maximum daily discharges. These findings revealed trends in the data, indicating three distinct hydrologic phases. These were identified as 1912-1930 (mostly  -33-  above average flows), 1952-1974 (consistently below average flows) and 1975-1990 (mostly above average). As the photo record examined in this thesis (1952-1991) corresponds exactly to two of these distinct flow regimes, the subsequent discussion will most closely examine these periods in particular.  Many studies have examined the relation between flow frequency and the size of the flow to describe dominant discharge. The dominant channel forming flow proposed for most rivers is that which roughly corresponds to the discharge represented by bankfull stage. The return period for these flows typically ranges between 1 and 2 years for many larger rivers, although this definition is not universal. Eaton summarized several studies on bankfull discharge and confirmed the findings of Bray (1973) who suggested the 2-year flood is the dominant discharge in Alberta rivers and should also apply to British Columbia rivers. Bray (1973) adopted the 2-year flood as a practical compromise between the 2.33 (mean annual flood) and the 1.5 year (median flood) discharge. The flood-frequency deistibution based on the annual maximum series is given in Figure 4. The graph clearly illustrates two distinct flow regimes since 1952, the latter of which is characterized by an increase in the magnitude of the largest floods. Using data from 1952 to 1974 only, the 2-year discharge is calculated as 275 m /s. For the period 1975 to 1991, the 2-year discharge is calculated as 313 m /s. Another frequently 3  3  cited reference value is the mean annual flood (return period of 2.33 years in the extremal value distribution). For Chilliwack River at Vedder Crossing, the Q  is calculated as 328 m /s for the 3  23 3  period 1952-74 and 343 m /s for the period 1975-91. The mean annual flood flow has statistical, 3  though not necessarily any morphologic significance for a given river.  -34-  Figure 4: Flood frequency analysis based on annual maximum series. Graph is based on data from 1952 to 1991 inclusive (Vedder Crossing gauge). Data are plotted log-normally and best-fit regression lines are shown.  2.4.4 Frequency Analysis of Flows Capable of Transporting Bed Material For studies relating changes in channel morphology to sediment transport, the flows that are important in terms of modifying channel form and transporting bed material are not necessarily related to flowfrequency.A more useful referencefigurewould be some critical threshold flow, below which the bed is stable. However, the detailed observations required to make this distinction are not likely to be available for most alluvial channels. In the absence of data relating bed material transport rates to discharge, either the mean annual flood or the 2-year flow may be a practical lower limit for sigriificant channel forming floods. For the Vedder reach of Chilliwack River, there are reasonably good datafromprevious studies that relates significant bed material movement to flow. Since these are the flows that are of interest in a study of morphologic channel changes, subsequent discussion will examine these "critical" threshold flows in particular.  The movement of bedload is generally episodic, as flows competent to entrain and transport bed material do not occur frequently. However, once flows exceed initial entrainment conditions, transport rates (and size of materials) generally increase. McLean (1980) observed that the D  5 0  of  trapped bedload increased between 225 and 250 m /s, above which bedload consisted primarily of 3  gravel and below which it was mostly sand (see Figure 5a). He also found that bedload and bed material size distributions for the Vedder River are similar above this threshold, indicating that all grain sizes are mobilized. Additional data on bedload transport rates were collected using trap samplers over a range of flow conditions (see Figure 5b). Although the data exhibit considerable scatter (and may show the effects of trap inefficiencies), the graph shows a trend of increased transport with discharge (note difference between sampler types). In the absence of direct bedload samples collected along the Chilliwack River, it is reasonable to adopt the 250 m /s threshold as a 3  -36-  100  £ E  10  o Q in  TS 10  "oD CD  CD  0.1  Note: DatareproducedfromMclean (1980). Original datafromWSC measurements, 1971-1975. Bestfitlines are approximate only. — i —  o  1  50  — i —  1  100  — i —  150  1  — i —  200  —i  250  1  300  1  350  400  Flow (m3/s)  Figure 5a: Bed load grain size on Vedder River near Yarrow  • Basket *VU  1000  •  a  -D  O  n  100 • A EA  TS CO O  *  ~u  •  A A  A A A  CD  CD  •  A  10 A  A  U  1  &  A  •  n  50  100  150  200  250  300  350  400  Flow (m3/s) Figure 5b: Relation between discharge and bedload transport on Vedder River near Yarrow. Bedload is directly measured from trap samplers.  lower limit for flows capable of transporting bed material.  Frequency counts illustrate the variability with which these flows can occur ( Figure 6a). In any given year, the number of mean daily flows exceeding this threshold is as high as 14; these are generally associated with periods of sustained high flows (see Figure 9) which can break up the bed surface and transport large quantities of sediment. Also significant is the number of years with few or no flows >250 m /s. Morphologic evidence based on photographic records will be used to support the 3  assumption that little bed material transfer occurs during these years (refer to Section 5.0).  A partial duration series may also be used to analyse flood records. It differs from the annual maximum series in that the average frequency of occurence of floods of a given size is determined, irrespective of when they occur (Dunne & Leopold, 1978). Rasmussen (1991) notes advantages of the partial duration series include higher accuracy, as the frequency distribution is estimated from a larger number of data points and better fit to theoretical distributions for smaller sample sizes. The principle advantage, of course, is that it includes only flows that are important in terms of some threshold condition. The partial duration series was calculated using the bed material mobilizing threshold of 250 m /s as a base flow. All daily mean flows which exceeded this value were selected 3  for the periods 1952-74 and 1975-91. As the number of flows exceeds the number of years of data for both periods, calculated return periods were scaled. For each period, this was completed by multiplying Tby the ratio number of years of record (n) to the number of observations. For example, there were 66 flows >250 m /s between 1952 and 1974 (23 years), so the return period was estimated 3  by:  T  PDS  V  66,  [2.1]  -38-  8 m 1952-74 • 1975-91 7  Figure 6b: Flow-frequency analysis based on partial duration series. All mean daily flows > 250 m /s between 1952 and 1991 are included. 3  The flood frequency curves for these data are shown in Figure 6b. Based on these observations, flows capable of transporting bedload have occurred 2.9 times per year on average between 1952 and 1974 and 3.4 times per year between 1975 and 1991. Comparative figures for the 2-year discharge based on the partial duration series are 330 and 440 m /s respectively. 3  Although flows greater than 250 m /s have been discussed as significant in terms of moving bed 3  material, flows near this threshold are unlikely to cause significant morphologic change. Therefore, channel morphology within a period during which there were no large floods will not change enough to be detected using historic aerial photography. Corresponding transport rates would be estimated as 0 m /yr ± measurement error. If the morphologic method is to be used for estimating bed material 3  transport rates, it is important that chosen survey dates roughly correspond to periods within which there has been significant gravel transported. The choice of a threshold discharge representative of major morphologic changes remains a significant point of discussion. McLean (1980) estimated that damagingflows(i.e. cause extensive bank erosion) occur at roughly 500 m /s or larger on Chilliwack 3  River. Martin and Church (1995) show that gravel transport rates past Vedder Crossing (downstream boundary of Reach 1) are small when peak flows (in 1-3 year study periods) were below 350 m /s. 3  Both of these values are significantly larger than the threshold value for bed material transport. The 2-year discharge based on the partial duration series may be a resonable compromise between these differentfigures.It is proposed that these flows (330 m /s from 1952-1974 and 440 m /s from 19753  3  1991) be accepted as reference flows, below which there is minimal change in the morphology of the channel. The validity of these reference values can be verified from morphogic evidence and sediment budget transport calculations.  -40-  It is interesting to note the incidence of large flows during the 1975-1991 period which includes 5 flows >500 m /s (compared to none in the earlier period). These high flows may be linked to a similar 3  period of high-magnitude, low-frequency rainstorms (see Church and Miles, 1987). There are several possible implications of the observed increase in the highest discharges. It is probable that the channel has adjusted its geometry in response to the first major flood (in 1975) by widening along lower reaches to accomodate larger flows (these reaches are not constrained by geology). Conversely, the channel may have degraded, increasing the capacity of the channel to convey flow. Increased sediment transport rates are expected for 1975-91 compared to the earlier period as the frequency of high magnitude flows (and of flows >250 m /s) is higher. It is known that these large flows have 3  the ability to transport significant quantities of sediment and they are known to modify channel morphology. Morphologic evidence and sediment budget calculations can further be used to determine these impacts.  2.4.5 Reach Strearnflow Estimates To consistently interpret aerial photographic evidence of channel morphology changes, strearnflow estimates for each study reach are needed. Reach flow estimates are used to calculate estimates of varying water depths and to correct channel bar exposure for the effect of different flows (stage changes). As reach breaks primarily occur where major tributary streams enter the Chilliwack River, flows should show a step increase in the reach immediately downstream of these tributaries. Within a given reach, however, an assumption is made that small streams, gully runoff and groundwater discharge contributions are not significant such that the flow within each reach is assumed constant. For each date of photography used, flows at each station were recorded from the WSC records and are provided in Table 2. An artificial 5th gauge (sum of Slesse and U/S Slesse) is not included as this  -41-  additional datum was found to have little influence on results.  Table 2: Streamflows for different dates of photographic records Flightline  Year  Date  Reach  Lake (329 km )  U/S Slesse (645 km )  Slesse (182 km )  Vedder (1230 km )  2  2  2  2  BC1622/23 BC1805/05  1952 1953  Oct 2 Sept 15  1-7 8-15  5.6 m /s 13.7  8.1 m /s 24.0  1.4 mVs 6.5  15.5 mVs 45.6  BC5215/13 BC5215/17  1966 1966  Aug 24 Sept 4  7-15 1-6  13.0 10.3  21.2 15.9  5.7 3.7  35.1 26.5  BCC088  1973  Sept 11  1 -15  7.5  12.5  4.2  20.4  BC83013  1983  July 22  1 - 15  37.8  60.2  15.9  109.0  BCB91157  1991  Sept 9  1-15  16.8  25.7  5.8  46.7  3  3  The 1952/53 data are based on regression estimates as only the Vedder Crossing gauge was operating on those dates. These estimates were made by comparing five years of daily flows (1963 to 1967) between Vedder Crossing and the other stations. This period was chosen to provide a sufficiently large sample of data and because this is the nearest period to 1952/53 in which all gauges were in operation. The complete channel was sometimes flown on two separate dates, so flows for each are tabulated separately. The anomalous result is that downstream reaches (eg. reach 7, 1952) may have lower flows than upstream reaches (eg. reach 8, 1953) when flows are higher during the upstream photography.  A basic observation common to most rivers is that streamflow is strongly correlated to drainage basin area and is typically represented as a function of the form Q = aAf (Gregory and Walling, 1973). The relation is consistent for watersheds with homegenous regional characteristics (surficial materials, geology, vegetation, precipitation) and for sub-basins within a larger watershed. These observations provide a basis for estimating runoff in ungauged basins (Chow, 1964). Discrepancies mainly result  -42-  from varying relief in the sub-basins and response rates to lake and groundwater storage, for example. By adopting this functional relation, the stepped increase in flow to each reach can be estimated. This involves determining the contributing area to each channel reach from various sub-basins.  A map showing 18 sub-basins within the Chilliwack watershed (including that portion in the United states) was digitized, and areas of each were calculated in Arc/Info GIS. The contributing area for each reach was then calculated as the sum area of all sub-basins flowing into, or located upstream of that reach. For each date of photography, regression was used to detennine the relation between basin area and strearnflow for each gauged site. Data were not transformed logarithmically for the analysis as this did not produce higher r values and predicted values showed greater variance. 2  For the 1952 data (refer to Table 2) the calculated relation is:  Q = 2.08 + (0.0366 *A )  [2.2]  d  This equation has r = 0.986 with a standard error estimate of ±2.49. A high r value (indicating a 2  2  strong or near perfect relation) is not unexpected with so few data points, but in reality is quite rare in the natural environment. The calculated value of t to determine the significance of the regression slope is 12.0. The critical value of t  0 1 2  (99% confidence) is 7.0; this means that the slope is  significantly positive in this case and the relation is meaningful. Similarly, slopes based on flow data for all other dates of photography were found to be significantly positive.  For each date of measurements, predicted values for reach 15 underestimate recorded values. This is not surprising given that the reach is lake controlled, 600 metres higher in elevation and 40 km  -43-  distance from Vedder Crossing. Regression of flows at the Lake outlet with those at Vedder Crossing showed 23% unexplained variation between the gauge sites. However, the magnitude of differences is not large enough to affect results. Figure 7a gives predicted strearnflow values for each reach and illustrates the assumed 'stepped' nature of flow downstream from the lake. Figure 7b compares predicted flow estimatesfromequation [8] with recorded flow at gauge sites (both figures are based on 1991 strearnflow data).  -44-  Figure 7a: Reach streamflow estimates  co E,  g o  LL  10  o  -I  250  1——I 350  450  1  1  1  1  1  550  650  750  850  950  1  1  1050 1150 1250  Drainage area (km2)  Figure 7b: Predicted streamflow estimates in comparison with gauged streamflow -45-  Chapter 3: Channel Change Data  3.1 Aerial Photographs and Maps Aerial stereo photographs have been taken by Federal and Provincial Governments since the 1930s as part of regular resource inventory programs and mapping surveys. As these photographs provide an accurate and permanent record of land-use activities, historical sequences can be used to provide summary information of landscape evolution. Kellerhals et al. (1976) note that photographic records of rivers and associated landforms are more complete than any other type of river records; this is particularly true of smaller and regionally isolated rivers. Despite the availability of these images and their potential value for both geomorphologic and engineering studies, they have been mainly used for land-use mapping, resource inventory and the interpretation of landforms. This project will demonstrate how photographic records can be used to provide quantitative data on river morphology changes.  The photographic record within the Chilliwack River watershed is extensive, dating from 1940 to 1993. A list of all catalogued (by Provincial government) stereo photography covering the entire river between Vedder Crossing and Chilliwack Lake is given in Table 3. Uncatalogued photographic records may also be available through the Federal government and private agency collections. Lower reaches of the river (generally below the Slesse Creek junction) have available stereo photographs for many additional dates including 1960, 1963, 1964, 1974, 1976, 1979, 1982, 1989, 1990 and 1993. Many of these photos were taken as special large scale (e.g. 1:5000 to 1:10000) projects.  -46-  Table 3: Available stereo photographic coverage for Chilliwack River  FLIGHTLINE  YEAR  DATE  SCALE  FLOW (m/s)"  BC207/209  1940  July 17  1:30000  no data  BC1622/23 BC1804/05  1952 1953  Oct 2 Sept 15  1:40000 1:40000  15.5 45.6  BC2151  1956  May 15  1:32000  76.2  BC5212/13 BC5215/17  1966  Aug 24 Sept 4  1:32000  35.1 26.5  A22234  1971  June / July  1:27000  n.d.  A23080/84  1972  June / July  1:49000  n.d.  BC5492  1972  July 30  1:70000  122.0  BC7472  1973  July 14  1:20000  81.6  BCC088  1973  Sept 11  1:20000  20.4  A24506  1976  June / July  1:50000  n.d.  A24777  1977  June / July  1:50000  n.d.  BC78101  1978  July 7  1:50000  98.0  BC80065  1980  July 21  1:57000  66.6  BC83013  1983  July 22  1:20000  109.0  BCC538  1986  July 9  1:10000  71.8  BC88073  1988  Aug 23  1:70000  26.2  BCB91157  1991  Sept 9  1:60000  46.7  3  a: flow given at Vedder Crossing (WSC gauge 08MH001)  In general, it is desirable to extract a sufficient photographic record such that: 1) coverage is always available either before or after each major flood event and 2) each date of coverage shows a similar river discharge, thereby rninimizing the need to perform water level adjustments For well documented areas such as the Chilliwack River, these two criteria can generally be met. However, some photographic records may not be suitable for inclusion in the study. Older  -47-  photography (pre-1950s) generally has too much radial distortion to be used by stereoplotters for example. Additionally, some scales may not be suitable for the area of interest. This is a function of the number of photographs needed and the minimum resolution of the images. Larger scales (ie. 1:5000) should be used only for smaller rivers; for larger systems, a single photograph may not even cover the bankfull width of the channel and hundreds of photographs will potentially be needed. Similarly, small scale (ie. 1:60,000) photographs are not useful for examining narrow channels or small instream features such as logjams. However, if there is little historical coverage for the area of interest, all available photography may have to be used simply to provide a sufficient data record. For further discussion on the suitability of different photographic scales for stream channel studies, see Ham (1995).  Other potentially useful sources of data include orthophoto mosaics and plariimetric maps. Orthophoto mosaics (reproduced as 1:5000 maps) are available for the Cliilliwack River based upon 1976, 1986 and 1991 photography, but coverage extends upstream to Slesse Creek only. For morphologic studies, planimetric maps are primarily used to orient aerial photographs, but they may also be used to further extend the data record. McLean (1990) found legal township survey maps showing islands and channel banks in the lower Fraser dating back to 1876. In most cases, however, the validity and accuracy of such maps must be scrutinized. For this project, 1:5000 planimetric maps of the lower six channel reaches are available, as well as regional district planning maps, forest cover maps and topographic maps at 1:20000 (Provincial) and 1:50000 (Federal) scales.  3.2 Photo Record Selection Photo records from 1952/53,1966, 1973,1983 and 1991 are used in this study. The period 1952 to  -48-  1991 includes the second largest estimated flood of record (since 1911) and most anthropogenic landuse changes. Flows for all dates are relatively low (15.5 to 46.7 m /s) except for 1983 (109 m /s). 3  3  Mean daily discharge at Vedder Crossing over the period of record is 60 m /s. Water level 3  adjustments are necessary when discharge on comparative dates of photography is different. This is particularly important for the 1983 mapped data to account for lower relative bar exposures due to higher water levels (which gives a false impression of erosion). Figure 8 relates the selected photo dates to flood flows, defined as flows greater than the 2-year discharge (refer to Section 2.4.4). Between each set of successive photo dates, there are three or four flow events that exceed the critical or 2-year discharge. Hydrographs based on mean daily discharge for most of these events are given in Figure 9. This graph shows that most high duration flow events occur for short periods (1 or 2 days maximum). Notable exceptions are floods in May, 1974 and November, 1990, when peak flows were followed by a temporary decrease in flows, then another increase. Martin (1992) suggests these events may be particularly effective at eroding and transporting bed material as the initial peak flow loosens the bed surface, which may not stabilize before the second peak flow.  It is obvious from the above discussion that considerable volumes of sediment transfer may occur between photo dates used in this study. In general, information on sediment which is stored, then entrained within the time resolution of the morphologic period is lost. Therefore, calculated transport rates represent lower bound estimates of actual transport rates. However, if the major floods are temporally isolated or erosion and deposition sites are spatially distinct, biases in the transport estimates may not be large. Temporal spacing of the surveys is important as the longer the time between surveys, the greater the probability of compensating scour and fill (Ashmore and Church,  in press). Lane et al. (1994) have shown that scour and fill volumes are increasingly underestimated  -49-  Figure 8: Significant floods in relation to photo dates used in study  -50-  800 to  700 i  600 _  at  a.  500  m E  400  o  100 1111111111111  I I I I II I I I II II I I I I I I I I I I I I I I I I I I I ! I I I I I  10/53 11/55 11/63 6/68 1/71  6/72  5/74  12/75 12/79 12/80 1/84  11/89  11/90  Flood date  Figure 9: Mean daily flood flow sequences for period of study  -51-  as the length between successive surveys is increased. Ideally, photo surveys would be available immediately before and after major floods, but this simply is not practical or possible in each case. Further, total scour and fill volumes may be roughly equal at some time scale (i.e. the channel is in equilibrium) yielding minimal change in storage. In this case, it is important that erosion and deposition sites be distinct, or information on morphologic changes is lost (again, resulting in biased estimates). This can be evaluated by examing erosion and deposition of island and fioodplain sediments as these sites are not subject to variations in flow (which can mask "actual" changes of gravel bar sediments). This can also be evaluated by exaniining morphologic changes at different temporal resolutions (see discussion below).  Despite these technical constraints, the study still allows a relative comparison of erosion, deposition, storage volumes and transfer rates to be made. As well, useful information on the long term sediment cascade of alluvial channels can still be gained and the relative stability of individual reaches can be examined. To improve the resolution of measured morphologic changes, several additional dates of photography were chosen for reach 1 to characterize the changes that occur in response to; 1) no events greater than 250 m /s 3  2) an event equal to 250 m /s 3  3) an event equal to the 2-year discharge and 4) an event greater than the 2-year discharge There is considerable debate as to whether high-magnitude, low-frequency flows or low-magnitude, high-frequency flows are most responsible as agents of sediment transport and channel change. Church (1992) points out that there is no universally consistent correlation between flood frequency and effectiveness for causing morphological change as there are many spatial and temporal factors  -52-  governing these changes for a given river. McLean's 250 m /s threshold corresponds to a relatively 3  frequent, low-magnitude event that is competent to transport bed material in lower channel reaches. By mapping the channel morphology for two dates between which events equal to, or less than this threshold occur, the transport regime in relation to this flow can be examined. It is expected that observed morphologic changes in response toflowsnear the threshold for bed material transport will be limited to bar migratioa Conversely, erosion of islands andfloodplainsediments is not expected, and in fact, channel recovery (defined as floodplain reconstruction and the re-vegetation of bar surfaces) may be observed if the time scale between significant sediment transporting flows is sufficient. Larger, less frequent flows have been observed to cause significant bank erosion and are predominantly responsible for shaping channel geometry. Photographic evidence will be used to estimate the magnitude of flow required to modify channel configuration. This knowledge can be used, for example, to estimate bank erosion rates (and even locations) in relation to the 2-year or larger flows.  