UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Linking fluvial dynamics to white sturgeon habitat in the Nechako River, BC. Gauthier-Fauteux, Simon 2017

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

Item Metadata

Download

Media
24-ubc_2017_may_gauthierfauteux_simon.pdf [ 81.07MB ]
Metadata
JSON: 24-1.0342969.json
JSON-LD: 24-1.0342969-ld.json
RDF/XML (Pretty): 24-1.0342969-rdf.xml
RDF/JSON: 24-1.0342969-rdf.json
Turtle: 24-1.0342969-turtle.txt
N-Triples: 24-1.0342969-rdf-ntriples.txt
Original Record: 24-1.0342969-source.json
Full Text
24-1.0342969-fulltext.txt
Citation
24-1.0342969.ris

Full Text

Linking fluvial dynamics to white sturgeon habitat in theNechako River, BCbySimon Gauthier-FauteuxBA, Geography, Concordia University, 2013a thesis submitted in partial fulfillmentof the requirements for the degree ofMaster of Scienceinthe faculty of graduate and postdoctoral studies(Geography)The University of British Columbia(Vancouver)February 2017c© Simon Gauthier-Fauteux, 2017AbstractConsiderable effort has been dedicated to restoring sturgeon habitat within dammed rivers.However, sedimentation causes long-term failure because interstitial voids provide critical habi-tat during early life-stages. Based on the premise that a better understanding of geomorphicprocesses will improve restoration design, this study characterizes flow and sediment transportdynamics through a white sturgeon spawning reach on the Nechako River, BC.An extensive dataset was collected throughout the 2015 flood. Bedload transport wassampled on 36 days with flows ranging from 44 m3/s to 656 m3/s. During a high flow of 525m3/s, channel bathymetry and water surface elevation were surveyed and velocity profiles werecollected across 9 transects. Banklines, bars and island topography were later surveyed duringlow flow.Sediment transport into the reach was positively related with discharge. This relation wasnon-linear and transport rates increased rapidly once flows exceeded 400 m3/s. The relationweakened with downstream distance and sediment transport peaked progressively later through-out the year. No relation was observed at the downstream end of the reach, where transportrates remained low and constant relative to upstream.Sediment was primarily transported through secondary channels conveying a disproportion-ate amount of sediment compared to flow. Within the single-thread channel, the locationsconveying the greatest amount of sediment remained spatially consistent over time.Hydrodynamic modelling indicates the Burrard Ave. Bridge causes backwatering once dis-charge exceed 225-275 m3/s. Velocity, shear stress and transport capacity at the downstreamend of the reach do not increase with discharge because of the backwatering and the expansionin channel width through the island complex. The locations of maximum shear stress andtransport capacity shift upstream with increasing discharge, but shear stress does not exceed23 N/m2 for flows up to 775 m3/s.The fluvial dynamics within the spawning reach create challenges and opportunities forhabitat restoration. Backwatering is problematic because it causes mid-reach deposition duringhigh flows and limits shear stress magnitude over the downstream spawning substrate. Mean-while, the presence of sediment transport pathways through secondary channels and within themainstem can be used to site restoration projects in areas apt to maintain suitable habitat.iiPrefaceThis thesis is a continuation of geomorphic research conducted within the framework of theNechako White Sturgeon Recovery Initiative. Portions of fieldwork and analysis were donein collaboration with Northwest Hydraulic Consultants Ltd. (NHC) and have been publishedelsewhere (2015 Sediment Transport Investigation on the Vanderhoof Reach of the NechakoRiver, available at www.nechakowhitesturgeon.org). All thesis work was completed under theguidance of a supervisory committee that included supervisor Dr. Brett Eaton (University ofBritish Columbia), Dr. Andre Zimmermann (Northwest Hydraulic Consultants) and Dr. SteveMcAdam (BC Ministry of Environment).iiiTable of ContentsAbstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiPreface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiiTable of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ivList of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viiList of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viiiAcknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 Context . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.2 White sturgeon spawning and habitat . . . . . . . . . . . . . . . . . . . . . . . . 11.3 Bedload sediment transport in relation to sturgeon habitat . . . . . . . . . . . . 21.4 Restoring white sturgeon habitat . . . . . . . . . . . . . . . . . . . . . . . . . . . 41.5 Numerical modelling and its role in sturgeon habitat restoration . . . . . . . . . 51.6 Study rationale . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72 Study Site . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82.1 Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82.2 Hydrology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92.3 History . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113.1 Bathymetry, topography and water surface elevation . . . . . . . . . . . . . . . . 113.2 Flow velocity and discharge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123.3 Bedload sediment transport . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174.1 Bathymetry and water surface elevation . . . . . . . . . . . . . . . . . . . . . . . 174.2 Velocity and flow conveyance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19iv4.3 Bedload sediment composition and transport rates . . . . . . . . . . . . . . . . . 244.4 Patterns of sediment transport through the study reach . . . . . . . . . . . . . . 275 Modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 315.1 Initialization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 315.2 Calibration and validation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 335.3 Simulation results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 376 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 456.1 Variation in flow dynamics with discharge . . . . . . . . . . . . . . . . . . . . . . 456.2 Characterization of bedload transport within the spawning reach . . . . . . . . . 466.3 Implications for larval habitat and restoration . . . . . . . . . . . . . . . . . . . . 506.4 Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 517 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56A Real-Time Kinematic (RTK) Survey on the Nechako River . . . . . . . . . 63A.1 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63A.1.1 Configurations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63A.1.2 Data collection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63A.1.3 Data processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66A.2 Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72A.3 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72B Velocity and Discharge Measurement in the Nechako River using an Acous-tic Doppler Current Profiler (ADCP) . . . . . . . . . . . . . . . . . . . . . . . 73B.1 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73B.1.1 Data collection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73B.1.2 Data processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78B.2 Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92B.3 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92C Bedload Sampling Protocol . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93C.1 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93D Geographic Information System (GIS) Data Processing and Analysis . . . 95D.1 Data processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95D.2 Maps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95vE Analysis of Annual Sediment Load using a Rating Curve Approach . . . . 100E.1 Bedload sediment transport . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100F Bedload Sediment Transport through the Nechako Spawning Reach . . . . 103F.1 Data processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103F.2 Maps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103G Two-Dimensional Flow Modelling of the Nechako River using Nays2DH . 108G.1 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108G.1.1 Configurations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108G.1.2 Processing the grain size data . . . . . . . . . . . . . . . . . . . . . . . . . 113G.1.3 Processing the model output . . . . . . . . . . . . . . . . . . . . . . . . . 118G.2 Model results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137G.3 Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156G.4 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156viList of TablesTable 4.1 Proportional flow conveyance and mean velocity across ADCP transects . . . 21Table 4.2 Bedload tranport through secondary channel MU-A and mainstem channelMU-C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28Table 4.3 Bedload tranport through mainstem channel ML-A compared to secondarychannels ML-B and ML-C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29Table 5.1 Simulated velocity and depth compared to ADCP data . . . . . . . . . . . . . 35Table 5.2 Simulated shear stresses with increasing discharge . . . . . . . . . . . . . . . . 40Table A.1 Configurations used for RTK surveying . . . . . . . . . . . . . . . . . . . . . . 63Table A.2 Comparison of PPP and GCM baseline methods . . . . . . . . . . . . . . . . . 67viiList of FiguresFigure 2.1 White sturgeon spawning reach on the Nechako River near Vanderhoof, BC . 8Figure 2.2 Pre- and post-regulation annual maximum daily discharge . . . . . . . . . . . 10Figure 3.1 Surveyed bathymetry, bar contours and bankline topography . . . . . . . . . 12Figure 3.2 ADCP velocity transects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13Figure 3.3 Moving bed at TRA causing offset between GPS sentences and Bottom Trackpositioning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14Figure 3.4 Bedload sampling locations in 2015 . . . . . . . . . . . . . . . . . . . . . . . . 15Figure 3.5 2015 hydrograph with bedload sampling and surveying dates . . . . . . . . . 16Figure 4.1 Digital elevation model of the spawning reach . . . . . . . . . . . . . . . . . . 18Figure 4.2 Bed and WSE profile along the mainstem channel at 525 m3/s . . . . . . . . 19Figure 4.3 Cross-channel velocity profiles at transects TRA, TRH and TRI . . . . . . . 22Figure 4.4 Depth-averaged velocity across transects TRA, TRH and TRI . . . . . . . . . 23Figure 4.5 D84, D50 and D16 grain sizes of bedload sediment . . . . . . . . . . . . . . . . 24Figure 4.6 Relation between bedload transport and discharge at progressively down-stream sampling locations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25Figure 4.7 Sampled bedload transport rates throughout the 2015 hydrograph . . . . . . 26Figure 4.8 Mean bedload transport rate through different channels (sampled between400-700 m3/s) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28Figure 4.9 Cross-channel bedload transport rates sampled in 2015 at the US and LPtransects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30Figure 5.1 Measured versus modelled stage at WSC gauge location . . . . . . . . . . . . 32Figure 5.2 Measured and simulated WSE profiles for a discharge of 525 m3/s after cali-bration of channel roughness . . . . . . . . . . . . . . . . . . . . . . . . . . . 34Figure 5.3 Measured versus simulated cross-channel velocity for a discharge of 525 m3/s 36Figure 5.4 Simulated versus measured water surface profiles . . . . . . . . . . . . . . . . 37Figure 5.5 Relation between discharge and velocity within the mainstem channel . . . . 38Figure 5.6 Simulated WSE profiles along the mainstem channel showing the develop-ment of backwater upstream of the Burrard Ave. Bridge . . . . . . . . . . . . 39viiiFigure 5.7 Reach-scale distribution of shear stresses with increasing discharge . . . . . . 41Figure 5.8 Reach-scale sediment transport capacity with increasing discharge . . . . . . 42Figure 5.9 Total cross-channel transport capacity with increasing discharge . . . . . . . 43Figure 5.10 Calculated transport capacity and sampled bedload transport at the US andLP transects plotted as a function of discharge . . . . . . . . . . . . . . . . . 44Figure 6.1 Schematic representation of bedload sediment transport through the spawn-ing reach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49Figure A.1 Establishing local control using GCM 653659 . . . . . . . . . . . . . . . . . . 64Figure A.2 RTK base-station positioned in front of the NWSCI . . . . . . . . . . . . . . 65Figure A.3 Installation used for bathymetric survey . . . . . . . . . . . . . . . . . . . . . 65Figure A.4 Surveying top-of-bank topography . . . . . . . . . . . . . . . . . . . . . . . . 66Figure A.5 PPP report for observation 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . 68Figure A.6 PPP report for observation 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . 69Figure A.7 PPP report for observation 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . 70Figure A.8 PPP report for observation 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . 71Figure B.1 Teledyne RDI RiverRay ADCP raft used to collect velocity profiles . . . . . . 74Figure B.2 Boat speed versus flow velocity across transect TRA . . . . . . . . . . . . . . 75Figure B.3 Boat speed versus flow velocity across transect TRB . . . . . . . . . . . . . . 75Figure B.4 Boat speed versus flow velocity across transect TRC . . . . . . . . . . . . . . 75Figure B.5 Boat speed versus flow velocity across transect TRD . . . . . . . . . . . . . . 76Figure B.6 Boat speed versus flow velocity across transect TRE . . . . . . . . . . . . . . 76Figure B.7 Boat speed versus flow velocity across transect TRF . . . . . . . . . . . . . . 76Figure B.8 Boat speed versus flow velocity across transect TRG . . . . . . . . . . . . . . 77Figure B.9 Boat speed versus flow velocity across transect TRH . . . . . . . . . . . . . . 77Figure B.10 Boat speed versus flow velocity across transect TRI . . . . . . . . . . . . . . 77Figure B.11 Submerged banklines and low-lying islands during data collection . . . . . . . 78Figure B.12 ADCP transect TRH used to set magnetic variation . . . . . . . . . . . . . . 79Figure B.13 Moving bed offsetting GPS and BT ship tracks at transect TRA . . . . . . . 80Figure B.14 Moving bed offsetting GPS and BT ship tracks at transect TRB . . . . . . . 80Figure B.15 Additional configurations used in WinRiver II for data processing . . . . . . 81Figure B.16 Cross-channel velocity profile at transect TRA . . . . . . . . . . . . . . . . . 81Figure B.17 Cross-channel velocity profile at transect TRB . . . . . . . . . . . . . . . . . 82Figure B.18 Cross-channel velocity profile at transect TRC . . . . . . . . . . . . . . . . . 82Figure B.19 Cross-channel velocity profile at transect TRD . . . . . . . . . . . . . . . . . 83Figure B.20 Cross-channel velocity profile at transect TRE . . . . . . . . . . . . . . . . . 83Figure B.21 Cross-channel velocity profile at transect TRF . . . . . . . . . . . . . . . . . 84Figure B.22 Cross-channel velocity profile at transect TRG . . . . . . . . . . . . . . . . . 84ixFigure B.23 Cross-channel velocity profile at transect TRH . . . . . . . . . . . . . . . . . 85Figure B.24 Cross-channel velocity profile at transect TRI . . . . . . . . . . . . . . . . . . 85Figure B.25 Summary statistics for ADCP transects . . . . . . . . . . . . . . . . . . . . . 86Figure B.26 Mean and standard deviation of depth-averaged velocity across-transect TRA 87Figure B.27 Mean and standard deviation of depth-averaged velocity across-transect TRB 88Figure B.28 Mean and standard deviation of depth-averaged velocity across-transect TRC 88Figure B.29 Mean and standard deviation of depth-averaged velocity across-transect TRD 89Figure B.30 Mean and standard deviation of depth-averaged velocity across-transect TRE 89Figure B.31 Mean and standard deviation of depth-averaged velocity across-transect TRF 90Figure B.32 Mean and standard deviation of depth-averaged velocity across-transect TRG 90Figure B.33 Mean and standard deviation of depth-averaged velocity across-transect TRH 91Figure B.34 Mean and standard deviation of depth-averaged velocity across-transect TRI 91Figure D.1 TIN generated from surveyed elevations . . . . . . . . . . . . . . . . . . . . . 96Figure D.2 2015 DEM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97Figure D.3 Water surface elevation during a discharge of 525 m3/s . . . . . . . . . . . . . 98Figure D.4 Water depth during a discharge of 525 m3/s . . . . . . . . . . . . . . . . . . . 99Figure E.1 Bedload rating curve developed for the Upper Site showing hysteresis . . . . 101Figure E.2 Predicted versus observed bedload transport rate at Upper Site in 2015 . . . 101Figure E.3 Bedload transport at the Lower Patch showing no clear relation with dis-charge in 2015 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102Figure F.1 Sampled transport rates (discharge below 100 m3/s) . . . . . . . . . . . . . . 104Figure F.2 Sampled transport rates (100-200 m3/s) . . . . . . . . . . . . . . . . . . . . . 104Figure F.3 Sampled transport rates (200-300 m3/s) . . . . . . . . . . . . . . . . . . . . . 105Figure F.4 Sampled transport rates (300-400 m3/s) . . . . . . . . . . . . . . . . . . . . . 105Figure F.5 Sampled transport rates (400-500 m3/s) . . . . . . . . . . . . . . . . . . . . . 106Figure F.6 Sampled transport rates (500-600 m3/s) . . . . . . . . . . . . . . . . . . . . . 106Figure F.7 Sampled transport rates (600-700 m3/s) . . . . . . . . . . . . . . . . . . . . . 107Figure G.1 Solver type calculation conditions . . . . . . . . . . . . . . . . . . . . . . . . . 108Figure G.2 Boundary conditions (for simulated discharge of 523 m3/s) . . . . . . . . . . 109Figure G.3 Entire modelling domain showing elevation in meters . . . . . . . . . . . . . . 110Figure G.4 Vegetation density used in simulations . . . . . . . . . . . . . . . . . . . . . . 110Figure G.5 grain size distributions input to the model . . . . . . . . . . . . . . . . . . . . 111Figure G.6 Manning’s roughness used for simulations . . . . . . . . . . . . . . . . . . . . 111Figure G.7 Rating curve developed to specify WSE at the downstream extent of themodel domain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112Figure G.8 Modelled vs measured stage at the WSC gauge used to develop the down-stream rating curve . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112xFigure G.9 Substrate image Region 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113Figure G.10 Substrate image Region 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114Figure G.11 Substrate image Region 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114Figure G.12 Substrate image Region 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115Figure G.13 Substrate image Region 5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115Figure G.14 Region 1 grain size distribution . . . . . . . . . . . . . . . . . . . . . . . . . . 116Figure G.15 Region 2 grain size distribution . . . . . . . . . . . . . . . . . . . . . . . . . . 116Figure G.16 Region 3 grain size distribution . . . . . . . . . . . . . . . . . . . . . . . . . . 117Figure G.17 Region 4 grain size distribution . . . . . . . . . . . . . . . . . . . . . . . . . . 117Figure G.18 Region 5 grain size distribution . . . . . . . . . . . . . . . . . . . . . . . . . . 118Figure G.19 Model velocity validation at transect TRA . . . . . . . . . . . . . . . . . . . 119Figure G.20 Model velocity validation at transect TRB . . . . . . . . . . . . . . . . . . . . 120Figure G.21 Model velocity validation at transect TRC . . . . . . . . . . . . . . . . . . . . 121Figure G.22 Model velocity validation at transect TRD . . . . . . . . . . . . . . . . . . . 122Figure G.23 Model velocity validation at transect TRE . . . . . . . . . . . . . . . . . . . . 123Figure G.24 Model velocity validation at transect TRF . . . . . . . . . . . . . . . . . . . . 124Figure G.25 Model velocity validation at transect TRG . . . . . . . . . . . . . . . . . . . 125Figure G.26 Model velocity validation at transect TRH . . . . . . . . . . . . . . . . . . . 126Figure G.27 Model velocity validation at transect TRI . . . . . . . . . . . . . . . . . . . . 127Figure G.28 Model depth validation at transect TRA . . . . . . . . . . . . . . . . . . . . . 128Figure G.29 Model depth validation at transect TRB . . . . . . . . . . . . . . . . . . . . . 129Figure G.30 Model depth validation at transect TRC . . . . . . . . . . . . . . . . . . . . . 130Figure G.31 Model depth validation at transect TRD . . . . . . . . . . . . . . . . . . . . . 131Figure G.32 Model depth validation at transect TRE . . . . . . . . . . . . . . . . . . . . . 132Figure G.33 Model depth validation at transect TRF . . . . . . . . . . . . . . . . . . . . . 133Figure G.34 Model depth validation at transect TRG . . . . . . . . . . . . . . . . . . . . . 134Figure G.35 Model depth validation at transect TRH . . . . . . . . . . . . . . . . . . . . . 135Figure G.36 Model depth validation at transect TRI . . . . . . . . . . . . . . . . . . . . . 136Figure G.37 Simulated shear stress during a discharge of 45 m3/s . . . . . . . . . . . . . . 137Figure G.38 Simulated shear stress during a discharge of 75 m3/s . . . . . . . . . . . . . . 138Figure G.39 Simulated shear stress during a discharge of 175 m3/s . . . . . . . . . . . . . 138Figure G.40 Simulated shear stress during a discharge of 275 m3/s . . . . . . . . . . . . . 139Figure G.41 Simulated shear stress during a discharge of 375 m3/s . . . . . . . . . . . . . 139Figure G.42 Simulated shear stress during a discharge of 475 m3/s . . . . . . . . . . . . . 140Figure G.43 Simulated shear stress during a discharge of 575 m3/s . . . . . . . . . . . . . 140Figure G.44 Simulated shear stress during a discharge of 675 m3/s . . . . . . . . . . . . . 141Figure G.45 Simulated shear stress during a discharge of 775 m3/s . . . . . . . . . . . . . 141Figure G.46 Estimated sediment transport capacity during a discharge of 45 m3/s . . . . 142Figure G.47 Estimated sediment transport capacity during a discharge of 75 m3/s . . . . 142xiFigure G.48 Estimated sediment transport capacity during a discharge of 175 m3/s . . . . 143Figure G.49 Estimated sediment transport capacity during a discharge of 275 m3/s . . . . 143Figure G.50 Estimated sediment transport capacity during a discharge of 375 m3/s . . . . 144Figure G.51 Estimated sediment transport capacity during a discharge of 475 m3/s . . . . 144Figure G.52 Estimated sediment transport capacity during a discharge of 575 m3/s . . . . 145Figure G.53 Estimated sediment transport capacity during a discharge of 675 m3/s . . . . 145Figure G.54 Estimated sediment transport capacity during a discharge of 775 m3/s . . . . 146Figure G.55 Profile of downstream transport capacity during a discharge of 45 m3/s . . . 146Figure G.56 Profile of downstream transport capacity during a discharge of 75 m3/s . . . 147Figure G.57 Profile of downstream transport capacity during a discharge of 175 m3/s . . . 147Figure G.58 Profile of downstream transport capacity during a discharge of 275 m3/s . . . 148Figure G.59 Profile of downstream transport capacity during a discharge of 375 m3/s . . . 148Figure G.60 Profile of downstream transport capacity during a discharge of 475 m3/s . . . 149Figure G.61 Profile of downstream transport capacity during a discharge of 575 m3/s . . . 149Figure G.62 Profile of downstream transport capacity during a discharge of 675 m3/s . . . 150Figure G.63 Profile of downstream transport capacity during a discharge of 775 m3/s . . . 150Figure G.64 Cumulative transport capacity for a sequence of low-flow hydrographs . . . . 152Figure G.65 Cumulative transport capacity for a sequence of typical hydrographs . . . . . 