In Section 2.4.4, it was proposed that the frequency of flows capable of transporting bed material and the magnitude of largeflowshad increased. It is possible that the channel may have adjusted to a new equilibrium condition in response to the increased frequency of significant floods during 1975-91. The floods of 1974 and 1975 may have been 'catastrophic' events, resetting the pre-flood river regime (cf. Desloges & Church, 1992). By contrast, there were no floods between 1952 and 1974 that significantly exceeded the threshold for significant floods, as determined for post 1974 flows. As there have been additional large floods in subsequent years, the channel has likely remained in this new regime, rather than recovering to the previous regime. As a result, average rates of sediment transfer may decrease over time as the channel is able to accomodate higher flows, so the effects of  -53-  moderate flows on morphology may be minimal.  3.3 Methods of Data Collection and Analysis 3.3.1 Photo Overlays In using the morphologic approach, one is concerned with measuring net volumetric changes in erosion and deposition over time, which is can be completed using historical aerial photography. These changes are used to calculate the A V term for the sediment budget for a given reach. Past studies of the morphologic approach involved superimposing mylar maps of traced images from the photographs, as described in Church et al. (1987). One year is traced to produce a base map containing all of the features of interest, as well as a number of planimetric control points. Control points are positions on the ground which do not change over time, such as road intersections, bridges, and buildings, such that positional changes in the river can be determined relative to these stable points. A second map from another year of photography is made by transferring the control points to a mylar sheet and scaling the photography (using a zoom transferscope) to match with the control points. This produces two maps at approximately the same scale which can be overlaid for analysis of areal differences.  Church et al. (1987) note that this approach has the advantage of fairly rapid mapping from a large number of photographs without the need for external survey control. However, accurate planimetry is complicated by the displacement of images due to camera tilting and topographic relief on the photographs. Therefore, the mapping may not reflect the true planimetric position of features, and as mapping proceeds upstream or downstream, these planimetric error displacements may be propagated. As the elevation or relief of the land changes, another complication arises - the scale of  -54-  the photos changes, and is difficult to determine precisely. These effects are minimized on alluvial channel reaches which have a low gradient, but become pronounced on steeper upland reaches, tributaries and over longer distances (eg. mainstem channel length). This problem is magnified by the fact that distortion on photographs increases radially outward from the centre of the photo, particularly for older (pre-1965) dates. The resulting complex geometry means that when direct linear or areal measurements are made from the overlays, scale changes may introduce significant measurement errors. By digitizing directlyfromaerial photographs using a stereoplotter, the features of interest can be accurately mapped, and these problems can be avoided or rectified.  3.3.2 Analytic Stereoplotter An analytical stereoplotter is a photogrammetric tool which mathematically relates two-dimensional image coordinates on photographs with the real-world three-dimensional ground coordinates of the same objects (Slama, 1980). Data were collected for this project using the Carto API90, a PC-based analytical stereoplotter. Aerial photos are oriented using standard photogrammetric equations for interior, relative and absolute orientation. Absolute orientation means that actual ground coordinates (analogous to the U T M grid on a topographic map) can be measured directlyfromthe stereo image. Points for orientation (control points) were bridged from 1:70000 TRIM diapositive airphotos. The T R I M photos are used to produce new 1:20000 scale topographic mapping throughout British Columbia. These photos contain up to 7 control points, the locations of which are detenriined through field surveys and triangulation. Each control point is marked with a reference number and can be associated with a coordinate file. Stated accuracy of individual points is roughly 0.5m rms or less. Absolute orientation coordinates may also be obtained from topographic maps, but with greater error (~ 5-10m rms).  -55-  The process for measuring changes with the stereoplotter is analogous to that earlier described for photo overlays. A set of photos for a given year (usually the most recent set, or scale closest to the TRIM photo date) is oriented with respect to the coordinates on the TRIM photos. Once this step is completed, control points are 'bridged' to the other years of photography such that all distortion and displacement errors are systematically minimized or removed. Bridging is a process by which the coordinates of a point(s) on the oriented stereo pair are transferred to the corresponding point(s) on the other photography. Through this process, all orientation parameters and scales are preserved so that direct overlays are positionally correct with respect to each other (relative accuracy) and the base photos (absolute accuracy).  There are several sources of errors that affect digitizing accuracy. The first of these arises from the resolution of the stereoplotter. The specified resolution of the micrometers is 20 microns, which corresponds to 1.4-m actual ground distance at 1:70000 scale; this accounts for most of the observed relative error. Second, there is placement error both in reading the true x,y,z coordinates of a point from the API 90 and digitizing the position of each reference point for absolute orientation. At a scale of 1:70000, a distance of only 0.1mm on the photos represents 7-m actual ground distance. This means that if one is not exact in placing the floating marks on the object or feature for which coordinates are to be obtained, then an incorrect set of coordinates will be recorded. Similarly, when digitizing points on the photo for absolute orientation, one can easily misplace the true position of the reference marker. At typical working scales of 1:15000 to 1:30000, this gives an additional error of 1.5 to 3m, assuming a placement accuracy of 0.1mm.  Given these problems, real observed errors are fairly small. The stereoplotter offers greater image  -56-  resolution over mirror stereoscopes or conventional photo overlays as the stereo image is rectified, has no parallax distortion, and is magnified by a factor of 6; these factors all greatly enhance image resolution, hence interpretation. A n experienced operator can digitize points with an absolute error that is commonly less than 2-m rms. On older photography which is partly distorted, this error can be roughly doubled. This means that a feature digitized over two consecutive periods will be out no more than ± 2 . 8 - m Over large distances such as 200 metres, the potential errors are small. However, interpretive errors caused by shadows, changing water levels, different scales and sun glare may all further reduce accuracy. A method of determining this error is to digitize the location of features that can not change over time (eg. bridge, bedrock confined sections of channel) and record the displacement. Errors are further discussed in Section 5.0.  3.3.3  Geographic Information Systems  A Geographic Iriformation System (GIS) is a set of software tools for storing, organising and analysing spatial data and associated attributes in a computer. GIS allows the integration and display of geographic information in a manner that readily permits understanding of the spatial relations in the data; this approach may simplify, or even broaden aspects of research and provide a basis for better management decisions. The use of GIS in many land based activities (including forestry, urban planning, and infrastructure design and maintainance) is well known and it has a recognized potential for managing enormous volumes of geographic data and assisting with environmental management problems. Despite this, GIS use has not made significant inroads into geomorphologic studies. In particular, there are few references in published literature concerning the use of GIS for studies in fluvial environments apart from mapping purposes and flood impact analysis (eg. determination of flood limits). Graf (1984) first implied the use of a raster GIS for predicting the nature and location  -57-  of channel migration (and resulting erosion damage) though its actual use was not explicitly stated. This thesis will describe the use of a GIS for mapping and measuring historic channel changes along the Chilliwack River and discuss its effectiveness as a research tool. Data calculated within the GIS framework are used as inputs for sediment budget calculations, to measure bank erosion and channel migration rates, to illustrate patterns of channel recovery and to predict patterns of future instability.  The study of sequential morphologic changes on rivers is a practical application of GIS. Fairly precise information about these changes (and processes) can be obtained if sequential imagery is available; if exact dates are known, quantification of these changes can also be made (Verstappen, 1977). This type of analysis also has applications for studies of coastlines, glaciers and aeolian dunes. For rivers, sequential maps or airphotos of the channel are obtained, relevent features (bars, islands, banks, fioodplain) are digitized and the data are overlaid. If, for example, the bankline for two separate years is combined, observed differences in position can be attributed to bank erosion or fioodplain reconstruction. Dividing by the time interval between the two periods establishes the rate at which this process occurs.  Overlays are one of four main types of spatial modelling functions found in GIS (ESRI, 1993). In the context of overlay analysis, several operations can be performed on the spatial and associated nonspatial attributes of geographic data to answer specific queries (Burrough, 1986). In the simplest case, maps of individual classes of features in a river channel (e.g. gravel bars) correspond to single layers in the GIS; these layers can then be displayed graphically or plotted together (see example, Figure 10). Each layer of mapped data can be stored as a polygon, or closed areal feature (e.g. an island). Use of this function generates a resultant polygon cover containing the unique attributes of  -58-  the individual layers. Burrough (1986) uses the following expression to define this relation: U=f{A,B,...}  [3.1]  As an example, let A and B represent the same gravel bar in a river for two different periods. The intersection (U) of these features then becomes a new polygonal feature, BarStable, which represents that part of the bar which is stable, or has not changed its apparent position over time. Other parts of the resultant overlay are defined as new (deposition of sediment) or missing (eroded and transported away) using a modelling algorithm. Typically, resultant polygons can be identified, classified and analysed using the rules of Boolean logic [AND, OR, X O R (and/or), NOT] to test whether a particular condition is true or false (Burrough, 1986). This is the main method used for analysing and calculating rates of changing channel morphology along the Cliilliwack River.  The GIS offers a number of advantages over manual overlays in terms of speed and accuracy of measurement. As all photo data are digitized according to the same reference system (ie. NAD83 datum, U T M projection) using the stereoplotter, they are exactly scaled to each other in the GIS. Thus, when overlays are produced and statistics compiled, one can be assured of measuring absolute changes within the limits of mapping error. Given the large number of erosional or depositional changes that may exist over several comparative periods, the speed of GIS over manual methods of calculating these changes is advantageous. Variables such as reach length and channel width may also be calculated quickly and efficiently. To calculate reach length, one simply has to trace a line that roughly follows the channel thalweg (path of deepest or fastest flow). A final important consideration in choosing GIS is the ability to perform additional spatial modelling functions. The GIS can be used to create a wide range of potential data analysis capabilities by combining analysis techniques, or  -59-  writing special simulation programs (Burrough, 1986). For example, if mapped data layers of forest cover, soils, terrain stability and topography were linked together and tied to algorithms for slope failure, hazard maps could be produced to predict sediment delivery to river channels. It may in fact be possible to link GIS with a watershed model to predict (estimate) patterns of channel morphology.  All GIS work in this thesis was completed using ARC/INFO, a vector-based GIS operating under an X-windows graphic environment on the Department of Geography network of Sun Sparcstations. ARC/INFO was chosen over PC-based GIS systems as the Sun workstations offer superior processing speed and are able to work with largerfilesmore efficiently. As well, the API90 is able to convert digitized data into a file format that ARC/INFO can import directly.  3.4 Creating and Analysing Chilliwack River Spatial Data All features digitised from the airphotos are imported into Arc/Info using the generate command. Each feature of interest (e.g. islands) is imported as a separate layer, known as a coverage. Mapped features are represented by a series of lines and polygons (lines with the same start/end coordinate) but have no attribute information attached. The procedure of attaching attribute information to lines and polygons is known as constructing topology and is completed by connecting all nodes, or the set of connected x,y positions that define lines and polygons. In overlay analysis, all features must be represented as polygons, including the river boundary which is 'closed' at either end. For each year of digitized data, features for bars, islands and channel banks are combined into a single temporary layer to ensure that contiguous features (e.g. lateral bar and channel bank) share common boundaries. This prevents the formation of sliver polygons, or false small polygons formed where common lines overlap. To construct bar polygons, for example^ the temporary layer is edited by removing all lines  -60-  that do not define bars. This ensures that the line defining the boundary between a lateral bar and the channel bank is exactly the same for the river polygon, thereby eliminating slivers when the two layers are overlaid.  Once this editing is complete, topology is constructed by using the arc command build; this constructs polygons with a unique data identifier and basic attribute information including perimeter and area. The polygon identifiers for each are also modified from randomly assigned values of 1 to n, to the year of mapping. For example, a bar with a random ID value of 204 will be changed to 66 for the year 1966. All coverage layers for a given year are then combined with each other (union command) and a layer that defines the 15 channel reaches (identity command). The final overlay coverage consists of gravel bars, islands, channel boundaries and a reach identifier. This layer is further edited to account for the physical area in the channel not occupied by bars and islands (ie. the water), and for the occurence of islands on gravel bars. This step is completed in the Info database; an example is given below: Select for Island-ID = 66 and Bar-ID - 66 Calculate Bar-ID = 0 This prevents areas from being counted as both islands and bars and results in several bar polygons with holes'. The final coverage layer can be used to identify areas of bars, islands and channel (water surface) for each reach. Figure 10 shows final channel maps for two periods in study reach 1. Differences in morphology between these two years are evident.  Channel maps constructed for each year of photography are combined with successive years (also with union command) to produce channel overlay maps (e.g. 1973 with 1983). A temporary map  -61-  representing the union of the two years is initially made, then intersected with the reach coverage to ensure all polygons are correctly identified by location. The final coverage contains attribute data of individual channel features for both years. The coding system described above allows morphologic changes to be shown graphically, and resultant changes calculated from the Info database. An example channel change map is shown in Figure 11 (top). This map presents the results of combining both channel maps shown in Figure 10. Thefinalmap, where all possible transitions are considered, is complex; maps for the entire river contain several thousand polygons. This type of map is too complex to be developed or analysed by hand and demonstrates the effectiveness of GIS for overlay analysis. A simplified version of this map (bottom) summarizes the fifteen different transitions by process. These processes include stable (no change), erosion (eroded by river), deposition (new deposits), stripping (removal of overbank fines) and recovery (floodplain reconstruction and vegetation of bars).  The colour coding system on the map is used to illustrate the types and locations of change that occur. Yellow regions, for example, show stable gravel bars; these are locations where bar material has not been eroded, or may be locations of compensating scour and fill. These areas are selected by the following boolean rule: IF Reach-ID = 1 A N D Barsl966 = 66 A N D Barsl973 - 73 All other transitions are similarly defined. Red colour codes illustrate bank erosion, regions that are classsified as floodplain in the earlier period, but are covered by active channel (water) in the latter. These areas are selected using the following rule: IF Reach-ID - 1  A N D Channell966 = 0 A N D Riverl973 = 73  If the areas of these polygons are divided by their respective lengths, bank recession rates can be  -62-  Reach 1 channel maps  N  A  1966  |  | Channel boundary Islands Gravel bars  1000  Figure 10: Channel maps for reach 1, 1966 and 1973  0 Meters -63-  Chilliwack River Overlay Maps Reach 1  v.  TRANSITION (1966 - 1973) Bar to Bar Bar to Island H i Bar to River Bar to Fioodplain Island to Bar Island to Island Island to River Island to Fioodplain River to Bar River to Island River to River River to Fioodplain Fioodplain to Bar Fioodplain to Island Fioodplain to River  PROCESS Erosion Stripping Stable Recovery Deposition Figure 11: Overlaid channel maps for reach 1  500  0 Meters -64-  determined by specific, reach averaged, or channel averaged location. These maps form the basis for most discussion on observed channel changes and are used to provide input data to the sediment budget; detailed results are presented in Section 5.0.  -65-  Chapter 4: Historic Channel Changes  This section of the thesis will examine the morphology of Chilliwack River on different dates and discuss changes as observed from sequential aerial photography and GIS analysis. Thefirstpart of the chapter will examine changes in channel pattern and lateral channel stability both downstream and over the period of record. The response of the channel to the large floods between 1975 and 1991 is investigated to determine their role in shaping channel morphology. The second part of the chapter will examine net observed volumetric changes over time in the context of a sediment budget. Between any two time periods, the observed net storage change (A V) represents the incremental increase or decrease in storagefromthe earlier period. Areas of erosion and deposition are determined in the GIS and multiplied by measured or estimated sediment deposit thickness to determine volumes for incorporation into the sediment budget. This further allows unit transfer rates of sediment transport to be estimated using a distance of travel assumption. An interpretation of these results is presented.  4.1 Morphology of Study Reaches An inspection of aerial photography along Chilliwack River reveals that the channel undergoes several distinct changes in character between Chilliwack Lake and Vedder Crossing. These changes are most easily recognized from photographs as changes in channel pattern, but include changes in sediment texture. Kellerhals et al. (1976) and Church (1992) note that the primary factors governing river morphology are the volume and timing of water flows and sediment transport (including type) from upstream, the nature of materials through which the river flows and the geomorphic and geologic history of the fluvial landscape. In this context, the river can be divided into different reaches, within which the factors governing morphology do not change appreciably. The most  -66-  noticable changes occur downstream of the confluence with Slesse Creek. Upstream of this location, the channel consists primarily of stable, sinuous reaches; below the confluence, the channel widens considerably and becomes braided in the lower reach. The river meanders locally, but most meander bends are partly confined. Rivers exhibiting this pattern of irregular instability have been termed wandering gravel bed channels (after Neill, 1973; in Church, 1983).  The Chilliwack River has been divided into 15 distinct study reaches (refer to Figure 2). Reach breaks were chosen at the confluence of major tributaries and where distinct changes in morphologic character could be observed. A brief description of reach characteristics is given below to place observed changes in context of their geomorphic setting. To simplify the discussion, several reaches are grouped together where their morphology is similar.  Chilliwack Lake to Foley Creek Above Foley Creek, the channel is fairly well incised,flowingwithin a broad U-shaped valley. Several large terraces are visible, evidence of downcutting through outwash sands and gravels within the former valley floor. The river flows in an irregular pattern along the the south side of the valley, its movement impeded and controlled by the terraces and bedrock walls. Between Reaches 11 and 13, there are multiple channel flow paths, but the river flows in a continous single thread narrow channel above and below these sites. This change in morphologic character is used to distinguish reach breaks between major tributaries. The reaches are fairly steep, ranging between 1.57% above Foley Creek to 3.06% below the lake. Moss on the large instream boulders and along channel banks serves as evidence that the channel is mostly stable in these upper reaches (see Figure 12). There are small localized deposits of cobble sized material, but there is very little storage of sediment overall. Flashy  -67-  Figure 12: Photo of left channel bank, reach 15. Note moss on largest boulders, indicating stability. Several small 'pockets' of cobble sized material can be seen intersperced among boulders. Channel is -20 m wide at this location.  flow events capable of transporting the coarser material are rare due to the moderating influence of Chilliwack Lake. Some small bars and island formations are observed in Reaches 11 to 13 where the channel widens locally and slope is decreased. The bar and island material is probably derived from fan deposits of tributary streams. Several large islands are found in the channel between Foley Creek and Reach 13, but the mature age of the forest cover on them shows they are stable.  Foley Creek to Slesse Creek  Morphology between Foley Creek and Slesse Creek is similar to upstream reaches. The channel is irregular, but becomes increasingly sinuous downstream where an alluvial valley flat emerges about  -68-  2 km below Chipmunk Creek (in Reach 8). A large meander bend is developed downstream, but has changed little over the period of record. Upstream of this location, the river is mainly confined by the valley wall to the north and a high, distinct terrace to the south which marks the distal edge of an extensive sandur that originates at Chilliwack Lake. This terrace is located upstream of a bedrock canyon above the Slesse Creek confluence, where the river has cut into its former valley floor (McLean, 1980). Channel gradient is similar to that upstream, except for the steep drop below the bedrock canyon which acts as a stable knickpoint. An interesting observation is that channel width is nearly constant from R10/R11 downstream to Slesse Creek, a testament to channel confinement and stability. There are very few gravel bars and only a few small islands covered with mature forest (see Figure 13).  Figure 13: Mid-channel island in Reach 8. Bank shows evidence of recent erosion, probably from 1989 and 1990 floods. Islands (and bars) in upper reaches are not affected by more moderate floods.  -69-  The lack of instream gravel storage suggests these reaches act primarily as a conveyance zone for small amounts of bed material delivered from upstream. Bed material is smaller than in upstream reaches (Hay & Co., 1992) but McLean (1980) notes that much of the bed would remain stable even during high magnitude (e.g. 1 in 10 year) flows. Airphotos from 1988 and 1991 do show increased bed material storage (lateral bars) and channel widening upstream of the Slesse Creek confluence as a result of the high discharge events of 1989 and 1990, though there is no evidence of this effect during earlier periods. The large alluvial fan of Slesse Creek has imposed a localized constraint on upstream gradient, resulting in this deposition.  Slesse Creek to Tamihi Creek Downstream of Slesse Creek the morphology of Chilliwack River changes considerably. Slesse Creek is the largest tributary, contributing about 15% of total basin flow and increasing downstream flow by about 25%. The channel below this point widens considerably and develops a more continuous alluvial fioodplain. Coarse gravel and cobble deposits are sometimes overlain with overbank sand; these finer deposits are rarely found upstream. The channel has developed large meanders in this section, most noticable in Reach 4 between Allison Pool and Slesse Park downstream. The location of these meanders is coincident with the location of the glaciolacustrine deposits described in Section 2.0; these sites are easily undercut by high flows, allowing extensive bank erosion and slides to occur (see Figure 14). Downstream migration of the meanders is limited because of partial confinement by the valley wall in Reach 5 and relict terraces in Reach 4. However, old channel scars provide evidence that some progression has occurred in the recent past. Between these meander belts, the river exhibits a wandering habit. Both islands and lateral and mid-channel bars are common throughout the entire reach with several short wandering stretches nearly braided in character (post-1989 only). This  -70-  Figure 14: Erosion of glaciolacustrine depoits in Reach 4, near Tamihi confluence  transition from a 'wandering' to 'braided' gravel-bed channel reflects a considerable increase in upstream sediment supply from Slesse Creek. Average channel gradient decreases from 1.32% below Slesse Creek to 0.74% above Tamihi Creek, and accordingly, mean sediment size also declines. McLean (1980) found that the D  50  of bed material decreased from 140 mm near Slesse to 70 mm  above Tamihi and that the two year flood would be capable of mobilizing all but the largest stones.  Tamihi Creek to Vedder Crossing From several hundred metres upstream of Tamihi Creek to the lower end of Reach 3, another distinct morphology can be observed. The river narrows and flows in a steep, fast, single thread channel. Although the floodplain is confined within steep kame terraces, the channel does not impinge directly  -71-  on them. The transition is due to the stabilizing effect of large boulder deposits (1 to 3-m diameter) along the bed and lower banks which limit bank erosion, even during very high flows (see Figure 15).  Figure 15: Large boulder deposits and stable banks near Tamihi Creek  As there is no obvious source of these boulders from landslides and their large physical size precludes fluvial transport mechanisms, they are likely of glacial origin (possibly an end moraine). These boulders serve to armour the bed, preventing downcutting, and locally steepen the slope (which is as high as 2%; Hay & Co., 1992) so that bed material is transferred through the reach, rather than being deposited within it.  Below Reach 3, the channel again begins to wander and develops a braided pattern with numerous islands. During extended periods of bed inactivity, the river also exhibits characteristics of split and  -72-  anastomosed channels as bars become vegetated. The floodplain widens and old side channels indicate that the channel actively migrates throughout most of this zone, often to the valley walls. Below Liumchen Creek, the floodplain expands to over 2 km in width before narrowing at the bedrock constrained Vedder Crossing. Lateral migration is currently limited in this wide section by a small terrace. Sediment calibre declines as the gradient decreases to 0.62% in Reach 1. McLean (1980) estimated the D at 40 mm and large deposits of sand sized materials can be found along the 50  channel (Hay & Co., 1992). The floodplain and channel islands also have overbank deposits of sandand silt-sized material up to 0.5-m thick in places. Low gradients combined with extensive sediment inputs from bank erosion and upstream sources lead to extensive bed material storage. During extended periods of aggradation in the lower reaches, channel avulsions are common and shifts up to 700 metres have been recorded. As the channel floodplain is forested with firs and cottonwoods, large inputs of organic debris occur regularly and numerous logjams are found within the active channel. These logjams can block side and backchannels and may aid in diverting flows.  It is important to note that bank erosion and flooding hazards are highest in these reaches because of property and irifrastructure development and has resulted in extensive bank protection works along the north edge of the present active channel zone (Figure 16). However, similar protection works have not been put in place along south banks, so high rates of bank erosion along these or other unprotected banks are likely during the next big flood(s).  -73-  Figure 16: Large gravel berm (dyke) along north side of channel Reach 2 designed to limit erosion near Chilliwack River Provincial Park. This area was subjected to significant lateral erosion during last 2 major flood events.  4.2 Patterns of Channel Stability Channel displacements, adjustments of size and changes in form and pattern are all natural responses of rivers to changes in water and sediment loads. Hickin (1983) notes that the study of channel changes is really the study of equilibrium behaviour and the nature of excursions from equilibrium conditions. Changes in environmental conditions (both natural and anthropogenic) are responsible for these excursions. In this thesis, channel changes are examined in the context of the 'engineering' time scale, which ranges from several years to decades (cf. Hickin, 1983). It is important to note that equilibrium at this time scale should be considered a transient property of rivers on longer time scales (e.g. decades to millenia) unless inputs of flow and sediment remain constant.  Hydraulic geometry and regime theory can be used to relate form and process in natural channels  -74-  over a decades-long time scale. Broadly stated, natural channels will develop a stable form in response to frequently occuring flows of water and sediment through them. A channel is assumed to be "in regime" when it has adjusted its slope and section (shape) to an equilibrium condition (Henderson, 1966). This implies that channel hydraulic geometry is in equilibrium with the incoming flow of sediment and water. The rate of bed material transport equals the rate of mcorning sediment supply so the bed is stable even though it may be moving (Haan et al., 1994). Although originally developed for stable canals and rivers with a limited range of discharge, regime equations can be used to compare measured morphologic parameters with predicted parameters for stable size and form. If the magnitude of the dominant discharge in Chilliwack River has increased, then changes in the hydraulic geometry of unstable reaches should be observed. If these changes are corifirmed by comparing measured channel width to predictions from regime equations, then it can be concluded that the river has experienced a significant change. The river would then be reset a new regime where the channel is in equilibrium with higher magnitude flows or sediment transport rates.  In the above view, a channel has a single, stable regime which it attempts to occupy. If the channel is not in regime with long-term flows of sediment and water, it will modify its form until a stable pattern is achieved. This process is known as channel recovery or post-flood adjustment. An opposing theory suggests that channel morphology is established by higher magnitude, relatively infrequent throughputs of water and sediment. In this view, channels may recover slowly or not at all; lag effects of the large flows may persist if introduced sediment is sufficiently coarse and smaller flows are unable to remove it. Such channels may remain in disequilibrium, or display non-equilibrium forms (Gupta, 1983). The distinction between these two channel forming theories is important for examiriing the relative importance of large floods and different land-use practices on shaping channel  -75-  morphology. By examining the morphologic response of Chilliwack River to individual flow events and longer term flow trends, a greater understanding of the processes responsible for governing observed changes can be realized.  Changes in channel width and bank erosion are discussed in the following sections to examine how these rates have changed over time. The normal regime of Chilliwack River will be estimated to determine whether large floods such as that which occurred in 1989 were catastrophic events (floods which transform the river to a disequilibrium regime). Gupta (1983) notes that these high magnitude floods can leave a stable imprint on channel form if there is a considerable supply of large organic debris (LOD), large floods are periodic (eg. 10-20 years) and the ratio of flood peak to mean annual discharge is high. The relict terrace in reach 1 may be an example of a stable imprint, or 'superflood' form (cf. Gupta, 1983), though this is difficult to confirm. It is most likely that lag effects of large floods persist in upper reaches only - the historic photo record shows very little change in morphology prior to 1991. Along these reaches, the bed and bank sediments are too coarse to be moved during frequent, smaller floods. By contrast, the photo records show continuous channel modification along lower reaches. In these reaches, the channel is competent to transport bed material during moderate flows (e.g. Q>250 m /s) that occur up to 3 times per year on average. As a result, 3  lower reaches can probably adjust their morphology continuously until equilibrium is reached. Variations in sediment storage and transfer rates in relation to changes in hydrology and land-use patterns are presented in Section 5.0.  -76-  4.3 Lateral Stability 4.3.1 Water Surface Width For each date of photography, water surface and active channel width are calculated for each study reach using the GIS. Water surface width is determined by dividing the area of water surface in each reach by the thalweg length (see Figure 17a). If flows for each date are similar, observed changes in width can be attributed to changes in sediment storage (i.e. aggradation / degradation) or flow velocity. For example, if surface width has decreased between two periods but flows are equal, then the flow must be deeper and faster and could result in local coarsening or armouring of the bed surface. There are no significant changes observed between 1952 and 1973. Surface width has increased for most reaches by 1983, but that is easily explained by the much higher flow on that date of photography compared to other years. The most significant changes appear in the 1991 data where water surface width has increased since 1983 in reaches 5,6,8,9 and 11 despite lower flows on the latter date of photography. This is likely due to aggradation of bed material resulting in widening and shallowing of the bed; evidence for aggradation is presented in Section 5.2. Changes in water surface width can also be used to provide an estimated correction in exposed bar area due to changes in discharge. These data are presented in Section 5.1.  4.3.2 BankfuII Width Channel width should remain roughly constant over time if the magnitude of the dominant channel forming discharge also remains constant, but may experience a permanent change if the river experiences a catastrophic flow. Bankfull width is estimated as the physical area occupied by the channel (defined by limits of terrestrial vegetation) excluding wooded islands, divided by the thalweg length (see Figure 17b). This defines the width of the mobile channel bed (after Karlinger et al.,  -77-  Figure 17a: Variations in water surface width, 1952 to 1991. Flows for each date of photography are given (at Vedder Crossing gauge).  Figure 17b: Variations in bankfull width, 1952 to 1991  1983) so is often referred to as active channel width. The thalweg was measured independently for each year of mapped data to account for changes in length caused by channel migration. Increases in width reflect active bank erosion or removal of islands. Decreases reflect fioodplain reconstruction or channel recovery, the usual evidence of which is revegetation of gravel bars. The bankfull width is distinguished from channel zone width (which includes wooded islands) as this latter figure can be misleading. If the river abandons a small channel flowing around an island, it appears as either fioodplain reconstruction or erosion of the island, although there has been little actual sediment transfered. Similarly, if the river changes course (i.e. there is an avulsion) and occupies a former backchannel, then a new island appears to have been formed, inflating average width values.  For all years, bankfull width increases significantly downstream of the confluence with Slesse Creek (break between reaches 6/7) where the river is mainly flowing through a well developed fioodplain of alluvial deposits. Width is fairly constant upstream of this point, reflecting channel confinement by valley walls (bedrock) and terraces formed as the river dissected the former valley floor (McLean, 1980). As a result of this confinement, channel width has not increased over the period of study. By comparison, widening of the active channel appears significant since 1983 in Reaches 1-2 and 4-7. Reach 3 downstream of Tamihi Creek is fairly stable due to the paving effect of large boulder deposits along the bed and lower banks and confinement within steep kame terraces.  Over the period of record, a definite trend towards widened reaches can be observed from Vedder Crossing to reach 7 and in reach 11, especially by 1991. Bankfull width had increased 2.0X in reach 2 and 2. IX in reach 6 when compared to the average width for all other years of record combined. These large increases in 1989 and 1990 are reflected by extensive lateral migration and bank erosion  -79-  in lower reaches. By contrast, width had actually decreased by 19%, 14% and 8% in reaches 12 to 14 respectively. Reasons for the decreases are likely related to growth of alluvial fans where steep tributaries enter the mainstem channel or infilling of smaller backchannels where the flow is insufficient to transport the coarser materials introduced during major floods. It is also possible that these differences are representative of interpretation or digitizing errors. The 1991 photography is at 1:50,000 scale compared to 1:15,000 to 1:30,000 for other years. A placement error of only 0.1mm translates into 5-m actual lateral displacement and could account for the percentage decreases observed. Given the small (narrow) size of these upper reaches and problems associated with digitizng due to shadows, the placement error described above is reasonable and can not be discounted.  The trends described above are clearly shown in Figure 18. From 1952 to 1966, there is little change in reach width, indicating a stable channel regime over this time scale. By 1973, there is a small decrease in active channel width for several reaches; this indicates channel recovery, characterized by the vegetation of gravel bars and lateralfloodplainaccretion (reconstruction). This is particularly evident in reach 1, the width of which had been steadily decreasing since 1952. By 1983, many reaches show a definite width increase and by 1991, the channel appears to have further widened in response to the floods of 1989 and 1990. These patterns suggest that very large flow events may be capable of causing 'catastrophic' changes, representing the adjustment of morphology to a new channel regime over the short term. To test this hypothesis, active channel width for all periods is compared to stable values as predictedfromregime theory. As regime equations are based on purely empirical relations, regionally specific examples should be used. For channel width, the common relation is w = a Q where b is a scale relation for different river or regime types. h  -80-  Figure 18: Comparison of bankfull width changes between dates of photography  -81-  Desloges and Church (1992) refer to equations presented by Bray (1982) for gravel-bed rivers in western Canada, in which w = 4.75  Q  053  is based on the 2-year flood on a mean daily basis. For  reach 1, it may be more appropriate to use w = 12.8 Q  0 4 5  from Ashmore (1991) which is based on  model observations of braided channels (Desloges and Church, 1992). For each equation, Q is 2  estimated as 330 mVs for 1952 to 1974 from the partial duration series. Regression of discharge with basin area for gauged sites (e.g. Eqn. [2.2]) gives a straight line relation. Therefore, flow for an individual reach was estimated as a scaled parameter of the 330 mVs flow at Vedder Crossing based on proportional contributing area of the reach. Predicted channel widths based upon these relations are given in Figure 19 a,b.  It is evidentfromthese data that between 1952 and 1974, Ashmore's relation represents the measured active channel width quite well in reach 1, though Bray's relation tends to overestimate width with the exception of reach 2. This may be because Bray's data were taken from wandering gravel-bed rivers; reaches 3 and 7-15 are largely confined by non-alluvial materials which limits channel widening and migration (i.e. bed material may be less mobile than in Bray's channels). As well, this period was characterized by below average streamflows and few large floods capable of eroding the banks. For reaches 4-6, the channel may actually be underfit due a decline in discharge, bedload transport, or both. It is known that the period 1911-1930 was characterized by above average flows, so stable channel geometry may have been in equilibrium with larger flows than observed during the period 1952-1974.  For further comparison, regime equations from Simons and Albertson (1960) are also presented in Figure 19a. The advantage of the Simons and Albertson equations compared to similar relations  -82-  Figure 19a: Measured bankfull width compared to predicted width from regime theory equations, 1952 - 1973. Q (from P.D.S.) is 330 rrrVs. 2  Figure 19b: Measured bankfull width compared to predicted width from regime theory equations, 1973 - 1991. Q (from P.D.S.) is 440 m /s. 3  2  from other authors is that the median grain size (D ) distribution is not required as an input. These 50  empirical equations are based on datafromcanals and rivers in India, Pakistan and midwestern North America so may apply to those sites only. Although designed for channels with steady discharge (not for natural channels), variations from these predicted values over time indicate non-equilibrium conditions. These equations (from Henderson, 1966) are given below:  w = 0.9P = K i Q  0  5  and [4.1]  W = 1.087w + 2.0 Where: w is average width in feet; P is the wetted perimeter in feet; Q is discharge in cfs and W is top width in feet. The value Kj is a constant depending on channel type. It is used to distinguish between variations in boundary materials and sediment load. In Figures 22a-b, a value of K = 1.75 t  (type 4, coarse noncohesive material) is chosen and compared with predicted width from Bray (1982) to provide a range of expected stable widths. In general, this equation seems to fit the observed data quite well with most reaches falling at, or between the expected range. These results suggest that the Chilliwack River is 'in-regime' over the period 1952 to 1973.  From 1973 to 1983, the channel appears fairly stable. Increases in width for reaches 1, 2 and 4 may be significant based on stable predicted values, but the magnitude of the widening is small. There were two floods exceeding 500 m /s during this period (1975 and 1980) but the impact of these on 3  shaping channel geometry seems transient, especially compared to the changes resulting from the larger floods of 1989 and 1990. Theseflows(in 1975 and 1980) are large enough to strip vegetation from gravel bars which, by definition, increases bankfull width and could explain observed changes. By 1991, there is extensive widening of the active channel zone along unconfined reaches (1 to 7 except for 3, and reach 11). A large localized input of sediment from Nesakwatch Creek during 1989  -84-  or 1990 may be responsible for this change.  Over the relatively short (40 year) time period examined in this discussion, the findings seem to confirm the idea of a regime shift in lower reaches of Chilliwack River. However, if longer time periods are considered, then floods such as occurred in 1989 and 1990 must be considered as relatively 'normal' as they have return periods of only 34 and 42 years respectively (Hay & Co., 1992). Equally large floods are also known to have occurred in 1917 and the latter 1800s for example. The similarity in measured bankfull width for reach 1 (1991) compared to the regime value predicted by Ashmore's relation serves as further evidence that these large floods are not catastrophic. Although subsequent smaller floods are likely responsible for reversing the effects of the largest floods, pre-fiood conditions such as observed from 1952 to 1974 may actually be in disequilibrium with dominant channel fonning flows. This is because the channel 'recovers' in low flow periods as more moderate flows (e.g. 250< Q <500 m /s) are competent to transport gravel and there is 3  encroachment of vegetation on gravel bars. In the absense of large floods comparable to those experienced in 1989 and 1990, lower reaches of the channel should 'recover' vWthin a decade.  4.3.3 Bank Erosion Although it is fairly obvious that bank erosion must occur in reaches which are widening, bank erosion is a natural process in migrating or wandering rivers. A particular reach or channel bend may be actively eroding its banks, thus propagating migration, but the average active channel width may not change unless there is a corresponding change in upstream sediment supply or discharge. Therefore, it is relevant to examine bank erosion rates separately from changes in channel width. Bank erosion and stability have an important impact on the shape and character of alluvial channel  -85-  reaches. Erosion of bank material is thought to be one of the major sources of bed material in Chilliwack River and thus has morphologic significance. The magnitude and location of bank erosion along Chilliwack River is of particular concern, however, because of property and infrastructure developments within the active floodplain. As such, it is perceived as the greatest single hazard to residents living along the lower channel reaches (Chilliwack Valley Ratepayers, 1993, unpub.). Large instream logjams are especially blamed for erosion of property along the channel, but this is difficult to quantify. The main problem appears to be that many homes are situated well within the erosion limits of actively migrating reaches (as indicated by old channel scars). Most residential expansion in the valley took place in the 1960s during a period of relative channel stability when many homes were located close the the river.  The dominant bank erosion processes which occur along the Chilliwack River include direct shearing of the banks at high flows (Hooke, 1980) or fluvial undercutting of unconsolidated lower banks and subsequent slumping of upper banks (Thorne and Lewin, 1979). There is direct field evidence of both of these processes. Bank erosion of the glaciolacustrine deposits may additionally be influenced by groundwater seepage. During periods of heavy rainfall, these materials may slump and slide into the main channel where they are easily eroded and carried downstream. These sites are then vulnerable to further erosion from slumping or fluvial undercutting. In upper reaches, bank erosion may also be controlled byfreeze-thawweathering cycles (Lawler, 1986) though there are no seasonal data on erosion rates to confirm this.  