152Figure G.66 Cumulative transport capacity for a sequence containing a high flow hydrograph153Figure G.67 Cumulative transport capacity for the 2015 flood . . . . . . . . . . . . . . . . 153Figure G.68 Cumulative transport capacity (uniform GSD) for a sequence of low-flowhydrographs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154Figure G.69 Cumulative transport capacity (uniform GSD) for a sequence of typical hy-drographs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154Figure G.70 Cumulative transport capacity (uniform GSD) for a sequence containing ahigh flow hydrograph . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155Figure G.71 Cumulative transport capacity (uniform GSD) for the 2015 flood . . . . . . . 155xiiAcknowledgmentsI would like to thank my supervisor, Dr. Brett Eaton, for creating an excellent learning en-vironment for me to develop as a geomorphologist. Dr. Eaton is an excellent teacher whosupports his students’ development through the stimulation of curiosity and critical thought.I am also very grateful to have had Dr. Andre Zimmermann and Dr. Steve McAdam on mycommittee. Both of these committee members provided me with incredible insight, attentionand support along the way. It is no surprise that I have forged great relationships, both personaland professional, with all three of these people.I would also like to thank all the members of the Nechako White Sturgeon Recovery Ini-tiative. This team is accomplishing incredible work and I feel very lucky to be a part of theeffort. The Nechako White Sturgeon Conservation Center was very welcoming and providedassistance, personnel and equipment, which made the project possible.I am also grateful to have had the opportunity to collaborate with Northwest Hydraulic Con-sultants throughout the project. NHC provided me with the necessary resources to accomplishthis research and allowed me to gain valuable professional experience.My work would not have been possible without the help of the Carrier-Sekani Tribal Council.Their generosity and effort is something I have rarely come across. I could never have expectedto make such good friends along the way.The district of Vanderhoof, BC, was a great place to work. I thank Wayne Salewski andthe community for their help and interest. I also thank Dr. Michael Church for his generosity.Finally, I gratefully acknowledge the financial support from the Natural Sciences and Engi-neering Research Council of Canada, the Province of British Columbia through the Ministry ofEnvironment and the Government of Canada through the department of Fisheries and Oceans.xiiiChapter 1Introduction1.1 ContextAt present, the Nechako, Columbia and Kootenay River populations of white sturgeon (Acipensertransmontanus) are experiencing recruitment failure primarily caused by the effects of flow reg-ulation (DFO, 2014). Geomorphic effects, including changes in substrate composition, occurin response to flow regulation as the fluvial system adjusts to the imposed flow and sedimentregimes (Church, 1995; Grant et al., 2003). White sturgeon are susceptible to these changesbecause of their reproductive life-history traits (Winemiller, 2005; Lytle and Poff, 2004) andbecause substrate characteristics determine survival during early life-stages (McAdam, 2011;McAdam et al., 2005).In response to declining sturgeon populations, considerable effort has been dedicated torehabilitate habitat within degraded river systems (NWSRI, 2012; KTOI, 2009; Crossmanand Hildebrand, 2014). However, the functionality of restored habitat has often been short-lived because altered fluvial dynamics continue to produce geomorphic change that is non-conducive to sturgeon survival (Crossman and Hildebrand, 2014; Johnson et al., 2006; NHC,2012). This study uses a biogeomorphic approach to understand how sturgeon habitat is affectedby sediment transport through a critical spawning reach of the Nechako River near Vanderhoof,BC. The premise of this research is that a better understanding of geomorphic processes withinthe spawning reach will increase the effectiveness of habitat restoration by allowing projects tobe designed in accordance with the changing fluvial system.1.2 White sturgeon spawning and habitatWhite sturgeon are a slow growing, long-lived species of fish found within the Fraser, Columbiaand Sacramento River systems of western North America (Hildebrand et al., 2016). In Canada,they inhabit the Fraser, Nechako, Columbia and Kootenay Rivers. Females may require 15-30years to reach sexual maturity, but once mature, they can spawn multiple times throughouttheir lives at intervals of 3 years or more (Hildebrand et al., 2016). This species has a periodic1life-history strategy, where high fecundity spawners reproduce periodically over time in a form ofintergenerational bet-hedging (Winemiller, 2005). This life-history trait is suited to large-scaleenvironmental variations because recruitment can be achieved through episodic reproductivesuccess despite long periods with unfavorable conditions (Winemiller, 2005; Coutant, 2010).Although the environmental/physiological cues that trigger spawning are poorly under-stood, water temperature has been described as a relatively good predictor of spawn timing(Hildebrand et al., 2016). In the snowmelt-dominated fluvial systems of western Canada, watertemperature typically rises to produce favorable spawning conditions in late spring or earlysummer. This timing generally corresponds to the spring freshet, creating a strong temporalassociation between the period of peak flow and sturgeon spawning activity. Spawning typicallyoccurs in relatively deep, moderate to high velocity flow and has been documented along me-ander bends (Paragamian et al., 2009), within side-channels (Perrin et al., 2003), at tributaryconfluences (Hildebrand et al., 1999) and below dam tailraces (Parsley and Beckman, 1994).Sturgeon spawn by broadcasting negatively buoyant eggs into the water column that adhere tothe substrate surface until hatch (Hatfield et al., 2013). Once the eggs hatch, yolksac larvaeimmediately seek refuge within the interstitial pore spaces, where they must remain hidden forapproximately 12-15 days until the onset of exogenous feeding to increase survival (McAdam,2011). After this period, larvae emerge from the substrate and drift towards lower velocityrearing habitat (Hatfield et al., 2013).Larvae cannot access interstitial refuge habitat if the gravel substrate has been infilledor covered by sand. Sedimentation and lack of interstitial habitat results in pre-mature drift,higher larval mortality and higher predation (Kock et al., 2006; McAdam, 2011). The infilling oflarval habitat has been identified as a cause of population decline within the Nechako (McAdamet al., 2005), Columbia (McAdam, 2015) and Kootenay Rivers (Paragamian, 2012). Substrateconducive to larval survival therefore consists of medium to coarse gravel (Bennett et al., 2007)with no more than a minimal proportion of fine sediment within the upper layer.1.3 Bedload sediment transport in relation to sturgeon habitatSediment transport determines the quality and availability of substrate habitat through its in-fluence on bed surface composition (Hassan and Church, 2000), grain stability (Hassan et al.,2007) and intergravel flow (Greig et al., 2005; Zimmermann and Lapointe, 2005; Lisle, 1989).The interaction between local hydrodynamics, substrate characteristics and sediment mobil-ity can produce geomorphic adjustment within spawning reaches (Eaton and Lapointe, 2001;Lapointe et al., 2000) that cause egg/larvae burial and suffocation (Kock et al., 2006). Al-though fluvial dynamics driving sediment transport are inherently variable (Ashmore, 1991)and influenced by external factors including sediment supply (Buffington and Montgomery,1999), hydrograph shape (Hassan et al., 2006) and the superimposition of sedimentological fea-tures (Hoey, 1992; Nicholas et al., 1995), certain regularities and characteristics can be used toidentify how suitable habitat is naturally maintained within a spawning reach.2Channel morphology directly influences habitat availability by generating flow dynamicscapable of spatially sorting, or segregating, sediment based on size. Suitable habitat foundalong the outer bank of large meander bends (McDonald et al., 2010) can result from lateralsediment sorting if inward acting, near-bed secondary flow circulation is sufficient to move finegrained sediment upwards along the point-bar slope (Dietrich and Smith, 1984; Dietrich andWhiting, 1989; Powell, 1998). Areas of flow convergence can also provide coarse substratehabitat as water super-elevation generates helicoidal vortices acting to excavate and maintainscour holes (Ashmore and Parker, 1983; Rhoads and Kenworthy, 1995). The junction angleand discharge ratio of confluent channels also influences how bedload is transported through,or around, such scour holes (Best, 1988; Roy and Bergeron, 1990).Habitat characteristics also vary depending on cross-channel location because a relativelysmall proportion of the total channel width can convey most of the bedload sediment (Gomez,1991; Habersack et al., 2008; Ashmore et al., 2011). The channel width conveying bedload sed-iment and undergoing short-term morphological change is known as the active width (Ashmoreet al., 2011). Even for a nearly straight, single-thread reach, slight asymmetry in the channelbed can cause cross-sectional variation in transport rates and bedload transport may be highestalong topographic lows (Habersack et al., 2008).Although the active width of a channel can increase with stream power, it commonly remainsonly a fraction of the total channel width (Ashmore et al., 2011). Significant areas of grainimmobility remain due to the development of high intensity transport zones (Lisle et al., 2000)and due to the feedback between partially mobile coarse fractions and bed structuring (Churchand Hassan, 2002). Sediment immobility can be beneficial or problematic for larval sturgeondepending on the pre-existing substrate characteristics and the composition of the bedloadsediment. Poor correlation between bed shear stress and particle size during bankfull flow(Lisle et al., 2000) can be beneficial because coarse, stable substrate located in areas with lowshear stress can provide refuge habitat for larvae during the high flow spawning period.The surficial composition of substrate in regulated rivers is often not conducive to larvalsurvival because bed texture adjusts to changing sediment input, stream power and hydrographshape (Church, 1995; Hassan et al., 2006; McDonald et al., 2010). Decreased stream competenceunder a regulated flow regime can result in a bed surface composed of coarse, immobile particles(Church, 1995) that progressively infills with fine sediment depending on sediment supply, flowconditions and substrate to bedload grain size ratio (Gibson et al., 2009). Medium to coarsesand is especially apt to infill a static surface layer because it is large enough to resist suspensionand fine enough to bridge inter-gravel spaces; creating a sand seal within the upper framework(Beschta and Jackson, 1979; Lisle, 1989). Once a sand seal has been formed, increasing flowand deceasing sediment supply does not entrain infilled fines to a significant depth (Beschtaand Jackson, 1979). Therefore, periodic mobility of coarse grains during high flow is a keymechanism in flushing stream gravels and providing suitable larval habitat.Sediment transport dynamics influence the availability and quality of sturgeon habitat3within the four major Canadian rivers they inhabit. In the Kootenay River, critical spawninghabitat has been identified along meander bends (Paragamian et al., 2009) where pre-regulationflows could maintain suitable larval habitat by sorting sediment and scouring the bed (McDon-ald et al., 2010). Within the Columbia River, sturgeon spawn immediately downstream of atributary confluence (Golder, 2006) which has complex hydrodynamic circulation (Fissel andJiang, 2008). The relative discharge between confluent channels affects eddy circulation and in-fluences the path of sediment transport, where pre-regulation flows may have effectively routedbedload around the spawning substrate rather than overtop of it (McAdam, 2015). In thelower Fraser River, spawning occurs (not exclusively) within side-channels (Perrin et al., 2003).This type of spawning habitat is interesting because bedload can be conveyed through braidedreaches by a subset of anabranches (Bertoldi et al., 2010; Ashmore et al., 2011), suggestingthat some spawning sites may not be exposed to active bedload transport during the spawningperiod. Finally, in the Nechako River, sturgeon spawn downstream of an anabranching reachwhere bedload sediment transport affects the availability of interstitial habitat and limits theeffectiveness of restorative measures (McAdam et al., 2005; NHC, 2012). Overall, the presenceand maintenance of high quality spawning habitat within each of these rivers is determined bythe dynamics linking flow and sediment conveyance through the spawning area.1.4 Restoring white sturgeon habitatSturgeon habitat restoration typically attempts to re-create or remediate natural spawninghabitat in regulated river systems (Dumont et al., 2011; Gendron et al., 2002; LaHaye et al.,1992). The most commonly employed technique is to increase the availability of interstitialvoids for egg and larval incubation by adding coarse sediment overtop the non-functional habitat(NWSRI, 2012; Crossman and Hildebrand, 2014; Dumont et al., 2011; Gendron et al., 2002;LaHaye et al., 1992; Trencia and Collin, 2006). Considering both white sturgeon and lakesturgeon habitat rehabilitation projects, the grain size of placed substrate has varied widely(0.02-1.5 m) and spawning area design has included the construction of mid-channel shoals,outer bank ridges, riffles, boulder cells and circular pads (Kerr et al., 2011).White sturgeon habitat restoration has been conducted on the Columbia, Kootenay andNechako Rivers, where sedimentation of the native substrate had occurred in response to flowregulation (DFO, 2014). In the Columbia river, a 1,000 m2 area of substrate was added withinthe thalweg at a known spawning location to increase complexity and availability of interstitialhabitat (Crossman and Hildebrand, 2014). The restored habitat, consisting of 90% large cob-bles and boulders to resist displacement and 10% coarse to very coarse gravel, proved successfulin providing refuge habitat for larval fish. Restoration on the Nechako system has consistedof adding 2,100 m2 of spawning substrate in two locations within a critical spawning reach(NWSRI, 2012), the details of which are presented in Section 2.3. Recent efforts on the Koote-nay system have adopted an ecosystem-based approach intended to restore fluvial dynamics andthe ecological functionality of three interconnected reaches (KTOI, 2009). Restorative measures4included bank-stabilization through grading and reestablishment of riparian vegetation, increas-ing floodplain and side-channel connectivity and the installation of instream structures to createscour pools, route bedload sediment and generate hydraulic complexity (KTOI, 2009).Restoration projects have often resulted in initial spawning success followed by long-termfailure attributed to progressive sedimentation (NHC, 2012; Kerr et al., 2011; Johnson et al.,2006; Gendron et al., 2002). To prolong the functionality of restored habitat, instream worksmust be designed in accordance with the fluvial and sedimentological dynamics of the spawningreach while remaining within the habitat requirements of the species. Pre-project hypothesistesting (Wheaton et al., 2004a; Wheaton et al., 2004b) and post-project monitoring are keyin understanding spawning site selection and in determining the sustainability of instreamworks. While mechanical or hydraulic remediation can be used to maintain the quality ofrestored habitat, designs that require minimal maintenance, for example periodic substrateaugmentation, are clearly more desirable.1.5 Numerical modelling and its role in sturgeon habitatrestorationFlow and sediment transport models are extensively applied to understand fluvial processes, linkfluvial processes with stream ecology and evaluate channel restoration designs (e.g. Pasternacket al., 2004; Biron et al., 2012; McDonald et al., 2010). While numerous hydrodynamic modelsare currently available, selecting an appropriate one depends on the purpose of modelling, thespatial scale of analysis and the resolution of input and output data (Brown and Pasternack,2009; Gard, 2009; Lane et al., 1999; Shen and Diplas, 2008).Depth-averaged, 2-dimensional (2D) models can be an effective means to simulate meso-scalehydrodynamics and evaluate restoration design based on sediment entrainment, flow complexityand habitat suitability (Crowder and Diplas, 2000; Shen and Diplas, 2008; Pasternack et al.,2006; Wheaton et al., 2004b). However, 2D model performance can be negatively affected byDigital Elevation Model (DEM) interpolation, mesh refinement and the presence of submergedinstream obstacles (Crowder and Diplas, 2000; Pasternack et al., 2004; Shen and Diplas, 2008).Surveyed point density and DEM inaccuracy have been found to be primary factors causingerror in predicted depth, velocity and shear velocity; however model output may still be withinthe range of measurement error in the field (Pasternack et al., 2006). Achieving representativehydrodynamics is critical in testing restoration design because error in simulated depth andvelocity propagates directly into habitat suitability metrics (Boavida et al., 2013).Given that local velocity refugia provided by submerged instream structures are often in-tegral to habitat restoration design, fully 3-dimensional simulations may be more appropriatedespite their greater cost, computational demand and required data collection. These modelshave been successful in reproducing complex hydrodynamics around instream structures andrepresent a more accurate treatment of bed shear stress (Lane et al., 1999; Shen and Diplas,2008; Biron et al., 2012). However, the predictive ability of complex 3D models remains lim-5ited by methodological issues arising from sensitivity to initial hydraulic conditions, channelcurvature and bed topography (Wheaton et al., 2004a; Lane et al., 1999). Acknowledging thatpredictive accuracy is fundamentally limited by the complexity of a fluvial system, the betterprocess representation achieved by 3D models can increase their utility in hydraulically complexareas, especially if restoration designs are being compared relative to each other (Lane et al.,1999; Shen and Diplas, 2008; Wheaton et al., 2004a).Hydrodynamic modelling has been used on the Nechako, Columbia and Kootenay systemswithin the context of sturgeon habitat use and restoration. On the Nechako River, a 2D model(RIVER2D) of the 6-km spawning reach predicted secondary channel flow conveyance andcross-sectional velocity fairly well, with discrepancies attributed to the resolution of topographicinput data and possible effects of submerged vegetation (NHC, 2008). Simulations revealed thatas discharge increases, a greater proportion of the flow becomes conveyed through secondarychannels and that the two highest velocity areas within the reach potentially correspond tohistorical sturgeon spawning sites.In the Kootenay River, a 2D flow and sediment transport model (FaSTMECH) was used todetermine hydraulic cues for spawning site selection, assess how flow regulation may affect thesecues and to determine the effects of pre- and post-dam flow conditions on substrate conditionwithin the 18-km spawning reach downstream of the Libby dam (McDonald et al., 2010).Results from the modelling suggest a strong spatial correlation between spawning location andarea of maximum cross-sectional depth and velocity and revealed that pre-regulation flowscould have maintained suitable substrate habitat at these locations through strong sedimentsorting and vertical scour. Modelling was also used in the Kootenay system to evaluate aproposed channel restoration design intended to convey high flows, transport available sediment,reduce bank erosion, improve flood plain connectivity and provide greater depth and velocityfor spawning fish (Logan et al., 2011). Simulations proved useful in identifying several designflaws, including the inability to mobilize sediment within the channel and the failure to achieveflood plain connectivity.On the Columbia system, three-dimensional modelling (COCIRM) was used to simulatecomplex hydrodynamics at a tributary confluence used by spawning sturgeon (Golder, 2006;Fissel and Jiang, 2008). The model domain incorporated a dam spillway into the upstreamboundary and was able to simulate standing waves reasonably well within the supercriticaloutflow region (Fissel and Jiang, 2008). Modelling revealed the presence of a highly dynamiccirculation pattern at the channel confluence, where the presence of up to three gyres circulatingin opposite directions dominate the eddy depending on the discharge ratio between channels(Golder, 2006; Fissel and Jiang, 2008). The effects of flow regulation on the local hydrodynamicsare likely to have caused geomorphic change within the spawning area which contributed torecruitment failure (McAdam, 2015).61.6 Study rationaleThe research presented herein was conducted within the framework of the Nechako WhiteSturgeon Recovery Initiative as a continuation of geomorphic investigations (NHC, 2014; NHC,2015). The goal of this study was to advance our understanding of fluvial dynamics withinthe critical spawning reach, based on the premise that an increased understanding of geomor-phic processes will improve future habitat restoration design. Specifically, this study sought tocharacterize the spatial and temporal pattern of sediment transport through the reach and toidentify the drivers of geomorphic change affecting sturgeon habitat. To accomplish this, an in-tensive sampling program was conducted throughout the 2015 flood hydrograph and the datasetwas analyzed in conjunction with 2D hydrodynamic modelling to supplement the interpretationof results.7Chapter 2Study Site2.1 DescriptionThe Nechako River at Vanderhoof, BC has been identified as a critical spawning reach forNechako white sturgeon. The river at this location has an anabranching morphology (type-3channel using classification from Nanson and Knighton (1996)) that flows through a complex ofstable, densely vegetated islands (Figure 2.1). Prior to the onset of flow regulation in the early1950’s, the low-lying islands and bars were largely devoid of vegetation suggesting the periodicmobility of sediment at high flow.Figure 2.1: White sturgeon spawning reach on the Nechako River near Vanderhoof, BC.The location of the spawning reach corresponds to a transition where the river is gravel-8bedded upstream and sand-bedded downstream (NHC, 2006). This section of river is a de-positional zone situated on a gradient break with an upstream channel slope of 0.06% anddownstream slope of 0.03% (NHC, 2013). Within the study area, substrate generally fines withdownstream distance starting from a pebble, gravel and imbricated cobble bed at the upstreamextent to a bed composed largely of sand, granules and infilled gravel at the downstream extent(NHC, 2012; NHC, 2014).Regarding the glacial legacy of the region, the Nechako River occupies a large meltwaterchannel valley produced during the Pleistocene glaciation (Rood, 1999). During deglaciation,remnant ice impounded large glacial lakes in the area that deposited a thick surficial layerof fine glaciolacustrine sediment (Plouffe and Levson, 2001). The river has since incised intothis deposit, resulting in terrace scarps that rise 30 m above the current floodplain elevationalong the outside of meander bends (NHC, 2006). The two major sediment sources upstreamof the study site are actively eroding terrace scarps and incising tributaries and the dominantsize-class of the sediment