The magnitude and location of bank erosion was determined using overlay maps calculated by the GIS. This is the preferred technique for determining bank erosion rates and can be used for periods  -86-  as long as 150 years where suitable maps exist (Lawler, 1986). It provides greater temporal and spatial coverage than field techniques (eg. erosion pins) and is more accurate than dating methods (eg. C , dendrochronology). Bank erosion was defined as polygons that were part of the fioodplain 14  on the early photo date (excluded from active channel zone) but were covered by water or gravel bars on the latter date of photography (included in active channel zone). This definition explicitly excludes channel avulsions (large changes in channel position) which inflate apparent erosion rates. The apparent transitionfromfioodplain to water might be partly dependent on flow conditions on the two mapping dates, but banks are generally too steep and well-defined to be mistakingly interpreted as eroded due to an increase in water stage. As it is not known if the transition from fioodplain to bars is the result of bank collapse or the staged process of erosion, bed material transport, and new deposition of material from upstream sources, these two processes are grouped together for subsequent discussion and calculations.  The magnitude and location of bank erosion along the lower six reaches of the Chilliwack is illustrated by a series of colour maps (Figures 20 to 23). Reach 3 is not shown as bank erosion rates are very low due to the stabilizing effect of large boulders on the bed and lower banks. Similarly, part of reach 5 is excluded where the river is confined by valley walls. Fioodplain reconstruction may occur between non-successive periods of erosion (ie. between 1952-66 and 1973-83). Where this occurs, erosion during the earlier period may be masked by erosion in the latter period because of the order in which the colours are represented and printed. However, it is known that this occurs infrequently, so erosional sites shown on the maps are approximately unique. Also, the record length for each period is not the same, so longer record lengths may show greater erosion, other variables being constant. The position of the bankline for 1952 and 1991 shows the maximum extent of lateral  -87-  Figure 21: Map of bank erosion, Liumchen Creek to Ryder Creek  -89-  Figure 22: Map of bank erosion, Tamihi Creek to Slesse Park  -90-  migration, bank erosion and fioodplain reconstruction over the period of record. Scale is indicated on each map by the 2 km spacing of the U T M grid. The scale varies slightly between maps, but 1 cm represents roughly 180-200 m of erosion over the period shown.  A cursory examination of the maps suggests that bank erosion rates have increased over the period of record, and were lower, in general, during 1966-1973. These observations are similar to those determined for active channel width, although this association is not invariable. The maximum erosion appears in reach 1, south of Baker Trails for the 1983-91 period. Erosion depicted at this site may reflect re-activation of an old fioodplain course (i.e. abandoned backchannel) in response to the large floods. The channel at this point is known to have shifted south 650 m within a few hours following the flood of 1990 (Hay & Co., 1992). Some of the eroded material may have accreted on the fioodplain downstream of this location.  Below Tamihi Creek, lateral migration of the channel by bank erosion is extensive (see Figure 21). However, meander belts are not developed here because the channel impinges on the valley wall in several places. Upstream of Tamihi Creek (reach 4, Figure 22) the river exhibits characteristics of a regular meandering channel. The channel banks are eroded as these meanders progress downstream, but the development of the meanders is limited below Slesse Park by a terrace on the right bank and the valley wall along the left bank. The meanders impinge directly on the glaciolacustrine exposures at most sites and are responsible for their periodic failure as the toe of the slopes is undercut. The channel appears to have straightened and widened by 1991. Schumm (1977) indicates that this is the expected channel response to an increase in both bedload and discharge. Upstream of Slesse Park, a large meander loop is developed, but its movement is limited by a high terrace on the left bank. In  -92-  reach 6 and the upstream section of reach 5 (Figure 23) the river again exhibits a wandering habit, similar to reach 2. Bank erosion and channel migration are relatively low for all periods except 198391. Meander development and progression is also limited by the valley wall in these reaches.  The average rate of bank recession for each reach is given in Figure 24a. These rates are derived in the GIS by summing all erosional areas for a period and dividing by the reach length. This technique allows unbiased comparisons to be made between periods and is simpler to calculate than measuring average bank recession at different sites. This is because bank erosion sites as depicted on the maps are composed of several individual polygons with irregular shapes whose appropriate length dimension is difficult to measure. Rates are normalized for each period by further dividing by the length of the measurement period in years. The maximum rate of bank erosion is also calculated and given in Figure 24b. These rates are determined by measuring the maximum bank retreat perpendicular to the thalweg within each reach; these data have also been normalized by dividing by the record length.  Average bank erosion rates for several reaches and periods is less than 1 m/yr and primarily reflects measurement error. The accuracy of locating the bankline in upper reaches is affected by the orientation of the aerial photos (±1.5 m), the precision of the stereoplotter (±0.5 m), the resolution of the photos when digitizing in shadows (±3 m), and uncertainty in bank position along islands and at confluences (unknown error). In lower reaches where the banks are more clearly visible, the latter two error terms are reduced. For an individual year, the error in positioning the bankline is therefore 3.4 m. The comparative resolution for any period can be estimated as the combined positioning error (4.8 m maximum) divided by the record length in years. Although a precise error limit is not easily  -93-  deterrriined, 0.7 m/yr (e.g. 4.8 m over 7 years) provides a good maximum limit, below which, there may be no actual bank erosion over the period of record. This error is shown as a dark line on Figure 24a.  Although streamflow discharge is greatest in the period 1983-91, average and maximum bank erosion in reaches 11 to 15 is never greatest during this same period. This indicates that failure mechanisms unrelated to flow and bank strength may dominate in these sections; frost and weathering disturbance of the banks in these reaches are the most likely candidates. Lawler (1986) made similar observations for upstream meander bends in Wales. The maximum rates shown in these reaches are probably related to local short-term channel disturbances such as tributary fans or logjams divertingflows.For all reaches downstream of reach 11, average and maximum erosion rates are all greatest for the period 1983-91 and are lowest for 1966-73. These findings are consistent with those made for channel width and flood patterns. Hooke (1980) cautiously noted that most bank erosion appeared to coincide with flow events of moderate magnitude and peaked at roughly bankfull discharge. By contrast, these results indicate that bank erosion along Chilliwack River is associated with the passage of major floods and most erosion occurs over a short period of time.  The range of erosion rates shown in Figure 24a,b for the lower six reaches varies considerably over time and space. Prior to 1983, reaches 1 and 2 were demonstrably the most active in terms of lateral migration with average erosion rates between 1 and 3 m/yr while erosion in the other reaches is comparatively minor. The maximum rates given prior to 1983 range between 2 and 14 m/yr. By 1991, average rates varied between 4 and 8.5 m/yr and maximum rates similarly increased to between 14 and 17 m/yr. Reaches 4 to 7 were also significantly affected. For comparison, published rates given  -94-  Figure 24a: Reach averaged bank recession rates  20 •  1952-66 • 1966-73 • 1973-83 • 1983-91  15  10  15  14  13  12  11  10  9  8  7  6  Reach  Figure 24b: Reach maximum bank recession rates  5  4  i 3  2  1  for mean and maximum erosion rates summarized in Hooke (1980) in Devon indicate rates between 1 and 10 m/yr are normal for rivers with a drainage area similar to the Chilliwack River. A study of the similar sized Green River in Washington state (by Dunne and Dietrich, 1978 given in Hay & Co., 1992) gives lateral migration rates up to 9 m/yr. These results indicate that values calculated in this study are relatively high, but nonetheless consistent with other studies.  4.4 Prediction of Bank Erosion and Channel Migration The erosion of channel banks and migration of the active channel represent significant hazards to property and infrastructure development along lower reaches of Chilliwack River. Mitigation of this hazard can be achieved by establishing erosion limits beyond which the probability of loss or damage is small. An understanding of channel dynamics is necessary to determine these limits and maximize the effectiveness of stabilization projects or protective structures. Empirical models have been successfully used to relate the average rate of bank erosion with environmental and geomorphologic variables. However, these models generally fail to predict the maximum erosion rate that can occur at a particular site because they do not account for bend migration. An alternative approach is presented in this study based on a probabilistic examination of channel behaviour. The probabilistic approach relies on historical evidence of channel occupation to predict potential instability.  Graf (1984) used historical airphotos and maps to determine the historical variations in position of an Arizona Creek over 107 years. He detenriined that the probability an individual cell would be inundated by the channel (eroded) during some time period was related to the magnitude and frequency of large floods and position of the cell with respect to the active channel. The technique was evaluated by comparing predicted erosional cells (between 2 dates) to known erosional sites over  -96-  the same period. A 100 m grid (cell) map depicting the overall migratory extent of the river over the 107 year study period was also produced. Although sites predicted as having a high probability of being inundated were not compared with this map, it was evident that there was strong association between erosion probability and historic channel occupancy. Downward et al. (1994) overlaid historic maps of the River Dee, Wales in a GIS to depict downstream changes in channel stability over a 115 year period. Their map was based on the percentage of time that the channel occupied different 1 km stretches of the fioodplain during the 115 year record. The historic GIS mapping completed for Chilliwack River could be used to produce a similar channel occupancy map.  The channel migration approach presented in this study is based on the location of bankline positions over the period of record (1952-1991). The bankline defines the channel zone, the extent of which includes the active channel zone and wooded islands (i.e. defines a region which contains no human development). For each of the five years of mapping, the polygons defining the entire channel were overlaid to produce a composite map showing the historic extent of the channel zone over the 39 year photo record. This composite map for all reaches of Cliilliwack River is comprised of 7350 unique polygons, in which each polygon has between one and five associated year values, indicating channel presence. Polygons with all 5 year values define channel locations that have not changed over the period of record. The number of unique combinations (no replacement) of years for each polygon is determined by the following function:  where n is the total number of mapped years in the historic record (e.g. 5 in this example) and r is the number of events. For example, the number of combinations involving 3 years (r=3) is 10. Varying  -97-  r from 1 to 5 and summing all values of C gives a total of 31 unique combinations. Additional years r  of mapped data will increase the resolution, but also the complexity of this analysis.  For each unique combination, a total year count was determined based upon the number of years that the river occupied each polygon. The maximum possible value is 40, or the length of the data record since 1951. For polygons which were occupied only once (e.g. 1983) the channel was assumed to have occupied that position since the previous large flood (e.g. 1980) until the following large flood (e.g. 1984) for a total of 4 years. Large floods refer to events that were likely to cause significant bank erosion, hence channel migration . For any two unique dates (e.g. 1966 and 1991) it was 2  assumed that the channel would have occupied those locations for 13 years (1955 to 1968) plus 1 year (1990 to 1991) for a total of 14 years. If the channel occupied a given location for any two consecutive periods, it was assumed that the location remained stable between the two dates, though this may not be correct if there were intervening large floods, such as between 1983 and 1991. In this example, the channel was likely to have remained in the same position from 1980 to 1991, or 11 years. All total year counts were assigned using the Info database by selecting the 31 possible combinations individually and assigning each set of selected polygons total year values as described above.  An additional numeric field was added to the database and used to calculate probabilities. Probabilities were estimated as the percentage of total years (plus 1 year) that the channel occupied a known position. If the channel occupied a position for all 40 years, the probability was calculated  2. Damaging floods are thought to be larger that bankfull floods, with a probable return period of about 5 years. For this analysis, large floods between 1951 and 1975 were estimated as being greater than 400 mVs. From 1975 to 1991, large floods were estimated as being greater than 500 m /s. There are 8 floods over the 40 year record that exceed this criteria. 3  -98-  as 40/41 or 0.976. The value n+1 years was used to avoid assigning a probability of 100% to any polygon. The fioodplain boundary provides a maximum limit of possible erosion in this case, and more accurately represents 100% probability (at least over the long term). For the convenience of display, all probabilities were classified into 7 ranges; an example map is shown in Figure 25. The map is also summarized in Table 4, which gives the percentage of the total channel area occupied by each probability range.  Table 4: % of total channel area occupied by each probability range Probability range  All reaches  Reach 1  0.0-0.15  35.1%  35.9%  0.15 -0.30  8.8  5.5  0.30 - 0.45  4  5.8  0.45 - 0.60  0.5  0.7  0.60 - 0.75  5  4.8  0.75 -0. 90  10.9  17.2  >0.90  35.8  30.2  The variability of the channel over the photo record can be used to predict sites of future instability. In a planning context, all 'coloured' regions would be avoided as potential sites of development because they are known to have been occupied by the channel at some point within the past 40 years and could be reactivated by the next large flood (if not currently active). This type of mapping can be used to complement existing planning tools, such as the 200-year flood mapping. It would be an interesting exercise to compare the two types of hazard maps on Chilliwack River, or other rivers where there has been damage to property and infrastructure developments from bank erosion.  As expected, reach 3 and most upper channel reaches show little variability over the period of record  -99-  Chilliwack River spatial variability map  Probability I  0.00 - 0.15 0.15 0.30 0.30 0.45 0.45 0.60 0.60 0.75 0.75 0.90 1.00 0.90 Fioodplain boundary  Reach 1  Figure 25: Spatial variability map of historic channel occupation  -100-  - i.e., the probability for channel migration and bank erosion is low. In uncorifined lower reaches, there is far greater spatial variability, but the highest probability range is continuous from Vedder Crossing to Chilliwack Lake. This means that the river never completely abandons its active channel, even in braided lower reaches subject to avulsions during large floods (refer to Table 4). However, there are several locations along the channel (in reaches 1,2,4 and 6) where the 'stable' band is very narrow. These narrow sites are consistently spaced from 1500 - 1800 metres apart. These locations may be analogous to end points of meander bends, between which the channel actively migrates.  Several of the largest floods of record have occurred during the past 40 years. In particular, the large floods of 1989 and 1990 account for much of the variability shown in Figure 25 (i.e. the low probability regions). Despite the magnitude of these flows, however, the maximum extent of the fioodplain is considerably wider than the total channel occupied zone in many locations. These areas may represent a hazard to property and infrastructure developments during subsequent large floods, particularly near the margins of the present active channel. The map could be improved by extending the data record prior to 1952 or by digitizing additional channel features such as backchannels. Dating of fioodplain deposits or calculation of sediment residence times would also be helpful in this context.  -101-  Chapter 5: Sediment Transfer and Yield This section of the thesis examines patterns of bed material erosion, transport and deposition along Chilliwack River based on morphologic changes as measured from aerial photos for 1952, 1966, 1973, 1983 and 1991. For each year of photography, net changes in bank position, gravel bars and islands were accurately planimetered with the stereoplotter, then mapped in Arc/Info GIS. Maps for successive years are overlaid in the GIS to determine relative changes in the erosional and depositional areas of these features (see discussion in Section 3.4). Volumetric changes are estimated as the product of these measured areal differences and the depth of the mobile bed material layer. An adjustment is made for exposed gravel bars to account for variations in river discharge. Net storage changes of bed material over the period of record are discussed.  A sediment budget framework will be presented for relating these volumetric changes to sediment transport rates. Bed material inputs from Slesse Creek (the largest tributary) are estimated and incorporated into the analysis. This allows the net volumetric changes in bed material storage along Chilliwack River to be assessed over a 40 year period. Observed variations in sediment transport rates over time can be used to make inferences about external controls on channel behaviour, but do not allow the impacts of shorter term (flood-length) processes to be determined. Channel response tofrequentlow-magnitude and infrequent high-magnitude flows is examined separately for reach 1 and compared to observed longer term trends.  5.1 Channel Change Assessment 5.1.1 Estimation of bed material depth Quantification of the sediment budget and bed material transport rates requires analysis of volumetric  -102-  (3-D) changes in net bed material storage. Volumetric changes can be estimated as the product of erosional or depositional areas and the depth of the mobile bed material layer (see discussion in Section 1.3.2). Of the three available methods for estimating bed material depth, the field survey and cross-section techniques were selected. Along upper reaches, scour depth was estimated from direct field measurements only; cross-sections are not available and the position of the banks can not be located precisely from the scales of photography available for these reaches (no photo scales larger than 1:15,000). The mobile sediment depth was estimated as the difference between the top of bar deposits and the deep part of the adjacent thalweg. Channel banks are effectively immobile in these reaches and were therefore not included in the total height estimates. As gravel bars are infrequent in these reaches, several measurements were taken at some bars and the results averaged as a single survey site. Measurements were made as frequently as possible downstream, but difficulties of access meant that some reaches could not be surveyed at all due to high, fast flows and confinement within steep terraces or bedrock walls.  The field locations of the surveys were marked on a set of airphotos, digitized and imported into the GIS. The GIS was used to calculate the distance downstreamfromthe lake for each survey site; these data indicate a general trend of increasing deposit thickness downstream (see Figure 26a). A best fit regression line was calculated to estimate deposit thickness for reaches which could not be surveyed. The mid-point of each reach was chosen as a representative position along the reach and used in the regression equation to estimate gravel depth based on cumulative distance downstream from the lake. This method of estimation also helps to smooth the very irregular spacing of the original field surveys. These results give bed material depth estimates of 1.2 m in reach 15 below Chilliwack Lake to 2.0 m in reach 7 above the Slesse Creek confluence (see Figure 26b).  -103-  4  Measurement type . • Field A X-S o Fines 3.5  u  3 2.5 •  2  • i •  i  • •  A  A  *  A  • •  A A  • A A  i •  •  A  A A  •  1.5 1  i  •  •  i  E,  A  A  A  s« =«  01  QO  •  B  • A  •  o  O  0.5  0  o  0  '"^  o  o  O  0 10  20  30  40  50  Distance downstream (km)  Figure 26a: Downstream variation of bed material depth  15  Reaoh  Depth  Reaoh  Depth  15 14 13 12 11  1.17 1.27 1.38 1.51 1.63  10 9 8 7  1.72 1.79 1.88 2  14  13  12  11  10  9  8  Reach  Bars  Banks  6 5 4 3 2 1  2.53 1.97 2.4 2.71 2.65 2.75  2.6 (est.) 2.7 (est.) 2.86 3.0 (est.) 3.04 3.27  I 6  i 5  i 4  I 3  I 2  Reach  Figure 26b: Reach averaged gravel bar and bank height estimates  A combination of methods was used to estimate bar and bank material thickness along the lower six reaches. Separate measurements were made for bars and banks to reflect the two different processes of sediment transfer that occur along Chilliwack River. Field measurements of thalweg to bar top height were made at 19 individual sites. A distinction was also made between the top of the basal gravel layer and the overlying deposits of fine sediments where they were found (as described in Neill, 1987). It was convenient to separate these data in the field so that this depth could be subtracted from bank height estimates as determined from the cross sections. Cross sections for 1990 and 1991 (reaches 1 to 4) were overlaid in a graphing program. Orthophoto maps were then used to locate the cross-sections on a recent set of aerial photos; the maps did not have the lateral extent of the surveys drawn to scale so the photos were used to help distinguish channel features (e.g. fioodplain, extent of active channel zone) and interpret the overlay plots (e.g. bank erosion, gravel bar scour and fill). Several of the cross sections could not be interpreted because the length of the survey was different on the two dates, so they were removed from subsequent analysis. The depth of the mobile sediment layer was estimated as the difference between the lowest elevation point on each cross section survey and the top of the interpreted gravel bars or banks.  Locations of all cross sections and field measurements were marked on the photos and digitized to calculate distance downstream of Slesse Creek. Field measurements and cross-section estimates of bar depth as a function of distance from Chilliwack Lake are given in Figure 26a. This figure clearly shows a positive trend of increasing bar depth downstream, though there is considerable scatter. It appears that there is some bias in the data, with many cross-section depths greater than field surveyed depths. Due to the high, fast flow in the channel, the deepest part of the channel could easily have been 'missed' during the field surveys because it was not possible to wade across the entire channel.  -105-  As well, there may have been aggradation in these lower reaches since the cross-sections were surveyed (1991) shortly after the last major flood (1990). As the last flood was so large, it is also possible that the maximum depths interpreted from the cross-sections are not representative of the entire surveyed section or are atypical indicators of depth over the longer term. It is significant, however, that the cross-section location nearest to each field survey location (the most closely juxtaposed pairs) shows similar values. As it is not known which data are more reliable, both sets were pooled together and used in the following analysis.  Given the considerable scatter in depth data downstream of reach 7 (km 25), regression estimates were not used to calculate reach averaged deposit thickness, as was done for upstream reaches. As well, regression (if used) would mask 'real' reach-to-reach differences in bed material depth that may exist. Alternatively, each surveyed point (from the pooled set of field and cross-section depths) was assumed to represent a length of channel equal to the sum of distances upstream and downstream from each point to the mid-points between adjacent survey locations. These distances were multiplied by the measured bar depth at each survey point location. For each reach, the sum of these distance weighted depths was divided by reach length to give a reach-averaged depth. These data are shown in Figure 26b. A trend line illustrates the near linear increase in gravel bar depth downstream. Departures from the trend line in reaches 5 and 6 may be due to the absence of available cross sections, which makes depth estimates less reliable than in the lower four reaches. However, greater than average depth in reach 6 is reasonable given the additional contribution of flow and sediment from Slesse Creek.  A second, parallel trend line illustrates downstream variations in bank height, as estimated from the  -106-  overlaid cross-sections along reaches 1 to 4. Banks along these reaches consistently appear to be about one-half metre higher than gravel bars deposits. As banks could not be easily identified from the cross sections along reach 3 (i.e. there is no fioodplain) the anomalously large depth value shown probably reflects the height of the stable kame terraces along the reach. Bank heights for reaches 3, 5 and 6 were alternatively estimated from the trend line fitted to reaches 1, 2 and 4. The difference in depth between bar and bank deposits reflects different styles of instability along Chilliwack River. Along upper reaches, channel stability is dominated by the transfer of material stored in gravel bars. Along lower reaches, stability is dominated by the erosion of channel banks and subsequent downstream transfer of this material.  Method [3] (airphoto measurement of water surface to bank top, plus the water depth estimate - see Section 1.3.2) was tested along reach 1 to provide an additional set of estimates and check the validity of the technique. This estimate gave an average depth of 2.49 to 2.62 metres (depending on roughness coefficient used) excluding the overbank sands. The method underestimates the bank height value of 3.27 metres. This underestimation is likely due to the fact that the method provides an average of bed geometry rather than the maximum depth range, which is the desired value. As there were many available cross sections for reach 1, method 3 was not used.  Uncertainty associated with the bar and bank height estimates may introduce a significant error to storage change volumes and sediment transport estimates. In general, this uncertainty increases downstream because the total area of erosional and depositional zones is much larger than in upstream reaches. Along upper reaches, the depth uncertainty is included as an absolute error term which incorporates interpretation errors and photo measurement errors (refer to Section 5.1.3).  -107-  Along each of the six lower reaches, average bar and bank height estimates were made based on a distance weighted average of direct field survey measurements and surveyed cross section overlays. Although there is no direct error associated with the field measurements, it was not always possible to survey the deepest part of the channel which may introduce a negative bias in the maximum depth estimates. There is further (unknown) bias associated with the cross sections as the maximum range of the mobile bed thickness may not be interpolated correctly. Additional uncertainty that must be considered includes the assumption that these spaced height estimates are sufficient to characterize actual variations in bed material thickness downstream and that the mobile gravel layer can be characterized by a single value in each reach. This obviously biases erosional and depositional volumes near reach breaks as there is a distinct 'step' change, rather than a more probable gradual change. A more accurate representation of these differences would involve costly topographic and bathymetric surveys, making the technique impractical for most studies. If these data were available, however, the assumptions of reach averaged bed material and bank thickness, and problems with water level adjustments could be avoided (e.g. McLean, 1990).  Uncertainty associated with the depth estimates was determined by calculating the standard deviation for the pooled group offieldand cross section estimates of bar height. Each value was compared to the distance averaged depth for the corresponding reach within which it was located. The standard deviation for gravel bars was calculated as 0.49 metres, which gives an adjusted estimate of depth ± 0.25 m. It is assumed that this error be applied equally to bank height estimates. For each reach, a range of storage change estimates was calculated based on the rninimum and maximum depth estimate for bars and banks. The average deviation of these two terms gives the estimated error for that reach (see Appendix A).  -108-  5.1.2 Areal exposure corrections Prior to any discussion of bed material storage and transfer rates, superimposed images (overlay maps) must be corrected for variations in river discharge. Higher discharge will increase the river stage (hence width of the water surface) and correspondingly decrease the surface exposure of bars, especially for wider, shallower reaches (assuming channel geometry is stable). Between any two dates of mapped photography, stage differences will give the incorrect impression of bed material erosion (higher flows on latter date) or bed material deposition (lower flows on latter date). These changes must be considered and accounted for prior to the estimation of erosional and depositional volumes (determined as areal changes multiplied by bed material depth) before these data can be used in sediment budget calculations.  An exposure correction is made for gravel bars in each reach based on the relation between discharge and water surface width. Over the range of flows examined in this study, it is assumed that the exposed area of islands and fioodplain does not change and can therefore be excluded from the discussion. Figure 27 gives an example cross section from reach 1 which illustrates that, as water width changes, bar exposure changes by an equal, but opposite amount. As the width of the active channel does not change significantly with respect to discharge for lower flow events (less than bankfull) and active channel width is equal to the sum of bar plus water surface width, this assumption seems reasonable.  As noted by Church et al. (1987), variations in channel form and mutual adjustments of water depth and velocity to discharge may cause the correction gradient to change along the river. As a result,  -109-  Figure 27: Exposure of gravel bars in response to variations in discharge  separate exposure correction curves were developed for each reach. For upper reaches (7 to 15), no change in water surface width with respect to discharge was observed. Although there is some scatter in the relation for reaches 11 and 13, there is no positive trend. These results indicate that areal exposure corrections are not required in any of the upper reaches over the range of flows examined in this study. Along lower reaches (1 to 6), changes in flow have a greater impact on water surface width; correction curves for reaches 1 to 6 are given in Figure 28. These diagrams show reach-length, at-a-station hydraulic geometry relations; straight line relations indicate that channel hydraulic geometry has remained stable over the period of record. These relations are defined by the equation w = aQ where b<l .0. Equation coefficents for each reach are given in Table 5. b  -no-  •-  Reach 1  %  m  " —  i  Reach 2 Row (m3/B)  Row (m3/s)  Row (m3/s)  • •  —•  • —  i  '  a  •  m  Reach 6  Reach 5 Row (m3/s)  Row (m3/a)  Figure 28: Strearnflow vs. water surface width along lower reaches  -11  Table 5: Exposure correction coefficients for reaches 1 - 6 t  s.e. ± %  0.976  15.51  6.4  0.268  0.946  7.23  13.8  34  0.13  0.726  2.82  35.5  4  23.8  0.206  0.902  5.26.  19.0  5  30.8  0.118  0.55  1.92  52.2  6  27.5  0.181  0.518  1.79  55.7  Reach  a  b  r  1  15  0.379  2  23.2  3  2  Note: Slopes for reaches 5, and 6 are not significantly greater than 0 at t j 3=2.35; reach 3 is marginally significant. 0  Reaches 1, 2 have remained stable over the period of record, but there appear to be adjustments of these relations for other reaches. The coefficient 'b' is the slope of the regression line when the variables w and Q are linearized. These coefficients show that changes in width (hence bar exposure) are less sensitive to discharge in reaches 3, 5 and 6 than in reaches 1, 2 and 4.  For each reach, the areal exposure correction is detennined as the proportional rate of change in water surface width per unit change in discharge. From differentiation of the equation w = aQ , the b  change in width for a corresponding change in discharge is given by:  =  abQ  [5.1]  bl  60 In practice, the corresponding correction in water surface width between any two consecutive dates can be estimated as:  AW  =  a(e  6 2  -e  6 1  [5.2]  )  For an upward adjustment (higher flows on the latter of two dates), the value Aw is positive. The correction factor for bars is determined as V  bc  = Aw * L  r  * d, where L and d define the reach r  -112-  length and bed material depth, respectively. The corrected volume of erosion between consecutive dates is equal to the measured volume minus the apparent erosion volume of bars caused from a stage increase. Correspondingly, accretion of bar material would be corrected as measured deposition plus the apparent erosion volume caused by the stage increase. As it is assumed that bar and bank slopes are similar in all locations (or can be averaged to the same slope over the length of each reach), the water level correction is applied equally to apparent erosion/deposition volumes. For any period then, the net storage change of bar material can be estimated as:  ^ - , ^ - ^  If discharge is lower on the latter date, the value V  + [^1,  b c  [5.3]  is negative, and is therefore subtracted from the  volumetric difference between bar deposition and erosion. For net aggradation of bar material in any period (e.g. 1966-1973), the estimated volume is added to deposition of island and fioodplain sediments in the same reach. For net degradation of bar material, the estimated volume is added to the calculated eroded volume of island and fioodplain sediments. The difference between the total volume of deposition and the total volume of erosion gives the net storage change for each reach. These net changes are used as inputs for sediment budget calculations.  Ideally, the water level on each date of photography would be similar so that no correction was necessary. As the water level adjustment is only an approximation of average channel geometry, it may introduce significant errors to the net storage change estimates. There are two error terms that must be considered. The first (and primary) of these is the uncertainty in the magnitude of the correction factor used. This can be estimated by the standard error of the regression slope and intercept (refer to Table 5). For each reach, the slope coefficient 'b' is corrected as b±s.e. which gives  -113-  a maximum slope (when the intercept value is minimum) and a minimum slope (when the intercept value is maximum) that can be used to define a range for the water level adjustmen between any two consecutive dates. For example, in reach 1, slope is corrected as 0.379±0.0244 and the intercept value as 1 0  1176±041  . The mean difference between this minimum and maximum adjustment gives the  storage change error for the reach. Alternatively, the net storage change of bar material can be estimated from Eq. [5.3] using this range of water level adjustments to provide new estimates of net bed material storage change for each reach. Following the above example, the net storage change for reach 1 could range between -16,868 m and -27,113 m (original estimate was -21,847 m ). The 3  3  3  mean of the difference (5123 m ) gives the uncertainty in storage change estimates. 3  The water level adjustment should not be used when flows on consecutive dates are similar, there is no apparent change inflowwidth between dates or when the stage-discharge relation is weak. There is an additional error introduced when the decision to ignore or apply the correction for a given reach is not certain, however. The significance of an incorrect application of the water level adjustment is that an aggrading reach may appear to be degrading (or vice-versa), thereby affecting transport rates. When the flow on consecutive dates is similar (e.g. 1966 and 1973) and the relation between flow and water width is not strong (e.g. reach 5) it is reasonable to assume that no correction is necessary and that there is no error in that decision. Similarly, where flows are disparate (e.g. 1973 and 1983) and the relation is very strong (e.g. reaches 1, 2) then it is assumed [correctly] that the adjustment is required. The adjustment was ignored for reach 4, 1966-73 and applied to reach 3, 1973-83 and 1983-91, each of which may be an incorrect decision. The error can be estimated by the net storage change of gravel bars and bed material when the adjustment is first ignored, then applied. The mean of these net changes gives the uncertainty term for storage change estimates. A summary of water  -114-  level adjustment error terms is given in Appendix A.  5.1.3 Channel change analysis Within reach channel changes for each time period were analyzed by overlaying channel maps from successive periods in the GIS. All resultant features of the composite map are tabulated and summarized using the Arc/Info database. The overlay channel maps used in this analysis are composed of several thousand polygons. The GIS serves as a useful tool for examining this amount and complexity of data. For each reach, there are a total of 15 possible transitions of channel features over time, of which six indicate erosion or deposition of bed material. A summary of these channel changes is given in Table 6.  The heading 'process'  indicates the presumed meaning of each observed transition between  consecutive dates. Stable channel features are defined as polygons having the same attributes throughout the period of study. Stable islands are generally characterized by mature trees (alder and conifers) and may persist for decades or longer, particularly in upper reaches. Stable river polygons are found in locations where the thalweg has not shifted and therefore provide an indication of lateral channel stability. Stable gravel bars are assumed to be inactive sedimentation zones. In particular, it must be assumed thay they do not experience compensating erosion and deposition in the same sites. It is assumed instead that bars change through a process of persistent erosion and accretion in distinct sites. All processes which are assumed to involve no transfer of sediment are excluded from the summary calculations by assigning a depth value of 0 (hence zero volume).  -115-  Table 6: Surrirnary of channel changes for Reach I, 1966-1973. Area (m )  Depth (m)  Volume (m )  stable  211636  0  0  island  recovery  141386  0  0  river  erosion  66595  2.75  -183136  1966 Feature  1973 Feature  Process  bars  bars  bars bars  1  3  bars  floodplain  recovery  97562  0  0  island  bars  stripping  28800  0  0  island  island  stable  206537  0  0  island  river  erosion  37473  3.27  - 122536  island  floodplain  abandoned  35877  0  0  river  bars  deposition  135988  2.75  373968  river  island  deposition  2250  3.27  7358  river  river  stable  98474  0  0  river  floodplain  deposition  2248  3.27  7352  floodplain  bars  stripping  16311  0  0  floodplain  island  avulsion  3972  0  0  floodplain  river  erosion  22816  3.27  - 74607  Eroded volume (islands and floodplain)  - 197143  Deposited volume (islands and floodplain)  14710  Net storage of bars ( V - V - A V )  +128930  Total (net) depositional volume  143640  bd  be  c  Erosional features are defined simply as bars, islands orfloodplainpolygons that are covered by water on the latter date. Similarly, depositional features are defined as polygons which are covered by water on the earlier, but not the latter period. Erosion and deposition of these gravel bar sediments account for the bulk of observed (and presumed) bed material transfers; these are identified as positive (deposition) or negative (erosion) sediment transfers. In addition, it is assumed that bar gravels underlie newly formed island and floodplain depositional sites. In Table 6, net gravel bar storage for  -116-  the reach is estimated as the difference between bar deposition and erosion, minus the water level correction (for a stage decrease between 1966 and 1973 photography). In the example, there is net aggradation of gravel bar storage, which is added to the deposition of island and fioodplain sediments to estimate total net deposition in the reach. Total erosion in the reach is simply the sum of island and fioodplain erosion. In reaches where a water level adjustment is not required, the eroded volume is estimated as the total of island, fioodplain and gravel bar erosion (similarly, all depositional volumes are added).  Recovery defines the growth of vegetation (islands) on gravel bars and does not presume any transfer of bed material, although deposition of fine (wash) material probably occurs. Channel recovery may also be measured by the rate of fioodplain reconstruction, but these transitions are recorded as sediment depositioa Abandonment occurs where the river simply stops flowing (except during some floods) such that gravel bars and islands are excluded from the active channel zone. The abandonment of islands and gravel bars typically results from logjams forming along the entrance to sidechannels. Subsequent inputs of bed material are trapped upstream of the jam, causing the river to take an alternate course. These pathways can generally be distinguished by their partial revegetation from sites which appear abandoned because of lower water levels. Channel avulsions occur when these abandoned channels are reactivated as the channel shifts course in response to increased sedimentation in the reach. New islands appear to be formed as depositional features by this process, but there is little actual transfer of sediment. The transformation of islands and floodplains to bars is thought to occurfromstripping. This process is defined as the removal of vegetation and fine (wash) material but does not imply erosion of bed material. In each case, it is possible that bed material was eroded, but this could only be measured at short (flood-length) time scales.  -117-  Given some of the limitations of photo-based studies, it is obvious that some transfer of bed material will be incorrectly tabulated or 'missed', particularly where several floods occur within the time frame of the photographic record. Therefore, transport rates presented in this thesis represent only lower bound estimates. Sediment budget calculations and transport rate estimates are given in Section 5.3.  5.2 Bed Material Sediment Budget This section of the study presents the bed material sediment budget for Chilliwack River. Sediment budget calculations are based on the inputs, outputs and storage changes of bed material along study reaches (refer to discussion in Chapter 1). To construct the budget, the transport rate at a reference location of the channel must be known. It is assumed that bed material transport from Chilliwack Lake (upstream end of reach 15) is zero, which is equivalent to sediment input for the first reach (15) downstream. The difference between sediment input and net storage changes gives the sediment output for a reach. As sediment output for any reach is equal to sediment input for the next reach downstream, the calculations can be extended downstream For any reach, the transfer of bed material can be expressed as:  V of  An additional term, V  0 (trib)  = Vo(f+l) + et V r  -  r  Vdl  T5  41  is required to assess the bed material contribution from significant  tributaries. Morphologic conditions (alluvial fans) and sedimentologic evidence (decrease in grain size) at tributary junctions suggest that only Slesse Creek is important in terms of delivering a significant bed material load to Chilliwack River (though it appears to be very important). This contribution is included as an additional V term for reach 6, i.e. it is added to the right side of 0  equation [5.4] for that reach. Transport rates estimatedfromchanges in channel morphology between  -118-  1952 and 1991 are presented.  5.2.1 Net storage changes For each reach, the net change in bed material storage (A V) is determined as the difference between deposition and erosion of gravel bars, islands and fioodplain sediments (e.g. see summary at bottom of Table 6). Adjustments for bar exposure due to changes in discharge are included. These changes are given in Appendix B and summarized in Figure 29.  In upper reaches, there is very little change overall, as deposition and erosion volumes are approximately equal. The absence of very large flows due to the attenuating influence of the lake and the large size of the bed material precludes significant mobilization of bed material. Consequently, the morphology in these reaches does not change much. Observed changes in reaches 9, 10, 12, 14 and 15 are probably the result of measurement error only. The maximum net difference recorded is 31,000 m (bulk volume) for reach 12 in 1952-1966, which averages only 2200 m /yr. This result is 3  3  roughly twice as large as the maximum recorded for the other periods and is likely due to the the comparatively poor quality of the 1952 aerial photos; that is, it might consist primarily of measurement error.  By comparison, aggradation and degradation rates are relatively large in reaches 11 and 13 where recent (since 1989) fan deposits from tributaries provide a source of bed material that may be stored or mobilized during larger flows. However, these volumes are considered minor overall when compared to downstream reaches. In reach 8, there were no net storage changes prior to the 1983-91 period, when there was significant degradation. This is almost entirely due to erosion of the channel  -119-  12000  n 1 1 1 1 1 15 14 13 12 11 10  1  9  1  8  1  7  1  6  1  5  i  i  i  r  4  3  2  1  Reach  Figure 29: Net storage change of bed material (bulk volumes) -120-  banks during the 1989, 1990 floods and removal (transport) of these sediments. In reach 7 above Slesse Creek, there is no net change in bed material storage although substantial aggradation of gravel bars was qualitatively observed and there has been extensive channel widening by 1991 (see Section 4.1). There has been a net increase of bar storage of 68,000 m (bulk volume) in the reach for the 3  period 1983-91 (confirming qualitative observations - refer to Section 4.1), much of which was likely derived from bank erosion in reach 8 upstream. Sediments eroded from reach 7 are transported downstream so that this reach has overall maintained an equilibrium state.  In general, the river becomes increasingly active downstream, reflecting an increase in sediment availabilityfromtributaries and floodplain deposits. There is no systematic downstream aggradation or degradation in any period, though the magnitude of changes does seem generally to correlate with the incidence of large floods. Between 1952 and 1966 there was no significant change in any lower reach which is an indication that the channel occupied an equilibrium regime during this period. Between 1966 and 1973, reaches 2 to 6 remained stable, though there was degradation above Vedder Crossing in reach 1. The lack of large loods capable of eroding the banks limited the supply of'new' material, so the river may have simply removed sediment stored in gravel bars.  