input is sand or finer (Rood, 1999).Overall, the Nechako River is relatively gravel-poor. Historically, this was because of thelarge proportion of fine sediment within the glaciolacustrine deposit, the lake headed nature ofthe fluvial system and the presence of gradient breaks which act as depositional zones along themainstem and tributary channels (Rood, 1999; NHC, 2006). Since flow regulation, however, theamount of gravel input to the channel has further decreased due to vegetation encroachmentand bank stabilization (Rood and Neill, 1987; NHC and McAdam, 2003a).2.2 HydrologyFlow regulation on the Nechako River began in 1952 with the construction of the Kenney Damand flow diversion tunnel to the Kemano Generating Station near Kitimat, BC. Historically,the natural hydrograph of the Nechako River was driven by spring snowmelt on the leewardside of the Coast Range and the Interior Plateau (NHC and McAdam, 2003a). Peak annualflow typically occurred in June, with the receding limb of the annual hydrograph periodicallyre-supplied by large frontal rainstorms during the latter portion of summer and into fall (NHCand McAdam, 2003a). Spring flows exceeding 1000 m3/s at Vanderhoof were not uncommonand the estimated mean annual peak daily discharge was 658 m3/s (NHC and McAdam, 2003a).The Nechako Reservoir was filled from 1952 to 1956, reducing the mean annual peak dailydischarge to only 233 m3/s (Figure 2.2). Since then, two water management strategies havebeen implemented; the first from 1957 to 1979 and the second from 1980 to present. The meanannual peak daily discharge during these periods has been 426 m3/s and 360 m3/s, respectively,which represents an approximate 45% reduction from historic flows. The timing of peak flow hasalso been changed from June to August because the current management plan was developedto control stream temperature during the sockeye salmon migration. The 2015 peak daily flowof 677 m3/s was the 3rd highest discharge since the onset of flow regulation in 1952.9Figure 2.2: Pre- and post-regulation annual maximum daily discharge.2.3 HistoryIn May 2011, as part of the Nechako White Sturgeon Recovery Initiative (NWSRI), coarsesubstrate was placed at two locations within the spawning reach to increase the availability ofsuitable larval incubation habitat (labelled Middle Spawning Pad and Lower Spawning Pad onFigure 2.1) (NWSRI, 2012). The two spawning pads are approximately 20 cm thick, with 20-30% of the placed material sized between 20-40 mm and 30-50% between 150-200 mm (NHC,2012). The cobble-gravel interstices of the downstream spawning pad, located downstreamof the island complex, began to infill with coarse sand and fine gravel shortly after place-ment (NHC, 2012). Subsequent monitoring has confirmed that the downstream spawning padcontinues to infill due to the immobility of coarse grains and because bedload is consistentlytransported over the pad (NHC, 2016). The upstream spawning pad has remained largelyfree of fine sediment deposition and infilling has only occurred in local areas near a tributaryconfluence and along pad margins (NHC, 2012; NHC, 2016).A series of geomorphic assessments (Rood, 1998; Rood, 1999; NHC and McAdam, 2003a;NHC and McAdam, 2003b) and sediment transport studies (NHC, 2014; NHC, 2015; NHC,2016) have been completed to gain insight into reach-scale sediment dynamics. Biological mon-itoring programs focused on spawning behavior and population dynamics have been ongoingfor over a decade and have included juvenile indexing, spawn monitoring, and telemetry (re-ports available at www.nechakowhitesturgeon.org). Additional work completed in recent yearshas included investigating the use of low-relief bedforms and mechanical remediation to main-tain/restore the quality of substrate habitat. This study represents a continuation of geomorphicresearch intended to increase the knowledge-base supporting future habitat restoration design.10Chapter 3Methods3.1 Bathymetry, topography and water surface elevationA Trimble Real-Time Kinematic (RTK) GPS was used to survey bathymetry, water surfaceelevation, banklines and bar topography within the study reach. A static observation fromGeodetic Control Monument (GCM) 653659, located approximately 8 km southeast of Van-derhoof, and Post-Processing Kinematic (PPK) procedures were used to position the GPSbase-station near the study site. To provide a measure of quality assurance between days, fivepre-and post-survey points were routinely logged on fixed markers located 300-500 m from thebase-station.Channel bathymetry was surveyed from May 12-15th, 2015, during a relatively high dischargeof approximately 525 m3/s. Surveying at high flow allowed access to a large wetted extent withinsecondary channels. To conduct the survey, the RTK GPS was mounted on top of a survey-grade SonarMite echo sounder and configured to take point measurements at equal frequencyintervals. This configuration resulted in simultaneous measurement of bed and water surfaceelevation.Bathymetric data were collected systematically from upstream to downstream with a typicaltransect spacing of approximately 30 m (Figure 3.1). The maximum transect spacing was 80m and occurred downstream of the Burrard Ave. Bridge. Two longitudinal profiles of bedand water surface elevation were also measured on May 14th, one collected along the southernmainstem channel and the other collected along the northern-most secondary channel.Top-of-bank and bottom-of-bank banklines, as well as bar surfaces and island topographywere surveyed during low-flow (64-118 m3/s) from August 25th to September 6th, 2015. Survey-ing was done with an irregular point spacing as determined by breaks in the natural topography.Island topography was surveyed somewhat opportunistically due to the high level of radio andsatellite signal interference caused by the tree canopy. Consequently, surveyed point density onsome vegetated islands remains relatively sparse.11Figure 3.1: Surveyed bathymetry, bar contours and bankline topography.The base-station coordinates initially obtained from the GCM 653659 baseline were verifiedagainst Precise Point Positioning (PPP) reports generated for the four longest duration base-station observations (CSRS-PPP service offered by the Geodetic Survey Division of NaturalResources Canada) (Appendix A). This comparison showed that post-processing of the surveydata was warranted and as a result, the base-station coordinates were shifted 0.359 m southand 0.786 m east to match the Northing and Easting obtained from the PPP results averagedover 33 hours of data logging. The PPP results were considered more accurate than the GCMbaseline location because GCM 653659 had not been recently maintained. The GCM baselineand PPP results were in good agreement regarding the elevation of the base-station, but forconsistency, it too was corrected to match the PPP results by decreasing it 0.007 m. For furtherdetail about the survey methodology, see Appendix A.3.2 Flow velocity and dischargeA Teledyne RDI RiverRay Acoustic Doppler Current Profiler (ADCP) was used to collectvelocity profiles and discharge estimates across nine transects distributed throughout the reach(Figure 3.2). Velocity measurements were taken from May 12-15th, 2015, at a discharge ofapproximately 525 m3/s. To collect the data, an RTK GPS receiver was mounted to the ADCPraft and set to transmit real-time position to the ADCP software via Bluetooth. This makes it12possible to detect whether moving bed conditions are present by comparing the GPS locationto the ADCP location referenced to the Bottom Track (BT). The ADCP raft was tethered tothe side of a motorized boat operating at a slow and constant speed. Average boat speed duringdata collection was 0.40 m/s (SD = 0.17 m/s). Each transect was repeated a minimum of fourtimes. If the estimated discharge from consecutive transect passes differed by over 5%, the passwould be flagged as an outlier, discarded and repeated. Only 3 out of 41 passes were flaggedas outliers. The percent difference between consecutive discharge estimates averaged 1.6% (SD= 1.2%), with a maximum error of 4.2%. Four compass calibrations were performed and errorvalues ranged between 0.6◦ and 1.8◦.Figure 3.2: ADCP velocity transects.The ADCP data were post-processed using the Teledyne RDI WinRiver II software. Mag-netic variation was set at 17.3◦ for all transects. This value was determined by aligning theGPS ship track with the ADCP BT referenced track along transects that were highly unlikelyto have moving bed conditions. Discharge and velocity were referenced to the ADCP BT for alltransects except TRA and TRB, which had sufficient bed mobility to offset the two ship tracks(Figure 3.3). For these two transects, discharge and velocity was referenced to the NationalMarine Electronics Association (NMEA) GGA sentence from the GPS receiver. All transectswere then cropped to exclude poor quality data nearest the banks. The post-processed datafor each of the nine transects was exported, depth-averaged and binned into 5-m cross-channeldistance intervals. Further information about the post-processing is presented in Appendix B.13Figure 3.3: Moving bed at TRA causing offset between GPS sentences (green and bluelines) and Bottom Track positioning (red line).3.3 Bedload sediment transportBedload transport rates were sampled at 10-m intervals across a total of 12 transects (Figure3.4). Two of the sampling transects, US and LP, were previously established from ongoingsampling programs (NHC, 2015; NHC, 2016). Sampling began two days after ice-off on March22nd and ended once flows receded below 45 m3/s on October 17th, 2015. A total of 36 days weresampled throughout the annual hydrograph and samples were collected during flows rangingfrom 44 m3/s to 656 m3/s (Figure 3.5). On average, sampling was conducted once every 38m3/s change in discharge. No samples could be collected from June 3rd to June 28th becausethe access to the river was restricted by the municipality due to flooding risk.14Figure 3.4: Bedload sampling locations in 2015.Samples were collected using a Helley-Smith bedload sampler with a 76.2 mm wide openingand 0.125 mm mesh bag. A larger Elwha River Sampler with a 203.2 mm wide opening was usedduring the period of peak flow when discharge exceeded 600 m3/s to reduce under-samplingof coarse bedload grains (Vericat et al., 2006). Each sample was collected over a duration of300 s, unless high transport rates caused overfilling of the sampler. In this case, two 150-second samples were collected. Additional details about the sampling protocol are provided inAppendix C.All bedload samples were dried and individually weighed. This allowed for a unit transportrate, or transport rate per meter width, to be calculated at each location by dividing the mass ofeach sample by the sampling duration and the width of the opening on the sampler. Measuredtransport rates were then used to estimate the total transport rate for each 10-m segment ofcross-channel distance as well as the total transport rate for each sampling transect.Samples collected on 13 dates were additionally sieved using phi sieves. Sieving was done toobtain the composition of the bedload sediment, to determine whether the grain size distributionchanged as flow increased from 62 m3/s to 656 m3/s, to identify any downstream trends inbedload composition and to see if coarse gravel became mobile during high flow. Due to timeconstraints, all samples from a transect were combined prior to sieving; results from the sievingtherefore represent the mean cross-sectional grain size distribution. All sieved samples wereeither from the US or LP transects.15Figure 3.5: 2015 hydrograph with bedload sampling and surveying dates.16Chapter 4Results4.1 Bathymetry and water surface elevationA 1-m resolution DEM of the study area was created in a geographic information system usingthe bathymetry, topography and bankline data (Figure 4.1). The DEM clearly defines the thal-weg along the southern mainstem channel with deeper pools located at channel constrictions,along bends and in areas of flow convergence. North of the mainstem, a complex of secondarychannels flows through the anabranching reach. Secondary channels within the northern por-tion of the island complex are seen to bifurcate at high angles, reaching 90 degrees in somelocations, due to the vegetated, cohesive nature of the banks.Water surface elevation (WSE) was interpolated and used to map depth throughout thereach for the time it was surveyed, which was from May 12-15th at a flow of about 525 m3/s(Appendix D). The mean and maximum depths within the mainstem channel were 3.07 m (SD= 0.78 m) and 6.32 m, respectively, while the mean and maximum depths within secondarychannels were 2.22 m (SD = 0.40 m) and 3.87 m. The deepest area within the entire reach waslocated downstream of the anabranching reach in the mainstem channel. Secondary channelshad a narrow distribution of depths, with 50% of the total secondary channel area being 2.0-2.5m deep. In comparison, the most frequent range of depths within the mainstem was 2.5-3.0m, but covered only 26% of the total mainstem area. The total areas covered by secondarychannels and the mainstem channel were approximately 0.28 km2 and 0.55 km2, respectively.The longitudinal profile of bed and water surface elevations collected along the mainstemchannel indicates that flow was non-uniform within the reach during high discharge (Figure 4.2).At the upstream extent of the reach, the steep water surface slope corresponds to high velocityflow through a relatively narrow channel width. Roughly 1 km downstream, the hydraulicgradient begins to decrease, indicative of backwatered flow conditions. Backwatering appearsto be controlled by the Burrard Ave. Bridge because water surface slope increases once againdownstream of the bridge. At a discharge of about 525 m3/s, the backwater is seen to extendapproximately 1.5 km upstream from the bridge.17Figure 4.1: Digital elevation model of the spawning reach.18Figure 4.2: Bed and WSE profile along the mainstem channel during a flow of approxi-mately 525 m3/s.4.2 Velocity and flow conveyanceThe cross-channel velocity profile taken across the upstream transect TRA (Figure 3.2) duringa discharge of about 525 m3/s shows high-velocity flow through a relatively simple, parabolic-shaped channel (Figure 4.3) (see Appendix B for all velocity profiles). The mean cross-sectionalflow velocity at this location was the highest within the study area, reaching 2.17 m/s. Meanvelocity then decreased downstream to a minimum of 0.57 m/s at transect TRH, located atthe downstream extent of the island complex approximately 675 m upstream from the BurrardAve. Bridge (Table 4.1). At this location, the channel shape is roughly rectangular and highervelocity flow is concentrated within a 40-m wide section beginning about 30 m from the leftbank (Figure 4.3). Approximately 200 m upstream from the bridge, mean velocity increasesonce again to 1.01 m/s at transect TRI. This location corresponds to the upstream portion ofthe Lower spawning pad (Figure 2.1). The cross-sectional profile at this transect is relativelycomplex with slower, deeper flow through the thalweg and higher velocity flow concentratedalong the inside of the meander about 40 m from the right bank (Figure 4.3). The downstreamspawning pad is located within the thalweg, about 40 m from the left bank.Bed mobility was detected at Transect TRA and Transect TRB due to the offset betweenGPS and Bottom Track (BT) referenced ship tracks (Figure 3.3). At the upstream transectTRA, the bed was mobile between about 10-20 m from the left bank and the greatest mobility19occurred near the 20-m mark. High sediment mobility at this location was also observed duringbedload sampling (US) on May 19th, seven days after the ADCP data were collected. On thatday, the highest cross-sectional transport rate of 138.62 g/s/m was sampled directly at the 20-mmark; a rate corresponding to 58.4% of the total cross-sectional sediment transport on that day.At transect TRB, the GPS and BT ship tracks became offset between 15-25 m from the leftbank. When bedload transport was sampled across this channel (MU-A), two days before andsix days after the velocity profiles were taken, the highest transport rates also occurred at a 20-m distance from the left bank. The sampled rates at this location on both days were 44.8 g/s/mand 19.6 g/s/m, respectively, representing 67.9% and 48.6% of the total cross-sectional bedloadtransport for each date. These findings suggests that bedload sediment is laterally concentratedinto lanes of higher sediment transport, rather than being evenly distributed across the channel.Around the time of data collection (500-550 m3/s), the amount of flow was not proportionalto the amount of bedload being conveyed through the different channels. On May 18th, 93% ofthe channel-wide bedload sediment was conveyed by 39% of the total flow through secondarychannel MU-A (see Figure 3.4) (bedload transect MU-A corresponds to ADCP transect TRBin Table 4.1). Further downstream, on May 5th, the sediment transport rate across transectML-A was only 0.3 g/s compared to 34.8 g/s at transect ML-B. The difference in bedloadconveyance between both transects occurred despite both channels conveying almost equalamounts of flow (bedload transects ML-A and ML-B correspond to ADCP transects TRG andTRH in Table 4.1). The disproportionate amount of sediment compared to flow being conveyedthrough different channels further supports that preferential pathways of sediment transportexist within the reach, both at the cross-sectional and reach-planform scales.Overall, a wide range of flow velocities were present within the reach during the time of datacollection. The decreasing trend in mean flow velocity from upstream to downstream (Figure4.4) is consistent with the longitudinal profile of water surface slope (Figure 4.2); TRA is in thehigh gradient section at the upstream extent, TRH is within the backwatered area of the islandcomplex and TRI is just upstream of the bridge where the hydraulic gradient increases onceagain. It is interesting that results from the ADCP provided insight into bedload conveyanceas well, as bed mobility was detected within narrow cross-channel widths.20Table 4.1: Proportional flow conveyance and mean velocity across 9 ADCP transects(data collected May 12-15th, 2015 at a discharge of approximately 525 m3/s)..Transect Percent of Total Discharge Mean Velocity (m/s)TRA 100 2.17TRB 39 1.56TRC 61 1.74TRD 19 1.02TRE 12 1.02TRF 31 1.00TRG 38 0.10TRH 39 0.57TRI 100 1.0121Figure 4.3: Cross-channel velocity profiles at transects TRA (top), TRH (middle) andTRI (bottom) during a discharge of approximately 525 m3/s between May 12-15th,2015.22Figure 4.4: Depth-averaged velocity across transects TRA, TRH and TRI during a dis-charge of approximately 525 m3/s between May 12-15th, 2015 (whiskers indicatestandard deviation within each cross-channel bin).234.3 Bedload sediment composition and transport ratesThe bedload being transported throughout the spawning reach was almost entirely composedof sand finer than 2 mm. The grain size distribution of the bedload showed no relation withincreasing discharge or transport rate and did not vary significantly between sampling locations.The coarsest D84 grain size sampled within the entire reach reached only 3 mm, and the D84 atthe US transect did not exceed 2 mm even when flow velocities were over 2.00 m/s during highflow (Figure 4.5). In fact, only 3.4% and 6.7% of the total sampled mass at the US transectwas coarser than 8 mm gravel during the two highest flows sampled; 597 m3/s on May 28th and656 m3/s on June 2nd. Very coarse gravel in the 45-64 mm range constituted only 1.4% and4.4% of the total sampled mass on these respective dates.Figure 4.5: D84, D50 and D16 grain sizes of bedload sediment.The rate of bedload transport at the US transect was positively correlated with discharge(Figure 4.6). However, this relation was non-linear and transport rates increased markedlyonce discharge exceeded about 400 m3/s. Using this threshold to split the data into two linearrelations, bedload transport increased with a slope of 1.48 below 400 m3/s (R2 = 0.39) and aslope of 8.97 above 400 m3/s (R2 = 0.79). The mean cross-sectional transport rate was 225.6g/s (SD = 203.1 g/s) for flows between 80-400 m3/s compared to 1,753.5 g/s (SD = 724.1g/s) for flows of 400-656 m3/s. The maximum cross-sectional transport rate past this samplingtransect reached 2,599.4 g/s during a discharge of 656 m3/s, on June 2nd, 2015. This datenearly corresponds to the timing of peak annual flow, that reached 677 m3/s on June 6th. Afterthis date, hysteresis was observed in the bedload-discharge relation. The lower transport rates24during the falling limb of the hydrograph suggest that the supply of bedload sediment upstreamof the US transect became limited during the period of peak flow.Figure 4.6: Relation between bedload transport and discharge at progressively down-stream sampling locations (Note: only regressions at the US and MU transects aresignificant).25Figure 4.7: Sampled bedload transport rates throughout the 2015 hydrograph.In contrast, the rate of bedload transport past the LP transect was not related to discharge.The maximum cross-sectional transport rate was only 1,384.8 g/s, or roughly half of the maxi-mum transport rate past the US transect. Peak sediment transport across transect LP occurredon August 31st at a relatively low discharge of 81 m3/s during the tailing end of the recedinghydrograph limb. Cross-sectional transport rates were also relatively constant compared tothe rates sampled upstream. The mean and standard deviation of bedload transport at theLP transect was 261.2 g/s and 288.1 g/s, compared to the US transect where the mean andstandard deviation was 807.7 g/s and 865.4 g/s. Overall, the relation between discharge andbedload transport weakened with downstream distance (Figure 4.6) and the timing of maximum26sediment transport occurred progressively later throughout the year (Figure 4.7).As part of the ongoing Nechako Sediment Transport Investigations, framed within theNechako White Sturgeon Recovery Initiative, an analysis was done to quantify the sedimentloads moving through the reach. The details of the analysis are not presented as part of thisthesis, but are provided for completeness in Appendix E and are summarized below.A bedload-discharge rating curve was used to derive the annual sediment load transportedinto the study area past the US transect; predicted and observed bedload rates were in goodagreement. The incoming sediment load was then compared to the amount of bedload beingtransported out of the reach past the LP transect. This output sediment load was estimated byinterpolating daily transport rates between sampled dates since transport was uncorrelated withdischarge. The predicted bedload transport rate across the US transect reached a maximum of190 m3/day during peak flow, exceeding the maximum transport rate across the LP transect of75 m3/day for 61 consecutive days between April 25th and June 24th, 2015. The total annualloads for the upstream and downstream locations were estimated at 9,250 m3 and 3,050 m3,suggesting net deposition of over 6,000 m3 of sediment within the reach.4.4 Patterns of sediment transport through the study reachBedload sediment was primarily transported downstream through the spawning reach by a sub-set of active secondary channels (Appendix F). As bedload entered the reach, a large portionof it was routed into the first secondary channel immediately downstream of the US transect,labelled MU-A in Figure 4.8. Transport rates within this channel ranged from 12.7% to 111.6%of the US cross-sectional transport rate while the amount of sediment transported in the main-stem channel MU-C was only 1.2% to 19.3% (Table 4.2) 1. Throughout the monitoring period,secondary channel MU-A transported on average 8 times more sediment than the mainstemchannel. Roughly 280 m downstream, the amount of