The large alluvial fan of Slesse Creek both contributes bed material directly to reach 6 and constrains local gradient, causing deposition of material suppliedfromupstream. Floods between 1973 and 1983 were large enough to remove this stored material (hence degradation occurred) though they were likely too small to erode banks or transport materialfromupstream sources. In comparison, there was significant aggradation of gravel bars at the upstream end of reach 6 during the floods of 1989 and 1990, though there was approximately an equal volume of sediment removed from bank erosion  -121-  resulting in net aggradation. In reach 5, significant degradation in the two later periods was caused by extensive bank erosion in the unconfined stretches of the river. Although bank erosion rates were even higher in reach 6 upstream, the total volume of material deposited in reach 5 was not sufficient to maintain equilibrium (as well, no large tributaries enter the reach). In 1973-83, much of this material was deposited as gravel bars in reach 4, which had the highest aggradation rate of any reach during the period of study (estimated as 17 cm averaged over the active channel zone). Later results show significant degradation in reach 4 by 1991.  There has been no significant change in storage along reach 3 in any period, which primarily acts as a conveyance zone for sediment. Bed material suppliedfromreach 4 and from Tamihi Creek probably has a very short (i.e. flood duration) residence time and aggradation probably occurs only during extended periods of moderate flows. Reach 2 had maintained equilibrium transport in all earlier periods, but there is substantial degradation by 1991. A qualitative assessment of this reach over the 1983-91 period shows that there was significant gravel bar aggradation, which is confirmed by the summary database statistics. However, there was even greater volume of erosion from island and floodplain deposits, resulting in net degradation. Some of this material was stored as bars in reach 1 downstream, but the bulk of the material was probably deposited even further downstream on Vedder fan. There was net degradation in reach 1, 1973-83 despite two large floods (in 1975 and 1980) capable of significant bank erosion. It is probable that subsequent smaller floods in 1981 and 1982 were competent to remove material stored in bar gravels. As in reach 2 upstream, there was no net aggradation of the bed in reach 1 between 1983 and 1991 despite a large increase in gravel bar storage. Extensive channel widening caused by bank erosion removed a large enough quantity of bed material to balance the volume of material deposited within the reach.  -122-  Overall, the periods 1952-1966 and 1966-1973 exhibit relatively small changes compared to those observed in the periods 1973-1983 and 1983-1991. The total sum of erosion and deposition in each period provides a relative comparison of channel activity; these results are summarized in Table 7.  Table 7: Annual erosion and deposition volumes of bed material (all reaches) 1952-1966  1966-1973  1973-1983  1983-1991  Erosion (m /yr)  92,300  118,000  192,000  275,000  Deposition (m /yr)  86,300  108,000  164,000  206,000  1.4/yr  1.7/yr  1.3/yr  1.9/yr  3  3  FIows>250 m /s 3  These patterns are coincident with the magnitude of the largest floods events. In each period, there are three or four flows that are greater than the 2-year flood, but the largest floods (i.e. >500 m /s) 3  occurred in the latter two periods only (refer to Figure 10). Nonetheless, eroded volumes exceed deposition in all periods. Some of the observed difference may also be attributed to the number of flows per year within each period capable of transporting bed material, though there is an obvious discrepancy for 1973-83. Flow duration and flood sequence may be significant factors.  5.2.2 Sediment contribution from tributaries Where a tributary introduces a significant sediment load to the main channel, this additional contribution must be accounted for in sediment budget calculations. Similarly, other inputs (i.e. landslides) or removals (i.e. gravel dredging) must be considered. All of the landslides that were observed over the photo record occurred within the lacustrine deposits and therefore contributed only fine sediments (e.g. wash material) that do not directly influence morphology. The emergence of lateral and mid-channel bars on Chilliwack River downstream of reach 7 serves as morphologic evidence that Slesse Creek contributes a significant amount of bed material. Similar evidence is not  -123-  found for other tributary systems, which are assumed to have a negligible coarse sediment contribution for purposes of the sediment budget.  If unaccounted bed material output from  tributaries is sufficiently large, then transport rates past Vedder Crossing will be underestimated, and might even appear to be negative! An assessment of the bed material contribution from Slesse Creek is desired. However, Slesse Creek is too small to reliably estimate transport from changes in channel morphology as measured from the scale of aerial photography used in this thesis. Although some larger scale (eg. 1:5000) photos are available for several years, the dates do not correspond with those used to measure changes along the Chilliwack River. Further, significant morphologic change on Slesse Creek probably occurs at much shorter time scales than on the larger Chilliwack River and information on sediment entrained between photo dates would be lost if photos from comparable dates were used. As no direct measurements of bed material transport are available for Slesse Creek, an alternative method of estimating bedload transport is required. Bed material output from Slesse Creek is estimated in the following sections based on Bagnold's (1980) transport equations.  Constructing the historical flow record sequence for Slesse Creek The gauge at Slesse Creek has been in continuous operation since 1957, but this record includes many missing data, notably during several high flow events. In order to estimate sediment transport for the entire period 1952 to 1991 using Bagnold's stream power function, a complete historic record of daily flows is required. Predicted (missing) daily streamflow data for Slesse Creek were estimated from regression equations calculated for each month of historical records. A summary of these results is given in Table 8.  -124-  Table 8: Re;gression analysis of mean montlily discharge for Slesse Creek and Vedder Crossing t  Month  Equation  r  t  Month  Equation  r  Jan  0.13X-0.4  0.89  14.6  July  0.15X+1.3  0.85  13.6  Feb  0.14X-0.7  0.84  12.4  Aug  0.15X + 0.2  0.83  12.5  Mar  0.13X-0.2  0.80  10.8  Sept  0.13X + 0.5  0.78  10.8  Apr  0.15X-0.6  0.69  8.1  Oct  0.17X-0.3  0.77  10.3  May  0.13X + 2.5  0.72  8.8  Nov  0.13X + 1.5  0.62  6.9  June  0.13X + 3.4  0.82  12.1  Dec  0.13X + 0.4  0.87  14.9  The critical value of t  0A 3 0  1  2  is 2.46 so all above relations are meaningful. For comparison, the ratio of  drainage areas between Slesse Creek and Chilliwack River at Vedder Crossing is 0.13, which is close to the slope of the regression line throughout much of the year. This ratio could also be used to predict missing daily values but was found to have slightly larger R M S E compared to the regression estimates when tested on a year of recorded (known) values. There are four months between 1952 and 1991 when both gauges were not operating continuously and daily strearnflow estimates could not be made. These missing values are significant as they all occurred at the time of known large floods in January, 1984 and November of 1989 and 1990. These flows were alternatively estimated from regression of the 40 largest recorded flows on Slesse Creek with station 08MH103, located on Chilliwack River upstream of the Slesse Creek confluence. The estimated flow of 139 m /s on 3  November 10,1990 is the largest for the period 1952 to 1991. The relation between large floods at the two gauge sites is not strongly correlated (1^=0.4, s.e. ±10.5) so flood magnitude (hence sediment transport volumes) may be incorrectly estimated. This error is not considered significant within the entire 1983-1991 period, however, as there were only four individual days when the estimated flow was sufficient to mobilize bed material.  -125-  Bedload sediment transport estimates for Slesse Creek This section of the thesis investigates the application of a theoretical bedload transport formula to provide an estimate of the bedload contribution from Slesse Creek. Gomez and Church (1989) examined 12 bedload transport formulae and found that Bagnold's (1980) stream power function performed well compared to the other formulae. The authors state that stream power equations provide the most straightforward scale correlation of flow and sediment transport and should be used when information on channel hydraulics is limited, as is the case for Slesse Creek. Bagnold's relation is also attractive to use as many of the hydraulic variables required are easy to measure or estimate.  In simple terms, Bagnold's equation states that transport will occur when stream power 'co' per unit bed area exceeds 'co ', a critical threshold value of co below which the bed is stable. This is analogous 0  to the assumption that bed material transport initiates on Chilliwack River when the flow at Vedder Crossing exceeds 250 m /s (refer to Section 3.5.4). The modified form of Bagnold's (1980) equation 3  (after Gomez and Church, 1989) is:  Y*  CD -CO  1  h =  Y." Y  'br  (<a-<0  3/2  (7/r r  2/3  (DID  y  [5.5]  m  where L, is the unit bedload transport rate in dry weight, y and y are the specific gravity of sediment s  and fluid respectively, Y is the flow depth and D is the characteristic bed material size (eg. D ). All 50  variables with subscript 'r' are reference values which Bagnold took from Williams's (1970) flume experiments; Bagnold (1980) adds that reference values can be taken from an alternate reliable data set if available. These variables are given the constant values of ^ = 0.1; (co-co ) = 0.5; Y = 0.1; G r  r  and D = 0.0011. The term y I (y -y) can be reduced to a dimensionless constant of value 1.61; it r  s  s  -126-  was introduced by Gomez and Church (1989) to convert the immersed weight of bedload sediment to a dry weight.  There remains the need to determine unit values for co and co . As given by Bagnold (1980), these 0  relations are defined as:  co = co = 5.75 [0.04 o  P  p)]  QS — w  3/2  and  (g/p)  m  [5.6] D  32  log (127AD)  The variables Y and D are the same as previously defined, g is the gravity constant, p is the mass density of fluid and p is the mass density of sediment, taken as 2650 kg/m . By substituting, co can 3  s  0  be reduced to co = 305 D  3/2  o  log (12 YID). The variables Q (strearnflow), S (energy gradient or  channel slope), w, Y and D are channel-specific hydraulic variables that must be known or measured to estimate bedload transport.  Daily mean strearnflow records were taken from historical records, as previously discussed, and grouped into the four periods based on the photo dates. Channel gradient was measured directly from aerial photos for 1983 (1:15000) and 1989 (1:5000). Repeated distance/elevation measurements were taken along the thalweg, beginning above the 'fan' and extending roughly 1 km upstream (limit of alluvial reach where most bar material is stored). Average channel gradient was calculated as 0.0272. Bankfull width was measured as 58 metres as determined from the 1983 photos only (best combination of scale and visibility). However, as the entire lateral extent of the channel bed is unlikely to be mobile at less than bankfull discharge (cf. Laronne and Duncan, 1989) a relation between discharge and flow width was developed (as for the water level adjustments). Width was allowed to  -127-  progressively increase with flow until the entire bed was assumed active (i.e. at the bankfull width of 58 m). Flow depth was estimated from the Marining equation (Eq. [1.6]) with n=0.045. The final variable, D  5 0  is a constant estimated as 0.14 m based on data from McLean (1980). Bagnold's  equation is extremely sensitive to many of the input variables, especially Y (depth), w (width) and D (grain size). Uncertainty in the accuracy of the input variables means that calculated transport estimates should be treated with caution.  The effect of varying discharge on co and co was examined in an attempt to validate the values 0  selected for input variables. The threshold value for bed material transport (co ) varied slightly with 0  flow depth, which ranged from 0.32 m at the mean daily flow (9.8 m /s) to 0.78 m at the flow of 3  record (139 m /s). Stream power (co) just exceeded the threshold value (co ) at Q=36 m /s (where 3  3  0  co =25.2 kg/m s"). The corresponding flow at Vedder Crossing (based on relations defined earlier) 1  0  was estimated as 270 m /s, which is reasonably close to the estimated transport threshold value at 3  Vedder Crossing of 250 m /s. This suggests that chosen values for the different variables used to 3  evaluate bed material transport on Slesse Creek are reasonable. However, actual rates of bed material transport may be higher than indicated on days when mean daily streamflow is near threshold values. The flow may have been much greater than the threshold value for part of the day, but when mean daily flow is determined, higher flows are attenuated and there is no apparent transport on those days. This negative bias is reduced on days when the mean discharge is much higher than the threshold value. Ideally, hourly flow data are desired to evaluate Bagnold's transport equation, but these data are rarely available. A comparison of co-co (shown as flow) with i,, (daily bedload transport rate) is G  shown in Figure 30.  -128-  Figure 30: Sediment transport rate on Slesse Creek as a function of discharge. There is minimal transport at flows less than 40 m /s. 3  On an annual basis, the maximum bed material load occurs when high flows are sustained for several days. The highest estimated annual yield of 12,200 m occurred in 1990, a year characterized by 3  three independent floods, one of which lasted 8 days and included the flood of record. There were 7 years during the period of record in which there was zero transport, the same as past Vedder Crossing. In 1972, there were 20 mean daily flows when co>co , the most of any year, although there 0  were no substantial floods. Consequently, total yield in 1972 was significantly lower than in years with larger floods (e.g. 1980,1989 and 1990). For each year, it is assumed that sediment transport is limited by the ability of Slesse Creek to convey the sediment (i.e. the transport capacity of the creek), not by the supply of available bed material. As there are considerable deposits of gravel bars along the lower reach, this assumption seems reasonable.  The formula (Eq. [5.5]) provides a unit transport rate in kg-m^-s" and must be integrated over both 1  active channel bed width (<58 m) and time (86,400 seconds/day) to calculate daily yields. For each period, all daily values were summed, then divided by the specific density of sediment to convert mass units to a volume. Estimated bed material transport rates for each period are given in Table 9.  Table 9: Bed material transport rates for Slesse Creek Period  Volume (m )  Annual yield (m /yr)  1952- 1966  3200  230  1966- 1973  3800  550  1973 - 1983  9500  950  1983 - 1991  23200  2900  3  3  Bedload transport equations are sensitive to small changes in input data, especially near threshold entrainment conditions. Bagnold's equation is known to be particularly sensitive to changes in grain -130-  size and flow velocity. McLean (1985) examined the uncertainty associated with five different bedload equations (including Bagnold's) by performing a sensitivity analysis on equation input and calibration parameters. For each variable, a sensitivity parameter (S ) was defined as the ratio of x  relative change in the transport rate relative to a change in the input variable (i.e. the first order derivative of the equation exponents). If the sensitivity parameter is high, a small uncertainty in the input value can cause a large error in the transport rate estimate. The sensitivity for all variables decreases with flow, so the largest uncertainty occurs near threshold transport conditions. For each input variable, sensitivity was plotted against a sediment mobility factor (F^, after Ackers-White, 1973) which is based on streamflow, channel depth, flow velocity and median grain size. For any given set of hydraulic parameters, the sensitivity of each input variable can be interpolated from the plots.  For any given flow, channel depth, flow velocity and width were estimated from previously described relations and used to calculate the sediment mobility factor. This was done for flows ranging between 30 m /s and 120 m /s on Slesse Creek. For the variables slope, velocity, depth and grain size, the 3  3  change in transport sensitivity relative to a change in flow was determined from linear regression (all relations were found to be significant, with r >0.98). These regression equations were then used to 2  estimate sensitivity parameters for each variable on each date over the period of record (i.e. for every mean daily flow). For each variable, sensitivity was found to varyfroma value of 3 at high flows, to 6 near threshold entrainment flows. For any variable, the sensitivity approaches infinity as flows decline, but there is no transport (hence zero error) on those days. These values are large compared with example values provided by McLean for the Fraser River because the D  5 0  of bed material on  Slesse Creek is much larger (140 mm compared to 23 mm) which lowers the mobility factor at any  -131-  comparative flow (e.g. the 2-year flood). An additional sensitivity of S =l was used to account for w  the uncertainty of channel width, which was estimated for flows less than 63 m /s (where maximum 3  width=58 m). This term was simply ignored for flows above this level.  McLean also examined the cumulative effect of random errors in all input variables using an error propagation model. This was expressed in terms of the sensitivity of each individual parameter as:  [5.7]  where each S term is the sensitivity for a given input variable and A X ! / X is the relative uncertainty x  t  error for each variable. For each date, total uncertainty is given as a percentage of the transport rate estimate for the same date. All of the input variables (channel slope, channel width, flow velocity, channel depth and grain size) used to evaluate Bagnold's transport equation along Slesse Creek were estimated, and therefore have an associated error. The slope error was calculated from the known vertical precision of the stereoplotter as ±1.5 metres over the 1 km reach, or 3% error. Errors for width, velocity and depth were estimated from a sensitivity analysis for each equation. For any given flow, these errors are conservatively given as 20%, 20% and 15% respectively. A technique for estimating the standard error of grain size distributions for a given sample size is given in Puce and Church (1996). For a sample of 100 stones (as used by McLean, 1980) the standard error of the D  5 0  is about 0.1 phi (cj>) units. This can be translated to a standard error of about 15%.  For each mean daily flow, the uncertainty in the transport rate estimate for Slesse Creek was calculated from equation [5.7]. The total of this uncertainty for each of the four periods provides an  -132-  additional error term in the sediment budget. The bed material transport rate for Slesse Creek, with uncertainty errors included, is given as 3200±5100 m for 1952-1966, 3800±5100 m for 1966-1973, 3  3  9500±12,900 m for 1973-1983 and 23,200±26,100 m for 1983-1991. In each period, the error term 3  3  is larger than the estimated transport rate, which is reasonable given that all input terms (except slope) were estimated, not measured. Despite this uncertainty, these values can be used to examine the effect of bed material inputs from Slesse Creek on the transport rate along Chilliwack River.  5.3 Estimating the sediment transport rate in Chilliwack River The sediment budget approach provides a means for estimating the average bed load transport rate of Chilliwack River. The net difference of erosion minus deposition (e.g. Table 7) for the entire channel gives the bulk volume of sediment that is transported past Vedder Crossing, excluding the sediment contribution from major tributaries. Erosion and deposition volumes are corrected for water level adjustments, and the contribution from Slesse Creek is included. A n additional term for sediment porosity is introduced here, as transport values are typically given as mineral volumes per unit time. All erosion and deposition volumes (hence storage change estimates) are multiplied by the term (l-p) wherep is given as 0.25±0.05 (after Martin and Church, 1995). By adopting the same porosity value here, transport estimates presented in this study can be compared directly to estimates given by the authors for Vedder River, which is the distal reach of Chilliwack River.  The bed material sediment budget for each period is summarized in Appendix B and presented in Figure 31. There was an estimated 4700 nrVyr of gravel transported past Vedder Crossing in 1952/66, increasing to 7900 nrVyr in 1966/73, 21700 mVyr in 1973/83 and 54600 m /yr in 1983/91. 3  The steady increase over the period of record is consistent with previous observations of channel  -133-  60000 1952-66  55000  1966-73 —1973-83 —1983-91  50000 r-  45000  £ Uin  40000  m CO  35000  <*)  30000 25000 20000 15000 10000 5000 0 -5000 15  14  13 12  11  10  9  8  7  6  Reach  Figure 31: Bed material transport rates for Chilliwack River  5  4  3  2  1  activity and the magnitude/frequency distribution of large floods. For all periods, there was minimal transport past reach 9, which is an indication that the small amounts of material that were available for transport became 'trapped'. In the earliest two periods, the transport rate remained low above Slesse Creek as there was no additional supply of bed material. During periods in which there were no significant floods (i.e. no maximum daily flows past Vedder Crossing >500 m /s) upper reaches 3  of Chilliwack River appear to have maintained an equilibrium channel regime. In contrast, the transport rate below reach 9 had increased in the later two periods. Floods greater than 500 m /s were 3  sufficient to remove sediments stored in minor bar accumulations capable of eroding channel banks along alluvial sections of the channel. Sediment contribution from tributaries in these upper reaches may also be greater than was originally assumed. The transport rate past reach 7 was 9300 mVyr in both 1973/83 and 1983/91, which is 30% and 17% of the transport rate past Vedder Crossing for the same respective periods. The rates are larger than the mean annual rates past Vedder Crossing during 1952/66 and 1966/73. It was originally thought that the transport rate past reach 7 was insignificant, which clearly is not correct when there are very large flows.  Below Slesse Creek, the transport rate generally increases because of the increased supply of mobile bed material. In the earliest periods, transport zones are not well defined because of a lack of significant channel activity. In both 1952/66 and 1966/73, there were no distinct erosional or depositional zones along the entire channel, although the transport rate increased slightly below reach 3. Although there was some bank erosion over these periods, much of the material transported was likely derived from instream deposits of bar gravels. This would account for the net degradation of bed material discussed previously. Overall, these results indicate that the channel between 1952 and 1973 was adjusting to a new regime in equilibrium with lower magnitude floods. In fact, the channel  -135-  may still have been recovering from a period of high flows between 1906 and 1921 and additional large floods in 1932, 1935, 1948, 1949 and 1951 (given in McLean, 1980). Evidence of channel recovery is clearly illustrated in Figure 32. In the absence of large floods, vegetation begins to establish on gravel bars, and the channel narrows as bed material is accreted to the fioodplain.  There was a significant increase in the transport rate by 1983, and the river had apparently developed three distinct transport zones. A n erosional zone extended from Slesse Creek downstream to reach 5, but declined in reach 4 which is a sedimentation zone (Church, 1983). Reach 4 had in fact become braided by 1983, a definite indication that deposition has increased. The large lag boulder deposits near the upstream end of reach 3 may have contributed to this pattern by trapping some of the bed material supplied from upstream, thereby decreasing upstream gradient. Below reach 3, another erosional zone emerged. The transport rate increased as the largest floods (1975, 1980)  caused  extensive bank erosion, thereby increasing the amount of available bed material. By 1991, the transport rate past Vedder Crossing had increased significantly compared to all previous periods. The channel was effectively one large erosional transport zone along lower reaches during that period. The largest floods were large enough to remove much of the bed material introduced to the channel from bank erosion and tributary inputs. Bank erosion along reaches 2 introduced more bed material to the channel than could be removed by subsequent flows. Much of this material was stored along reach 1, causing extensive aggradation of bar gravels. However, erosion from the same reach removed an equal volume of sediment from the channel banks, so that overall, no net change in the storage volume of bed material was observed.  -136-  22000 20000  • 1952-66 0 1966-73 • 1973-83 • 1983-91  18000 16000 14000 12000 10000 8000 6000 4000 2000 0  15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 Reach  Figure 32a: Rate of vegetation growth on gravel bars  50000 45000  • 1952-66 B1966-73 •1973-83 • 1983-91  40000 35000 g  30000  iL £  25000  |  20000 15000 10000 5000  mi  •I lil nJ l l  I  0 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 Reach  Figure 32b: Rate of fioodplain reconstruction  5.4 Discussion of transport estimates There are a number of assumptions and biases that may affect the reliability of the transport rate estimates. As some transfer of bed material may occur between photo dates, transport estimates represent only lower estimates of the actual transport rate. The average long-term transport rates given in this study are compared to short-term (flood-length) changes and to results given in independent studies. In addition, an error model is presented where the total uncertainty associated with the transport rate is estimated.  5.4.1 Short-term (flood length) changes Changes in bed material storage and transport rates were examined for short-term (1 to 2-year) periods to investigate variations that may not have been observed over the longer term. Additional photos for 1954, 1972, 1978, 1979, 1980 and 1989 were chosen to examine channel changes in response to frequent low-magnitude and infrequent high magnitude floods. Observed changes should permit inferences about channel behaviour during the short term, which can then be compared to longer term trends. Results are presented for reach 1 only; additional reaches were not mapped and analyzed because of the considerable effort involved. A summary of flow conditions and observed channel changes for the different periods is given in Table 10. For each period, the net storage change of bed material was estimated as previously described.  -138-  Table 10: Morphologic changes and bed material transfer rate, reach 1 Period  Q , ™ (m /s) 3  # flows > 250 m /s  Net Storage change (m )  Bank Erosion (m )  Unit transfer rate (m /yr/km)  3  3  3  3  1952-1954  351  1  87728  84332  12400  1972-1973  222  0  -227287  33483  4900  1978-1979  249  1  -69386  81562  12000  1979-1980  447  1  116074  86020  12700  1980-1983  533  5  -79047  241679  35600  1989-1991  715  10  181122  302340  44600  Notes: For periods 1980-83 and 1989-91 there are 4 independent flood events in whichflowsare capable of transporting bed material. However, there is only 1 independent event in each period in whichflowsare sufficiently large to erode channel banks. Net storage change and bank erosion are given as bulk volumes; transfer rate is given as mineral volume.  There appears to be no clear relation between the magnitude of the largest flow in any period and the net change in bed material storage. The significant degradation shown for 1972-73 is unreasonable, and probably reflects errors due to the difference in flows on the two dates (hence a large water level correction was needed) and the small scale of the 1972 photos. As there should have been little or no movement of bed material between these dates, the net storage change during this period was ignored. Water level adjustments are small for 1979-80 or 1989-91 so net storage changes on those dates should most accurately reflect the actual processes that occurred. Between 1989 and 1991, it is estimated that there was net aggradation of 180,000 m or 18 cm in reach 1, roughly twice the 3  aggradation that occurred in 1979-80 when the peak flood was much smaller. There was also significant aggradation during 1952-54 of roughly the same magnitude as in 1979-80 if the uncertainty of water level adjustments is considered. In each of these periods, the largest mean daily flood exceeded the 2-year discharge. During these floods, the bed material supplied from bank erosion exceeded the capacity of the channel to convey all of the sediment so aggradation occurred.  -139-  There was apparent degradation of69,000 m of bed material in 1978-79 (equal to an 11 cm lowering 3  of bed) despite only a single flood at the transport threshold. Although the magnitude of changes observed seems too high, degradation was also observed in the longer term (i.e. 1966-73) when there was an absence of large floods. In these periods, there is little 'new' bed material supplied to the channelfromupstream or bank erosion, and the river may have simply removed sediments stored in gravel bars. There was also apparent degradation during 1980-83 which does not fit the general pattern observed during other periods - there should have been net aggradation. As the largest flood occurred in 1980, it is probable that subsequent smaller flows were adequate to remove sediments supplied from bank erosion.  The relation between erosion of channel banks (and islands) and flow magnitude is more distinct. In 1972-73, the erosion volume is probably not greater than measurement error; this volume is equivalent to an average recession rate of 2.2 metres per metre of channel which is within the precision estimates of the photo overlays. This supports the earlier assumption that there is little bed material transport when mean flows are less than 250 m /s (though there may have been some transfer 3  of gravel bar deposits). In periods when the flood peak was greater than 250 m /s but less than 500 3  m /s, total erosion volumes are the same. These volumes are equivalent to 5.6 metres of lateral bank 3  erosion, averaged over the reach length. Peak flows in 1952-54 and 1979-80 were equal to the 2-year flood so some bank erosion would be expected during those periods. By comparison, the volume of bank erosion in 1978-79 should have been near zero given that the largest flow in the period was only equal to the estimated transport threshold. This is probably the result of photo interpretation errors as the scale of photography on the two dates (1:50000 for 1978, 1:10000 for 1979) is very different. There were difficulties distinguishing islands from bars at the 1:50000 scale because of the irnmature  -140-  age of the vegetation cover (only 3 years after the major flood of 1975). Therefore, it can be assumed that measured changes do not accurately reflect actual changes within this period. In 1980-83 and 1989-91, bank erosion was significantly greater as peak floods in both periods exceeded 500 m /s. 3  This appears to be a threshold value for the re-entrainment of sediments stored in the floodplain, causing short term aggradation above Vedder Crossing.  Erosion volumes were used to estimate the bed material transfer rate past Vedder Crossing based on the approach described by Neill (1987; see Eq. [1.4]). McLean (1990) notes that the step length between major morphologic features in a meandering channel should be about five times channel width, or about 1100 m along Chilliwack River. The repeating distance of riffles in the reach is 830 m (11 riffle zones). Both of these values are close to 900 m, which is one-half the meander wavelength distance estimated for this reach (see section 4.3.4). The unit transfer rate of bed material is determined as the average annual volume eroded per kilometre of channel. Average reach length over the photo record is 4.58 km. Although photo periods range from one to three years duration, all rates can be given as single year values because the bed was active in only one year (i.e. no bed activity in 1981, 1982 or 1991). The bed material transport rate can be estimated as the product of the unit transfer rate and the 900 m step length (see Table 10).  Transport rates were lowest during 1972-73, when only an estimated 4900 m of bed material passed 3  Vedder Crossing. Actual transport in this period may have been close to zero in light of uncertainty errors. Transport rates averaged between 12000 and 13000 m /yr in 1952-54 and from 1978 to 1980. 3  These are all presented as annual rates, though as there was only one flow in excess of 250 m /s 3  within each period, they can be considered unit values. These volumes, multiplied by the number of  -141-  flows per year in excess of the nominal transport threshold, should provide a reasonable indication of annual bed material transport rates past Vedder Crossing. The transport rates are much larger in periods when there were significant floods, reflecting an increase in bed material supply. Annual transport estimates for 1980-83 are 35,600 m and 44,600 m for 1989-91. As there were 4 3  3  independent flood events within each of these periods, it is reasonable to adjust these values accordingly (i.e. 4x the step length). Revised transport rates are given as 142,000 m from 1980-83 3  and 178,000 m from 1989-91. For comparison, Martin and Church (1995) estimated a volume of 3  1 5 7 , 0 0 0 ± 1 6 , 7 0 0 m in 1989-90 (note that there were no daily mean flows greater than 250 m /s in 3  3  1991).  These results indicate that most of the erosion and transport of bed material on Cliilliwack River occurs during a few largefloods.Ultimately, it is the frequency/magnitude distribution of these large floods that determines the morphology and transport regime of Chilliwack River. By contrast, there is very little sediment transport in years with no significant flows because the supply of material available for transport is limited. Therefore, average longer-term transport rates presented in Section 5.3 underestimate annual rates, the variations of which are large.  5.4.2 Results from other studies The validity of transport estimates given in this study can be further examined by comparing estimates to those given in other studies. Using a flow duration curve and direct bedload measurements, McLean (1980) estimated an annual transport rate of 76,000 m /yr past Vedder Crossing. He also 3  averaged these results with estimates from other sources (cross sections, bank erosion rates) and produced a revised estimate of58,000 m /yr from 1940 to 1976. This rate can be roughly compared 3  -142-  0  with the 4700 rrrVyr for 1952/66 and the 7900 rnVyr estimated for 1966/73. His value is significantly larger than the estimates presented here (no measure of uncertainty was provided). Although there has been compensating scour and fill of bed material within the time frame of this study, McLean's estimates appear too high. Martin and Church (1995) estimated the mean annual gravel transport into Vedder Crossing over the period 1981-90 to be 36,600 ± 5600 m /yr (mineral volume). Annual 3  estimates were also given; these ranged from a low of 19,000 nrVyr in 1981-82 to a high of 157,000 nrVyr in 1989-90. To directly compare these results with the transport rate given in this study (for the period 1983 to 1991), the estimates for 1981-82 and 1982-83 were excluded. In addition, it was assumed that at least as much material was transported in the 1990 flood as during the smaller 1989 flood. Based on these changes, a revised transport estimate of47,000 nrVyr can be given. This value is remarkably similar to the estimate of 54,600 nrVyr presented in this study, and well within the uncertainty error of both studies.  Although these results appear to validate the procedures followed in both studies, it is probable that there has been compensating erosion and deposition between survey (photo) dates. In each period, there are several floods capable of eroding channel banks, and as many as 20 flow events capable of transporting bed material. As a result, transport rates presented here should be considered as lower bound estimates only. This may partly explain the large difference between rates discussed here, and those given in McLean (1980). Although Chilliwack River was in a transient state of equilibrium between 1952 and 1973, this does not preclude bed material transport. However, there is little morphologic evidence that the channel had steadily degraded or that there was sufficient throughput (i.e. material from upstream reaches or tributaries that is not deposited in lower reaches) to sustain high sediment yields during this period. As well, it seems unlikely that transport rates pre-1975 would  -143-  be higher than in 1981-90 (cf. Martin and Church, 1995), given the frequency and magnitude distribution of floods in these periods. Despite these differences, transport estimates presented in this study still provide important information on the relative rates of bed material transport over the 40year period of study.  5.4.3 Error Analysis Errors in the precision of mappingfromaerial photographs, estimates of bed material depth and water level corrections (in lower reaches) all introduce uncertainty in the accuracy of computed results. In upper reaches, low channel banks, the small (narrow) size of the channel and shadows all make accurate digitizing more difficult than in lower reaches. Therefore, when the bankline for adjacent years is digitized and overlaid, there may be locations where there appears to have been bank erosion or reconstruction, though there may have been no actual transfer of sediment.  This error was  estimated as 3.8m (for 1 year) or 4.8m (for successive years) in Section 4.3.3. If channel displacements are equally distributed, erosion and deposition errors will be compensating and no net storage change will be observed. However, this 'measurement error' may cause apparent negative transport rates along some upper reaches, which is clearly not possible. It may be more accurate to determine errorsfromnet volumetric storage changes in reaches which are unlikely to have changed over the period of record. For example, reaches 14 and 15 have very stable banks and little stored bed material so these reaches should have no significant aggradation or degradation. The maximum net volumetric storage change that is observed over some period should therefore represent a precision error in [stable] upper reaches.  Alternatively, errors in any reach can be estimatedfromthe precision of photo overlays, reach lengths  -144-  and bank height estimates. No water level correction is used above the Slesse Creek confluence (reach 6/7) so this uncertainty can be ignored in all upper (and some lower) reaches. Photo overlay errors (see discussion in Section 3.2.2) were estimated as ±2.8 metres, including both photogrammetric and operator error. This means that if a feature boundary is digitized over two consecutive dates, a maximum displacement of 2.8 metres may be estimated, though there may have been no actual change in position. As any feature (channel bank, island, gravel bar) is potentially subject tc^this error, precise determination of the maximum possible error is very difficult. Complicating the error analysis is the fact that many of these errors will be compensating. For example, a shift in channel position caused by placement error will result in apparent erosion of one channel (or gravel bar) bank and deposition on the other. On average, these errors should be roughly equal and may not affect net storage changes or transport rates, but will nonetheless propogate downstream, inflating the magnitude of uncertainty. Compensating errors will also inflate erosion and deposition volumes estimated for any reach.  A method for calculating uncertainty when there are compensating measurement errors is given in Martin and Church (1995). For each reach, errors are based upon uncertainty in the transport rate estimates. The transport rate for reach n is defined as V = V - A V. Uncertainty in the transport o  {  rate for each reach can be calculated as:  E =  3A,  6A  ( dV. dp,  + ESlesse  [5.8]  Where SVj is the uncertainty in the transport rate at a reference point (i.e. lake), SAj is the uncertainty in the storage change estimates for each reach, 5pj is is the uncertainty of the porosity  -145-  estimates and E  s f c s s e  is the error in the transport rate for Slesse Creek. As the transport rate from the  lake is known to be zero, the left term in the equation can be eliminated and error propagation is therefore based only on the precision estimates of storage changes for each reach. The transport rate for each period is therefore estimated as V ± E . The additional error term for Slesse Creek is added D  n  to the total error estimated for each of the six lower reaches as it affects the transport rate estimates, rather than the storage change estimates.  For all periods, the uncertainty in storage changes along upper reaches was estimated as the average of net absolute storage changes for reaches in which A V should equal zero. Uncertainty estimates for upper reaches were estimated as 780, 730, 390 and 930 mVyr (bulk volumes) for the different periods. It is assumed that these averaged errors apply equally to any individual upper reach. The error becomes increasingly large downstream because of cumulative additions in the uncertainty equation. Total cumulative error estimates for upper reaches are 1800 mVyr in 1952/66, 1600 mVyr in 1966/73, 900 mVyr in 1973/83 and 2100 mVyr in 1983/91. Some of the error in 1952/66 can be attributed to the comparatively poor quality of the 1952 aerial photos, while some of the supposed error in 1983/91 may be the result of actual erosion and transport.  Below Slesse Creek, uncertainty in the transport estimates increases sharply as several additional error terms are included (refer to Appendix A). For each of the lower six reaches, there is an uncertainty associated with and bank height estimates, the adjustment for different water levels between photo dates and errors in applying the water level adjustment to reaches where no correction is necessary (see discussion in section 5.1.2). In general, these latter two terms are much larger than depth errors, but they are not included for every reach. The results of the error analysis  -146-  give an uncertainty of the bed material transport rate of V ± 2500 m /yr for 1952-66, V ± 3900 3  0  0  mVyr for 1966-73, V ± 12000 mVyr for 1973-83 and V ± 10000 nrVyr for 1983-91. Transport 0  0  estimates shown with this range of errors is given in Figure 33. These errors represent about 50% of the transport rate past Vedder Crossing in each of the first three periods, but only 18% in the period 1983-1991 when the transport rate was proportionately much greater.  For comparison with these results, Martin and Church (1995) gave a precision estimate of ± 5600 m /yr for the gravel transport rate on Vedder River based on volume changes estimated from repeated 3  cross-section surveys (their precision estimate also included uncertainty associated with grain size and sediment porosity). This indicates that transport rates estimated from photo overlays are approximately as reliable as from cross-section surveys if the flow on comparative photo dates is similar. However, the errors may be unacceptably large for periods when the flow on comparative dates is disparate and bed material transport rates are small. The morphology-based sediment budget should probably be avoided in that situation.  -147-  Figure 33: Bed material transport rate with estimated error ranges  Chapter 6: Summary and Conclusions  This study has examined changes in channel morphology on Chilliwack River between 1952 and 1991. Channel features were initially mapped from historic aerial photography for five separate dates using an analytic stereoplotter. Maps for consecutive dates were then overlaid and analyzed with a GIS. Observed changes were used to examine the processes governing both short- and long-term patterns of channel development. Changes were also examined within the context of a sediment budget to provide an estimate of bed material transport rates and describe the sediment cascade along the channel. Many of the procedures and techniques discussed in this study have been generalized so that they can be applied to examine sediment yields and land-use impacts on other channels. In particular, this study may have have important implications for the management of large gravel-bed rivers within British Columbia.  Our understanding of the long-term stability of alluvial rivers can be improved by examining patterns of erosion and deposition along channel reaches. Reaches which are in equilibrium at some time scale will neither be actively aggrading nor degrading. The volume of bed material eroded from each reach will be equal to the volume deposited, so the net storage change will be zero, even though the bed may be active (i.e. there is bed material transport). Reaches which are either aggrading or degrading at some time scale may experience important changes in aquatic and riparian habitat quality. Changes in bed material storage may reflect changes in the flow regime or sedimentation as the consequence of land-use activities. Aggradation is of particular concern along Chilliwack River as it can lead to channel avulsions along lower reaches which can reduce access to overwinter habitats and cause extensive bank erosion, hence property and infrastructure loss. The procedures presented in this study  -149-  provide a valuable means of measuring these changes and observing their impacts.  There are two main patterns of instability along Chilliwack River. Upper channel reaches (from Chilliwack Lake downstream to Slesse Creek confluence) have remained relatively stable over the 40 year period of record examined in this study. In general, the large size of bed material, partial confinement of the channel by terraces and the moderating influence of the lake on flow magnitude all preclude significant modification of the channel. Frequent small floods (i.e. bankfullflows)are not sufficiently large to transport the available limited supply of bed material, allowing the channel to maintain an equilibrium regime. During exceptional floods such as occurred in 1989 and 1990, there was increased mobilization of bed material supplied from tributary streams and some erosion of channel banks and wooded islands. This 'new material was transported downstream past Slesse Creek 1  confluence during subsequent smaller flows until the supply diminished. It was originally assumed that there was minimal bed material transport in upper reaches, which is not correct. However, the increase in flow magnitude, flood frequency and sediment transport in these reaches did not cause a shift to a disequilibrium channel regime.  Below Slesse Creek, the channel widens considerably as it flows within an alluvial fioodplain along much of its length. This transition in morphology is due to an increase in sediment supply, a decline in bed material size and an increase in flood flow magnitude and variability. The morphology of Chilliwack River in lower reaches is dominated by high-magnitude, low-frequency floods. In general, floods greater than 500 m /s which have a return period of about 5 years can be considered as 3  significant channel forming events. These floods cause extensive bank erosion, introducing a considerable supply of'new* sediment to the active channel. In the short-term, this causes aggradation  -150-  above Vedder Crossing as the river is not competent to remove all of the available material. However, as bed material transport occurs several times per year on average when flows are larger than about 250 m /s, there is no long-term aggradation above Vedder Crossing. There was significant growth 3  of gravel bar storage in all reaches (except reach 3) by 1991, giving the impression of overall aggradation in the channel, but the results presented here do not support that assumption. In all periods, erosion of material stored in the floodplain removed an even greater volume of sediment.  This study has shown that contemporary channel processes on Chilliwack River primarily reflect the re-entrainment of sediments which may persist in the floodplain for extensive periods of time. Madej (1984) estimated residence times in the 720 km Redwood Creek basin from tens to thousands of 2  years. The average annual flux of these sediments may be small relative to the total volume stored in the floodplain, such that this supply is not generally limited. A cursory examination of the airphoto s shows this to be true for Chilliwack River, although the actual volume of bed material stored in the floodplain was not estimated. Therefore, during a period of sustained high flows or numerous large floods, sediment transport rates remain high. It was estimated that about 55,000±10,000 mVyr passed through Vedder Crossing during the period 1983-1991, or about 10 times the average annual volume during 1952-1966 when there was little bank erosion.  