bedload transported in channel MU-Dwas 97% and 224% of the upstream cross-sectional transport rate at MU-A on April 26th andMay 31st, respectively. For these same dates, the transport rate in channel MU-B which leadsback into the mainstem channel was 37% and 16% of the upstream transport rate at MU-A.Further downstream within the anabranching section, bedload was primarily conveyed bychannel ML-B through the middle of the island complex (Figure 4.8). The amount of sedimentbeing transported through channel ML-B increased with discharge, as well as over time, with thetwo maximum rates occurring during the peak of the hydrograph and during the receding limb(Table 4.3). Compared to channel ML-B, transport rates were very low within the mainstemchannel ML-A and within the northern secondary channel ML-C. The two highest transportrates in channel ML-A were sampled during the rising and falling limbs of the flood hydrographand transport dropped to zero during the period of peak flow.1If transects were not sampled on the same date, values were obtained by interpolating transport rates betweensampled days.27Figure 4.8: Mean bedload transport rate through different channels (sampled between400-700 m3/s).Table 4.2: Bedload tranport through secondary channel MU-A and mainstem channelMU-C in the upstream portion of the study reach..Date Discharge (m3/s) MU-A % of US MU-C % of US3/24/2015 115 104.0 8.73/31/2015 223 60.6 13.54/10/2015 280 111.6 19.34/26/2015 488 45.6 3.85/18/2015 552 17.4 1.25/31/2015 636 13.3 5.46/30/2015 473 12.7 3.728Table 4.3: Bedload tranport through mainstem channel ML-A compared to secondarychannels ML-B and ML-C within the island complex..Discharge (m3/s) Hydrograph Limb ML-A (g/s) ML-B (g/s) ML-C (g/s)100 - 200 Rising 24.5 No Data No Data200 - 300 Rising 8.6 64.4 No Data400 - 500 Rising 0.3 34.9 2.3500 - 600 Rising No Data 101.9 0.0600 - 700 Rising 0.0 743.2 0.0500 - 400 Falling 5.1 33.6 3.4300 - 200 Falling 9.0 470.7 2.9The location of highest sediment transport past the US and LP transects remained spatiallyconsistent over time (Figure 4.9). At the US transect, where the total channel width is approx-imately 100 m, bedload was typically transported within 50 m of the left bank and the highesttransport rates were located 15-35 m from the bank. This narrow 20-m width had the highesttransport rates of the entire cross-section on 16 of the 21 sampled days. The eleven highesttransport rates sampled across the US transect in 2015 were located within this 20-m widthand ranged from 32.6 g/s/m to 181.3 g/s/m. At the LP transect, the highest transport rateon 21 of 24 sampled days was located 15- 65 m from the left bank. The total channel widthat this location is roughly 150 m. The six highest transport rates across this transect weresampled 15-35 m from the left bank and ranged from 15.2 g/s/m to 134.1 g/s/m. Four of thesepeak transport rates were sampled between August 31st and October 17th when discharge wasbetween 44 m3/s and 80 m3/s. The consistency in cross-channel location with high sedimenttransport corroborates results from the ADCP (described in Section 4.2) stating that bedloadis laterally concentrated within discrete lanes at the US transect. In addition, the sampling re-sults described above confirm that preferential pathways of sediment transport also exist at theLP transect, a spatial dynamic which was undetected by the ADCP due to the lower intensitytransport rates.Overall, bedload was conveyed through the reach in a relatively consistent pattern. Sedimentis primarily transported through the island complex by a subset of active secondary channelswith comparatively minimal bedload transported along the mainstem channel. At the US andLP sampling transects, where the channel is single-thread, bedload is conveyed within a 50-mportion of the total channel width and highest transport occurs within a 20-m subsection nearthe left bank.29Figure 4.9: Cross-channel bedload transport rates sampled in 2015 at the US and LPtransects, distances provided relative to a fixed location on the left bank.30Chapter 5Modelling5.1 InitializationThe Nays2DH hydrodynamic model, accessed through the International River Interface Coop-erative (iRIC) platform, was used to conduct all simulations. Details about the 2-dimensionalsolver are not presented herein and can be found elsewhere (www.i-ric.org). Nays2DH wasselected over alternative models due to its numerical stability, flexibility in setting the initialwater surface profile and because the effect of vegetation can be introduced separately fromMannings roughness as an additional drag force. For further detail regarding the model data,configurations and results presented below, see Appendix G.Topography was input to the model using the reach-scale DEM (Appendix D) coarsened toa resolution of 2 m. The modelling mesh, composed of 5 x 5 m grid cells, was created froma polygonal channel center-line and specified domain width of 900 m. Simulations were runfor 10,000 seconds and the time-step was adjusted between 0.10-0.15 seconds depending onthe simulated discharge to achieve model stability. Solution results were output at 10 secondintervals and the last 1,000 seconds were averaged to generate the final simulation result.Discharge and water surface elevation at the downstream boundary were held constant foreach simulation. Simulations were run for every 50 m3/s discharge increment between 75 m3/sand 775 m3/s, with an additional low flow simulation of 45 m3/s. Preliminary runs usinga uniform flow calculation to set the downstream boundary condition revealed that flow isnon-uniform downstream of the bridge during moderate to high discharges. Consequently, astage-discharge rating curve was developed to specify WSE at the downstream boundary. Thiswas done by iteratively adjusting the input WSE until good agreement was achieved betweenmodelled output and measured stage at Water Survey of Canada (WSC) Gauge 08JC001 (Fig-ure 5.1). This WSC gauge is located approximately 1 km upstream of the model boundary.Including the entire range of simulated flows, the mean absolute error of modelled WSE at thegauge location was 0.066 m.Manning’s roughness coefficient (n) was varied during model calibration to achieve agree-31ment between simulated and observed WSE profiles. Calibration resulted in a low Manning’sroughness value of 0.0215 assigned to the channel, with a slightly higher roughness of 0.024assigned to a localized area of flow convergence to reduce instability at high flow. Bar contours,bottom-of-bank and top-of-bank shapefiles were imported to the model and assigned additionaldrag to account for vegetation. Bar tops and bank slopes were assigned a vegetation densityof 0.1 stems/m2 and all overbank areas were assigned a vegetation density of 2.0 stems/m2.These values were determined by comparing simulation output with water surface elevation andvelocity data.Figure 5.1: Measured versus modelled stage at WSC gauge location (simulations wererun using a rating curve to specify the downstream WSE boundary condition).The grain size distribution imported to the model was obtained by photo-sieving a seriesof 30 underwater images taken across the US, MU-A, MU-B, MU-D, M-A and LP bedloadsampling transects (Figure 3.4). Image resolution was 1.76 pixels per 1.0 mm and the surfacearea captured in each image was about 627 cm2. The photo-sieving code used a WolmanPebble Count approach where 100 grains within the image were digitally measured at griddedintervals. The finest size class used for classification was sand and all grains finer than 2mm were included within the sand fraction. The grain size distribution of the bed surface wasdetermined for each photo and combined by transect to generate averaged cross-channel results.The averaged grain size distributions were used to create five polygons, corresponding to thesubstrate characteristics of channels US, MU-A, MU-B/MU-D, M-A and LP. The grain sizedistribution showed a trend of downstream fining from the US transect (D50 = 36 mm, 12%sand) to the LP transect (D50 = 8 mm, 28% sand).32For all simulations, the model was run with a fixed bed and the output flow parameterswere used to calculate the sediment transport capacity for each grid cell. Transport capacitywas calculated using the Wilcock and Crowe (2003) transport function. This surface-basedsediment transport model is defined by Eqs.5.1, 5.2 and 5.3:W ∗i =0.002Φ7.5 for Φ < 1.3514 (1− 0.894Φ0.5)4.5for Φ ≥ 1.35(5.1)τ∗rm = 0.021 + 0.015 exp[−20Fs] (5.2)τriτrm=(DiDsm)b(5.3)where,b =0.671 + exp(1.5− DiDsm) (5.4)τ∗rm =τrm(s− 1)ρgDsm (5.5)W ∗i is the dimensionless transport rate of size fraction i, Φ = τ/τri, τ is the shear stress, τriis the reference shear stress of size fraction i, τ∗rm is the reference dimensionless Shields stressfor the mean size of the bed surface, τrm is the reference shear stress of the mean size of thebed surface, Fs is the proportion of sand in the surface size distribution, Di is the grain sizeof fraction i, Dsm is the mean grain size of the bed surface, s is the ratio of sediment to waterdensity, ρ is water density and g is gravitational acceleration.For the calculation, each cell within the model domain was attributed a grain size distri-bution by interpolating between the six locations where underwater images had been collectedand photo-sieved (Appendix G).5.2 Calibration and validationThe low channel roughness (n = 0.0215) calibrated to a discharge of approximately 525 m3/sachieved good agreement between simulated and observed WSE profiles in both the southernmainstem channel and the northern-most secondary channel (Figure 5.2). Minimum, maximumand mean absolute error in the mainstem channel and secondary channel was -0.033 m, 0.082m and 0.025 m, and -0.061 m, 0.100 m and 0.015 m, respectively. Such a low roughnessvalue suggests that the wetted channel boundary is hydraulically smooth, lacking bedformsand vegetation. This seems possible given the D50 of the substrate within the study reach istypically immobile medium to coarse gravel (8-36 mm) infilled with 12-28% sand. While noclear spatial trend in the error was observed within the secondary channel, a downstream trend33does exist within the mainstem channel where simulated WSE around the island complex isslightly higher than measured elevations.Figure 5.2: Measured (May 12-15th, 2015) and simulated WSE profiles for a discharge of525 m3/s after calibration of channel roughness.The model was validated against velocity and depth data collected across nine transects(Figure 3.2). The mean absolute error in simulated velocities across each transect ranged from0.09 m/s to 0.19 m/s (Table 5.1). The most accurate simulation results were obtained within themainstem channel, while poorest results occurred within relatively narrow secondary channels.34Mean absolute percent error was lowest at transects TRA and TRC where flow velocity wasthe highest (Table 4.1). Regarding modelled depth, mean absolute error was less than 0.10 mfor all transects except TRA, but the mean absolute percent error at this location was still lessthan 5%. Overall, modelled patterns of cross-channel velocity (Figure 5.3) and depth were inclose agreement with measured data given the complexity of channel. Data from all 9 ADCPtransects are compared to simulated results in Appendix G.Table 5.1: Simulated velocity and depth compared to ADCP data collected May 12-15thduring a discharge of approximately 525 m3/s.Transect Velocity MAE (m/s) Velocity MAE (%) Depth MAE (m) Depth MAE (%)TRA 0.14 6.7 0.13 4.6TRB 0.17 15.5 0.04 2.2TRC 0.14 8.7 0.06 2.6TRD 0.17 28.0 0.07 2.8TRE 0.19 17.8 0.08 4.1TRF 0.19 19.5 0.07 3.3TRG 0.11 11.7 0.09 3.1TRH 0.09 17.0 0.10 3.6TRI 0.10 11.5 0.09 3.035Figure 5.3: Measured (May 12-15th, 2015) versus simulated cross-channel velocity for adischarge of 525 m3/s.36Additional data collected in 2006-07 by NHC (NHC, 2006; NHC, 2008) were used to evaluatemodel performance for discharges of 78 m3/s, 460 m3/s and 800 m3/s. At 78 m3/s, modelledWSE was generally high with a mean