Between 1952 and 1966, channel width remained stable (and even decreased) as the channel continued to recover from an earlier sequence of large floods in the 1930s and 1940s. Recovery continued until 1973, even though transport rates increased over this time. It is probable that the length of the photo interval (14 years) between 1952 and 1966 was too large to adequately characterize actual annual transport rates. However, resolved transport rates were constant over the  -151-  entire 1952-1973 period within the margins of uncertainty errors. As development in the watershed expanded during this period of low channel activity (i.e. since the 1950s), increased erosion and flooding in recent years has been blamed on timber harvesting in tributary watersheds. However, this study has demonstrated that the mobilization of bed material is directly related to the magnitude of the large floods in 1989 and 1990. It is probable that similar high rates of channel activity occurred in the decades prior to this study and near the beginning of the century. This finding has significant implications for studies of land-use impacts and channel change on other coastal British Columbia rivers. Studies which relate observed changes within consecutive time periods to each other, rather than to an absolute reference such as stable regime form, may incorrectly assume that land use impacts from [e.g.] forestry are responsible for modifying the channel. It is important to examine flow trends over the same periods (where these data are available) for specific land-use impacts to be identified.  The methodologic approach presented in this study is mainly limited by the availability of suitable historic airphotos. Ideally, all photography should be at roughly the same scale to minimize interpretive errors, although the positional accuracy of the stereoplotter (including magnification of the images) assists in minimizing these errors. Significantly greater uncertainty in channel change measurements and transport rate estimates will occur if a similar device is not available for channel mapping. The most common scale of available photography is 1:15000 to 1:20000 which is appropriate for channels from 30 m to several hundred metres wide. On very large rivers (i.e. bankfull width > 1 km) scale of 1:40000 to 1:70000 are ideal as this reduces the total number of stereo models that must be used.  -152-  The time interval between successive photo dates is also important to characterize the temporal scale at which actual morphologic changes occur: On smaller rivers, significant morphologic changes may occur on an annual basis. Therefore, photos taken at yearly intervals should be used when available. This may also apply to larger rivers with an annual nival flood. The most important factor limiting the accuracy of the transport estimates is difficult to control. If the flow on different dates of photography is disparate, there may be considerable [apparent] erosion or deposition between successive dates, even if no actual changes occurred. Ideally, the flow on all dates of photograpjhy should be similar, particularly if there are wide, braided reaches. This limits the usefulness of the airphoto/GIS approach compared to transport estimates based on cross-section changes (cf. Martin, 1992) or bed topographic changes (cf. McLean, 1990; Lane et al., 1994). These two techniques also avoid the requirement to measure bed material depth along the channel. These limitations could be mitigated by measuring at-a-station hydraulic geometry at a number of channel cross-sections. Nonetheless, the method presented in this study allows a much longer history of channel development to be studied compared to more conventional techniques such as repeated cross-section surveys or direct bedload sampling, for which good data are rarely available on most rivers. This is of particular importance on larger rivers where channel response to different land-use impacts may not occur for years to decades following the initial disturbance.  -153-  REFERENCES  Ashmore, Peter and Church, Michael (in press). "Sediment Transport and River Morphology: a Paradigm for Study." In Klingeman, P.C., Beschta, R.L. and Bradley, J. [eds.]. Gravel-bed Rivers in the Environment. Proceedings, 4th International Workshop on Gravel-bed Rivers. Westview, CO., Water Resources Press. Bagnold, R.A. (1980). "An empirical correlation of bedload transport rates in flumes and natural  rivers." Proceedings Royal Society, London, Series A, Vol. 372, pp. 453-473. Benda, Lee (1990). "Changing patterns of erosion due to land use." EOS, Transactions, American Geophysical Union, Vol. 71, No. 3. Beschta, Robert L. (1984). "River Channel Responses to Accelerated Mass Soil Erosion." In C.L. OLoughlin and A.J. Pearce [eds.]. Symposium on Effects of Forest Land Use on Erosion and Slope Stability. Environment and Policy Institute East-West Center, University of Hawaii, Honolulu, pp. 155-163. Bonham, Eric (1980). "Resolving the Fish vs Flood Control Controversy on the Vedder River." Canadian Water Resources Journal, Vol. 5, No. 1, pp. 38-53. Bowman, Jim (1992). "The Chilliwack: The River that Couldn't be Tamed." B. C. Historical News Fall, 1992. Bray, D.I. (1973). "Representative Discharges for Gravel-Bed Rivers in Alberta, Canada." Journal  of Hydrology, Vol. 27, pp. 143-153. Burrough, P.A. (1986). Principles of Geographical Information Systems for Land Resources Assessment. Clarendon Press, Oxford. Caine, Nel and Swanson, F.J. (1989). "Geomorphic coupling of hillslope and channel systems in two small mountain basins." Zeitschrift fur Geomorphologie, Vol. 22, No. 2, pp. 189-203. Carling. P.A. (1992). "Hydraulics and Sediment Transport." In Calow, P. and Petts, G.E. [eds.]. The  Rivers Handbook, Vol. I. Blackwell, pp. 101-125. Carson, M . A . and Griffiths, G.A. (1987). "Bedload Transport in Gravel Channels." Journal of Hydrology, New Zealand. Vol. 26, No. 1, pp. 1-151. Channel Assessment Procedure Guidelines (1996). Participant Manual, Workshop Proceedings. Chatwin, S.C., Howes, D.E., Schwab, J.W. and Swanston, D.N. (1991). A Guide for Management ofLandslide-Prone Terrain in the Pacific Northwest. B.C. Ministry of Forests, Land Management Handbook Number 18. -154-  Chilliwack River Valley Ratepayers Environmental Management Committee (1993). ''Chilliwack River Hazard Management Study Interim Report." Unpublished supplemental report submitted to the Regional District of Fraser-Cheam. Chow, Ven Te (1964).  Handbook ofApplied Hydrology. New York: McGraw-Hill, Inc.  Church, Michael (1983). "Pattern of instability in a wandering gravel bed channel." In Collinson, J.D. and Lewin, J. [eds.]. Modern and Ancient Fluvial Systems. Special Publications, International Association Sedimentology, No. 6, pp. 169-180. Church, Michael (1985). "Bed Load in Gravel-Bed Rivers: Observed Phenomena and Implications for Computation." Proceedings, Canadian Society for Civil Engineering, 7th Hydrotechnical Conference, Saskatoon, pp. 17-38. Church, M . and Miles, M.J. (1987). "Meteorological antecedents to debris flow in southwestern British Columbia; Some case studies." Geological Society of America, Reviews in Engineering Geology, Volume VII, pp. 63-78. Church, M . , Miles, M . , Rood, K. (1987). "Sediment Transfer along Mackenzie River: A Feasibility Study." Environment Canada, Western and Northern Region. IWD-WNR(NWT) WRB-SS-87-1. Church, M.A. (1992). "Channel Morphology and Typology." In Calow, P. and Petts, G.E. [eds.] The  Rivers Handbook, Vol. I. Blackwell, pp. 126-143. Church, M.A. and McLean, D.G. (1994). "Sedimentation in Lower Fraser River, British Columbia: Implications for Management." In Schumm, S.A. and Winkley, B.R [eds.]. The Variability of Large Alluvial Rivers. American Society of Civil Engineers Press, pp. 221-241. Coble, D.W. (1961). "Influence of Water Exchange and Dissolved Oxygen in Redds on Survival of Steelhead Trout Embryos." Technical Paper No. 1381, Oregon Agricultural Experimental Station. Collins, C M . (1990). "Morphometric Analyses of Recent Channel Changes on the Tanana River in the Vicinity of Fairbanks, Alaska." U.S. Army Corps of Engineers, Cold Regions Research & Engineering Laboratory, Report 90-4. Desloges, J.R. and Church, M . A . (1992). "Geomorphic implications of glacier outburst flooding: Noeick River valley, British Columbia." Canadian Journal of Earth Sciences, Vol. 29, No. 3, pp. 551-564. Dietrich, W.E. and Dunne, Thomas. (1978). "Sediment budget for a small catchment in mountainous terrain." Zeitschrift fur Geomorphologie N.F., Suppl. Bd. 29. pp. 191-206. Dietrich, W.E., Dunne, T., Humphrey, N.F. and Reid, L . M . (1982). "Construction of sediment budgets for drainage basins." In Swanson, F.J., Janda, R.J., Dunne, T. and Swanston, D.N. [eds.]. Sediment budgets and routing in forested drainage basins. Forest Service General Technical Report PNW-141,pp. 5-23.  -155-  Downward, S.R., Gurnell, A . M . and Brookes, A . (1994). "A methodology for quantifying river channel planform change using GIS." In Olive, L.J., Loughran, R.J. and Kesby, J.A. [eds.]. Variability in Stream Erosion and Sediment Transport, IAHS Publication No. 224. Dunne, T. and Leopold, L.B. (1978). Water in Environmental Planning. New York: W.H. Freeman and Company. Eaton, Brett (1995). "Bankfull Discharge Return Periods." Unpublished research paper prepared for Professor M . A . Church, Department of Geography, The University of British Columbia. Environmental Systems Research Institute, Inc. [ESRI] (1993). Understanding GIS the Method. Longman Scientific & Technical.  ARC/INFO  Eschner, T.R., Hadley, R.F. and Crowley, K.D. (1983). "Hydrologic and Morphologic Changes in Channels of the Platte River Basin in Colorado, Wyoming, and Nebraska: A Historical Perspective." Hydrologic and Geomorphic Studies of the Platte River Basin. Geological Survey Professional Paper 1277-A. Farrell, George (1987). DFO/MOEP Fish Habitat Inventory and Information Program Stream Information Summary - Chilliwack River. Goff, J. and Ashmore, P.E. (1994). "Gravel transfer rates and morphologic changes in braided Sunwapta River, Alberta, Canada." Earth Surface Processes and Landforms, Vol. 19, pp. 195-212. Gomez, Basil and Church, Michael (1989). "An Assessment of Bed Load Sediment Transport Formulae for Gravel Bed Rivers." Water Resources Research, Vol. 25, No. 6, pp. 1161-1186. Graf, W.L. (1984). "A Probabilistic Approach to the Spatial Assessment of River Channel Instability." Water Resources Research, Vol. 20, No. 7, pp. 953-962. Gregory, K.J. and Walling, D . E . (1973). Drainage Basin Form and Process. London: Edward Arnold. Gupta, Avijit (1983). "High-magnitude floods and stream channel response." In Collinson, J.D. and Lewin, J. [eds.]. Modern and Ancient Fluvial Systems. Special Publications, International Association Sedimentology, No. 6, pp. 219-227. Haan, C.T., Barfield, B.J. and Hayes, J.C. (1994). Design Hydrology and Sedimentology for Small Catchments. San Diego: Academic Press Inc. Ham, D . G . (1995). "A Comparison of Methods to Monitor Stream Channel Morphology." Unpublished Report prepared for the Watershed Restoration Program, B.C. Ministry of Environment, Lands and Parks.  -156-  Hay & Company Consultants Inc. (1992). "Chilliwack River Hazard Management Study - Interim Report. Volume 1: Text and Volume 2: Figures and Drawings." Prepared for the Regional District of Fraser- Cheam. Henderson, F.M. (1966). Open channel Flow. Macmillan Series in Civil Engineering. Toronto: The Macmillan Company, Collier-Macmillan Canada, Ltd. Hickin, E.J. (1983). "River channel changes: retrospect and prospect." In Collinson, J.D. and Lewin, J. [eds.]. Modern and Ancient Fluvial Systems. Special Publications, International Association Sedimentology, No. 6, pp. 61-83. Hickin, E.J. and Nanson, G.C. (1984). "Lateral Migration Rates of River Bends." Hydraulic Engineering, A S C E , Vol. 110, No. 11, pp. 1557-1567.  Journal of  Hooke, J.M. (1980). "Magnitude and Distribution of Rates of River Bank Erosion." Earth Surface Processes, Vol. 5, pp. 143-157. Jordan, P. and Slaymaker, O. (1991). "Holocene Sediment Production in Lillooet River Basin, British Columbia: A Sediment Budget Approach." Geogaphie physique et Quaternaire, Vol. 45, No. 1, pp. 45-57. Jordan, Peter (1990). "Hydrology of the November, 1989, Chilliwack River Flood, and some Observations on the Impacts of Forest Management." Unpublished report prepared for B.C. Ministry of Forests, Chilliwack District. Karlinger, M.R., Eschner, T.R., Hadley, R.F. and Kircher, J.E. (1983). "Relation of Channel-Width Maintenance to Sediment Transport and River Morphology: Platte River, South-Central Nebraska. Hydrologic and Geomorphic Studies of the Platte River Basin. Geological Survey Professional Paper 1277-E. Kellerhals, R., Church, M . and Bray, D.I. (1976). "Classification and Analysis of River Processes."  Journal of the Hydraulics Division, A S C E , Vol. 102, No. H Y 7, pp. 813-829. Kellerhals, Rolf and Church, Michael (1989). "The Morphology of Large Rivers: Characterization and Management." In Dodge, D.P. [ed.]. Proceedings of the International Large River Symposium. Canadian Special Publications Fisheries Aquatic Science, Vol. 106, pp. 31-48. Kelsey, H . , Madej, M.A., Pitlick, J., Coghlan, M . , Best, D., Belding, R. and Stroud, P. (1981). "Sediment Sources and Sediment Transport in the Redwood Creek Basin: A Progress Report." Redwood National Park Technical Report Number 3. Lane, S.N., Chandler, J.H. and Richards, K.S. (1994). "Developments in monitoring and modelling small-scale river bed topography." Earth Surface Processes andLandforms. Vol. 19, No. 4, pp. 349368.  -157-  Laronne, J.B. and Duncan, M.J. (1989). "Constraints on duration of sediment storage in a wide, gravel-bed river, New Zealand." In R.F. Hadley and E.D. Ongley [eds.]. Sediment and the Environment, IAHS Publ. no. 184, pp. 165-172. Lawler, D . M . (1986). "River bank erosion and the influence of frost: a statistical examination." Transactions Institute British Geographers. N.S. 11: 227-242. Lyons, J.K. and Beschta, R.L. (1983). "Land Use, Floods, and Channel Changes: Upper Middle Fork Willamette River, Oregon (1936-1980)." Water Resources Research, Vol. 19, No. 2, pp. 463-471. Madej, Mary Ann (1984). "Recent Changes in Channel-Stored Sediment, Redwood Creek, California." Redwood National Park Research and Development, Technical Report 11. Martin, Yvonne E. (1992). "Sediment Budget from Morphology: Vedder River, British Columbia." M.Sc. Thesis, Department of Geography, The University of British Columbia. Martin, Yvonne and Church, M . A . (1995). "Bed-Material Transport Estimated from Channel Surveys: Vedder River, British Columbia." Earth Surface Processes and Landforms, Vol. 20, pp. 347-361. McLean, David (1980). "Flood Control and Sediment Transport Study of the Vedder River." M.A. Sc. Thesis, Department of Civil Engineering, The University of British Columbia. McLean, David (1985). "Sensitivity Analysis of Bedload Equations." Proceedings, Canadian Society for Civil Engineering, 7th Hydrotechnical Conference, Saskatoon, pp. 1-15. McLean, David (1990). "Channel Stability on Lower Fraser River." PhD Thesis, Department of Geography, The University of British Columbia. Nanson, G.C. and Hickin, E.J. (1986). "A statistical analysis of bank erosion and channel migration in western Canada." Geologic Society of America Bulletin, V . 97, pp. 497-504. Neill, Charles R. (1971). "River Bed Transport related to Meander Migration Rates." Journal of the Waterways, Harbors and Coastal Engineering Division, Proceedings of the American Society of Civil Engineers, Vol. 97, No. WW4, pp. 783-786. Neill, C R . (1987). "Sediment Balance Considerations Linking Long-term Transport and Channel Processes." In C R . Thorne, J . C Bathurst and R.D. Hey [eds.]. Sediment Transport in Gravel-bed Rivers. John Wiley & Sons Ltd., pp. 225-240. Orchard, Imbert (1983). Floodland and Forest: Memories of the Chilliwack Series Number 37. Victoria: Provincial Archives.  Valley. Sound Heritage  Parker, G . , Klingeman, P . C and McLean, D.G. (1982). "Bedload and size distribution in paved gravel-bed streams. Journal of the Hydraulics Division, A S C E , Vol. 108, No. HY4, pp. 544-571.  -158-  Peters, Neill J. (1978). "Vedder River Salmon and Flood Control." M.A.Sc. Thesis, Department of Civil Engineering, The University of British Columbia. Phillips, Robert W. (1971). "Effects of Sediment on the Gravel Environment and Fish Production." Symposium on Forest Land Uses and Stream Environment Proceedings, Oregon State University, October 19-21, 1970. Popov, I.V. (1962). "Application of Morphological Analysis to the Evaluation of the General Channel Deformations of the River Ob'." Transactions State Hydrologic Institute (Trudy GGI). pp. 267-324. Rasmussen, Peter F. (1991). The partial duration series approach to flood frequency analysis. Institute of Hydrodynamics and Hydraulic Engineering, Technical University of Denmark, Series Paper no. 55. Reid, L . M . , Dunne, T., and Cederholm, J. (1981). Application of Sediment Budget Studies to the Evaluation of Logging Road Impact. Journal of Hydrology, New Zealand. 20 (1), pp. 49-62. Reid, I. and Frostick, L.E. (1994). "Fluvial sediment transport and deposition." In Pye, Kenneth [ed.]. Sediment Transport and Depositional Processes. Blackwell Scientific Publications, pp. 89-155. Rice, Stephen and Church, Michael (1996). "Sampling Surficial Fluvial Gravels: The Precision of Size Distribution Percentile Estimates." Journal of Sedimentary Research, Vol. 66, No. 3, pp. 654-665. Roberts, Richard G. and Church, Michael. (1986). "The sediment budget in severely disturbed watersheds, Queen Charlotte Ranges, British Columbia." Canadian. Journal Forest Research, Vol. 16. pp. 1092-1106. 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. Ministry of Forests, Land Management Report No. 34. Saunders, I.R., Clague, J.J. and Roberts, M . C . (1987). "Deglaciation of Chilliwack River Valley, British Columbia." Canadian Journal of Earth Sciences, Vol. 24, pp. 915-923. Schurnm, S.A. (1977). "Applied Fluvial Geomorphology." In John R. Hails [ed.]. Applied Geomorphology. Elsevier Scientific Publishing Company, New York. Scrivener, J.C. and Brownlee, M.J. (1989). "Effects of Forest Harvesting on Spawning Gravel and Incubation Survival of Chum and Coho Salmon in Carnation Creek, British Columbia. Canadian Journal of Fish Aquatic Science, Vol. 46,1989. Simons, D.B. and Albertson, M . L . (1960). "Uniform Water Conveyance Channels in Alluvial Materials." Journal of the Hydraulics Division. Proceedings of the American Society of Civil Engineers, Vol. 86, No. H Y 5, pp. 33-71.  -159-  Slama, Chester C. Manual Photogrammetry, 1980.  of Photogrammetry.  Falls Church, Virginia: American Society of  Swanson, F.J.. Benda, L . E . , Duncan, S.H., Grant, G.E., Megahan, W.F., Reid, L . M . and Ziemer, R.R. (1987). "Mass Failures and Other Processes of Sediment Production in Pacific Northwest Forest Landscapes." In Salo, E.O. and Cundy, T.W. [eds.]. Streamside Management: Forestry and Fisheries Interactions. University of Washington Institute of Natural Resources, Contribution No. 57.  Thomson, Bruce and VanDine, Doug (1995). "Slope Hazards Investigation Trip - Chilliwack River Watershed." Unpublished field trip guide. Thorne, C R . and Lewin, J. "Bank Processes, Bed Material Movement and Planform Development in a Meandering River." In Rhodes, D.D. and Williams, G.P. [eds.]. Adjustments of the Fluvial System. Dubuque, Iowa: Kendal Hunt. Thurber Consultants Ltd. (1988). "Geotechnical Assessment of Slope Stability in the Chilliwack Valley." Report to Regional District of Fraser Cheam. Vancouver. Tripp, D.B. and Poulin, V . A . (1986). "The Effects of Mass Wasting on Juvenile Fish Habitats in Streams on the Queen Charlotte Islands." B.C. Ministry of Forests, Lamd Management Report No. 50. Tripp, D.B. and Poulin, V . A . (1992). "The Effects of Logging and Mass Wasting on Juvenile Salmonid Populations in Streams on the Queen Charlotte Islands." B.C. Ministry of Forests, Lamd Management Report No. 80. Verstappen, H.Th. (1977). Remote Sensing in Geomorphology. Amsterdam: Elsevier Scientific Publishing Company.  -160-  APPENDIX A: Summary of Storage Change Error Terms  1952-1966 Reach 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1  Upper 10930 10930 10930 10930 10930 10930 10930 10930 10930 n.a. n.a. n.a. n.a. n.a. n.a. (±m3)  Depth n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. 1507 755 5484 3294 7055 11007 (±m3)  Water Level n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. 11356 n.a. 10660 5123 (±m3)  W/L Application n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. (±m3)  Cumulative Error 586 829 1015 1177 1317 1441 1556 1663 1764 2129 2130 2254 2267 2387 2490 (± m3/yr)  Upper 5090 5090 5090 5090 5090 5090 5090 5090 5090 n.a. n.a. n.a. n.a. n.a. n.a. (±m3)  Depth n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. 718 151 495  Water Level n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. 5443 2811 (±m3)  W/L Application n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. 15883  Cumulative Error 546 772 953 1099 1231 1346 1454 1553 1646 2539 2539 3698 3699 3834 3925 (± m3/yr)  1966-1973 Reach 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1  476 2682 3400 (±m3)  Note: Cumulative error given in mineral volume. All other error terms given in bulk volumes.  n.a. n.a. n.a. (±m3)  APPENDIX A: Summary of Storage Change Error Terms  1973-1983 Reach 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1  Upper 3913 3913 3913 3913 3913 3913 3913 3913 3913 n.a. n.a. n.a. n.a. n.a. n.a. (±m3)  Depth n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. 9438 8353 13868 4606 30777 46159 (±m3)  Water Level n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. 47245 28176 51581 32832 (±m3)  W/L Application n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. 37775 n.a. n.a. (±m3)  Cumulative Error 293 415 545 623 700 760 815 866 944 2792 3115 6567 8381 10585 12188 (± m3/yr)  Upper 7398 7398 7398 7398 7398 7398 7398 7398 7398  Depth n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. 2242 9366 6747 1432 7480 10827 (±m3)  Water Level n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. 28154 n.a. 31798 21769 (±m3)  W/L Application n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. 21663 n.a. n.a. (±m3)  Cumulative Error 694 982 1207 1393 1568 1717 1854 2020 2136 5426 5953 6148 7697 9385 10099 (± m3/yr)  1983-1991 Reach 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1  n.a. n.a. n.a. n.a. n.a. n.a. (±m3)  Note: Cumulative error given in mineral volume. All other error terms given in bulk volumes.  Appendix B: Sediment Budgets for Chilliwack River  1952-1966 Reach 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1  Dep  Eros  Av  Vi  Vo  6973 29120 27293 34124 8729 18959 2492  5379  1594 4244 6713 -23110 -14283 5770 -6270 5274 11116 11040 -10572 -3139 -27918 -6521 -16385  0  -1594  -1594 -5838  -5838 -12551 10559 24842 19073 25343 20069 8953 1089 11661 14800 42718 49238 65623  31802 32253 66525 75772 125208 18441 216059 212764  24876 20580 57233 23012 13189 8762 26528 21137 55485 86344 128347 46359 222580 229149  -12551 10559 24842 19073 25343 20069 8953 1089 11661 14800 42718 49238  4687 ± 2500 m3/yr  1966-1973 Reach  Dep  Eros  Av  Vi  Vo  15 14 13  6397  8862 13229 26360 42050  -2465 4241 -13133  0 2465 -1775 11357  2465 -1775 11357  7088 -3889  -3889 -6099 -12001  12 11 10 9 8 7 6 5 4 3 2 1  17469 13228 46319 17190 15022 9326 24487 36578 40595  6213 12812 3425 29039 35203 35133  4269 10977 2210 5902 -4553  56329  57166  1376 5462 -837  119271 18659 38036 107730  131686 14716 54056 147857  -12415 3943 -16020 -40127  -6099 -12001  7088  -7448 -8824  -7448 -8824 -10467  -10467  -9630  -9630 2785 -1158 14862  2785 -1158 14862 54989  7856 ± 3900 m3/yr Note: storage change estimates given as mineral volumes  Appendix B: Sediment Budgets for Chilliwack River  1973-1983 Reach 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1  Dep  Eros  Av  Vi  Vo  5060 12665 9884  3412 11315 30670  1648 1350 -20786  0 -1648 -2998  -1648 -2998 17789  59189 22232 13961 5013  66290 36034 10802 6431  -7102 -13802 3158 -1418  17789 24890 38692 35534  24890 38692 35534 36951  32036 13784 23928 49946 217582 41507 281903 442637  35848 38422 96889 114637 161555 23778 303381  -3812 -24638 -72961 -64691 56027 17729 -21478 -56546  36951 40763 65401 147895 212585 156559 138830 160308  40763 65401 147895 212585 156559 138830 160308 216853  499183  + Sl=9500  21685 ± 12200 m3/yr  1983-1991 Reach  Dep  Eros  Av  Vi  Vo  15 14  3583 9787  13 12  28370  4668 11590 15070  -1085 -1803 13301  0 1085 2888  1085 2888 -10412  11 10  58901 22676 16379  54728 45821 26889  4173 -23144 -10511  -10412 -14585  9 8  2116 24071  12287  -10171  8559 19070  -14585 8559 19070  7  58673 133064  72192 55409  -48122 3263  29240 77362  118502 308856 188846 51305 356231 327149  14563 -130368 -73689 -15748 -127474  74099 82754 213122 286811 302559 430032  6 5 4 3 2 1  178488 115157 35558 228758 320018  -7131  29240 77362 74099 82754  + Sl=23200  213122 286811 302559 430032 437163  54645 ± 10100 m3/yr Note: storage change estimates given as mineral volumes  


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