absolute error of 0.107 m (Figure 5.4). However, thiserror may be due to changes in channel topography which have occurred since the data werecollected about 10 years ago. Results from the 460 m3/s simulation are closer to observed valueswith a mean absolute error of 0.038 m and the 800 m3/s simulation can be considered in generalagreement with the data, given that the data represent estimates of the high-water mark fromthe 2007 freshet (NHC, 2008). Overall, these comparisons confirm the validity of the modellingapproach and are likely to overestimate simulation error due to the effects of potential changesin channel bathymetry.Figure 5.4: Simulated versus measured water surface profiles (data collected by NHC in2006-07).5.3 Simulation resultsDischarge and velocity show a positive relation within the mainstem channel at the upstreamextent of the reach and downstream of the bridge. However, they are negatively related withinthe downstream portion of the island complex (Figure 5.5). At the upstream extent of the37reach, velocity reached a maximum of 2.54 m/s during a simulated discharge of 775 m3/s and aminimum of 0.62 m/s during a 45 m3/s simulation. Comparatively, velocity reached a maximumof 1.35 m/s within the downstream portion of the island complex during low flow simulationof 45 m3/s and a minimum of 0.37 m/s during a high flow of 775 m3/s. Velocity varied withthe greatest magnitude within the upstream reach, changing by 1.92 m/s over the range offlows compared to more moderate variations of 0.67 m/s, 0.98 m/s and 0.76 m/s for threeprogressively downstream locations.The negative relation between discharge and velocity within the island complex agreeswith measured WSE data indicating a non-uniform flow profile upstream of the Burrard Ave.Bridge during moderate to high flows (Figure 4.2). The reversal from a negative to a positiverelation downstream of the bridge (Figure 5.5) further supports that the bridge is controllingthe backwater within the reach. Model results suggest that backwatering begins at a dischargeof approximately 225-275 m3/s and that a clear non-uniform flow profile develops once dischargeexceeds 325 m3/s (Figure 5.6).Figure 5.5: Relation between discharge and velocity within the mainstem channel (plot-ted by 100 m3/s discharge intervals).38Figure 5.6: Simulated WSE profiles along the mainstem channel showing the develop-ment of backwater upstream of the Burrard Ave. Bridge.The maximum shear stress within the reach increased with discharge, reaching 22.6 N/m2at 775 m3/s. Maximum shear stress for all simulated flows varied only within a narrow range,between 22.6 N/m2 and 16.2 N/m2 (Table 5.2). Similarly, mean shear stress within the main-stem channel increased with discharge but only from 3.2 N/m2 to 4.4 N/m2. The location ofmaximum shear stress shifts as discharge increases from mid-reach channel constrictions andmeanders to the upstream, mainstem channel (Figure 5.7).Similar to the pattern of maximum shear stress, the area of highest transport capacity shiftsupstream as discharge increases from 45 m3/s to 375 m3/s (Figure 5.8). The spatial patternof high transport remains constant once discharge exceeds 425 m3/s, although the magnitudeof the transport capacity continues to increase. At a discharge of 75 m3/s, the total cross-sectional transport capacity reaches a maximum of 8.7 kg/s at the downstream extent of theisland complex (Figure 5.9). At this flow, the cross-sectional capacity at US and LP are 0.0kg/s and 2.8 kg/s, respectively. As discharge increases to 375 m3/s, the transport capacitybecomes similar at US and LP, with rates of 2.1 kg/s and 1.2 kg/s, while capacity within theisland complex drops to 0.0 kg/s. For a discharge of 775 m3/s, the transport rate at US, withinthe island complex and at LP are 5.7 kg/s, 0.0 kg/s and 3.2 kg/s, respectively.39Table 5.2: Simulated shear stresses within the reach with increasing discharge (meanshear stress calculated using the mainstem channel)..Discharge (m3/s) Max Shear Stress (N/m2) Mean Shear Stress (N/m2)45 16.9 3.275 19.3 3.5175 20.9 3.9275 19.7 4.1375 16.2 4.0475 18.8 4.0575 20.2 4.1675 21.3 4.2775 22.6 4.4Sampled bedload transport rates were less than the estimated transport capacity at both theUS and LP transects. This overestimation may have been produced because the modelled shearstress used to calculate the capacity represented the total shear stress, rather than only the skindrag, which is the proportion of the total shear stress responsible for grain mobility in sedimenttransport functions. In addition, transport capacity may have been overestimated because grainsize distribution has a strong influence on estimated rates and significant interpolation wasrequired to attribute a substrate composition to the entire model domain (further discussed inSection 6.4). While it is possible that the difference between transport capacity and observedtransport at the US transect could indicate supply limited conditions, this is unlikely to bethe case at the LP transect given the large supply of sand within the island complex locatedimmediately upstream. Despite the discrepancy between observed and predicted transportrates, the ranges of flows producing the highest sampled rates correspond to the ranges of flowswith the highest transport capacity at each location (Figure 5.10).40Figure 5.7: Reach-scale distribution of shear stress with increasing discharge; areas with shear stress less than 1 N/m2 not shown(colored white).41Figure 5.8: Reach-scale sediment transport capacity with increasing discharge.42Figure 5.9: Total cross-channel transport capacity with increasing discharge.43Figure 5.10: Calculated transport capacity and sampled bedload transport at the USand LP transects plotted as a function of discharge.44Chapter 6Discussion6.1 Variation in flow dynamics with dischargeIncreasing discharge produces very different responses in hydraulic conditions depending onlocation within the spawning reach. Within the anabranching portion of the reach, shear stressincreases with discharge from 45 m3/s to approximately 325 m3/s (Appendix G). The relativelyuniform depths and rectangular channel geometries of the secondary channels (Appendix B)suggest that total wetted width increases rapidly once the water surface reaches a thresholdelevation. Most secondary channels across the middle of the island complex have bed elevationsthat are 1.0-1.3 m higher than the thalweg elevation within the mainstem. This initial wettingstage produces an increase in the total surface area exposed to moderate shear stress withoutcausing a major increase in shear stress within the mainstem despite higher flows (Table 5.2).Velocities within the mainstem channel near the island complex even decrease slightly oncesecondary channels become wetted due to the increase in total channel width (Figure 5.5).As discharge begins to exceed 225-275 m3/s, the Burrard Ave. Bridge reduces flow con-veyance enough to cause backwatering. Backwatering decreasing the water surface slope pro-gressively further upstream as discharge increases (Figure 5.6). The near-zero hydraulic gradi-ent reaches the downstream extent of the island complex once flows reach 275 m3/s, causingmainstem velocities to decrease abruptly by 39% from 1.18 m/s to 0.72 m/s (Figure 5.5). Back-watering further extends into the island complex as discharge increases from approximately 275m3/s to 525 m3/s, resulting in decreased shear stresses (Figure 5.7).Velocity (Table 4.1) and shear stress (Figure 5.7) over the downstream spawning pad remainhigher than within the island complex during backwatered conditions. Higher velocities aremaintained because the local hydraulic gradient increases with discharge immediately upstreamof the bridge as backwatered flow passes through the constriction (Figure 5.6). The spatialdistribution of shear stresses over the spawning pad remains relatively constant once dischargeexceeds 325 m3/s, as does the magnitude of shear stress, varying between 3 N/m2 and 8 N/m2.Shear stress at this location peaks at a flow of 125 m3/s, reaching just over 8 N/m2.45At the upstream extent of the study reach, velocity and shear stress are positively correlatedwith discharge. At this location, flow through the single-thread, parabolic-shaped channel isnearly uniform over a wide range of flows. Flow is described as nearly uniform because thehydraulic gradient through this section increases locally once discharge exceeds about 475 m3/s,an effect caused by the relatively narrow channel slightly constricting high flows. This area islocated far enough upstream from the bridge constriction that the water surface profile remainsunaffected by backwater, even during the highest simulated discharge of 775 m3/s.Results from the hydrodynamic model suggest that flow downstream of the bridge is alsonon-uniform during periods of high discharge. When simulating a discharge of 525 m3/s withuniform flow as the downstream boundary condition, the simulated WSE is 1.22 m lower thanthe measured elevation at the downstream extent of the model domain. Under these same con-ditions, the simulated WSE is 1.00 m lower than measured values at the WSC gauge locatedimmediately downstream of the bridge. Though the model does not extend far enough down-stream to simulate the origin of backwatering, a large meander downstream of the study areais suspected to be the cause. The uniform flow boundary condition is only valid for dischargesbelow 125 m3/s, confirmed by accurate prediction of WSE at the WSC gauge location.Backwatering that occurs downstream of the Burrard Ave. Bridge raises the question ofwhere fish may have spawned and what the hydrodynamics of the spawning reach were priorto bridge construction. Currently, as the fish swim toward the spawning reach during a highflow year, the first higher velocity zone they encounter is immediately downstream of the bridgewhere the local hydraulic gradient increases as backwater from upstream of the bridge flowsthrough the constriction. Prior to bridge construction, however, this local increase in hydraulicgradient may have been less pronounced and the backwater originating downstream of the bridgecould have extended upstream into the spawning reach. Under these conditions, combined withhistoric high flows, the first high velocity zone encountered by spawning fish would have beenfarther upstream within the spawning reach. Suitable egg and larval habitat may have beenavailable at this upstream location, which likely had a coarse substrate and multiple gravel barsthat were largely devoid of vegetation prior to flow regulation.6.2 Characterization of bedload transport within thespawning reachSpatial variation in the relation between discharge, hydraulic conditions and sediment availabil-ity influences bedload transport rates from upstream to downstream. The positive relationshipbetween bedload transport and discharge at the US transect (Figure 4.6) suggests that oc-casional high flow years input a pulse of sediment into the spawning reach while moderateflow years input relatively minimal sediment; a finding that is consistent with previous data(NHC, 2014; NHC, 2015). The magnitude and timing of peak sediment transport into thereach corresponds with peak flow (Figure 4.7), which concurrently corresponds to the maxi-mum backwatered extent and minimum transport capacity within the island complex (Figure465.9). The increased sediment availability and decreased transport capacity mid-reach producestransport-limited conditions and results in sediment deposition, primarily within secondarychannels that convey the most bedload (Figure 4.8). As the hydrograph recedes and velocity(Figure 5.5), shear stress (Figure 5.7) and transport capacity (Figure 5.8) increase within thedownstream portion of the spawning reach, the sediment previously stored within secondarychannels is transported downstream at relatively constant, moderate rates.It is interesting that most of bedload transported into the reach from upstream is immedi-ately routed into the first northern secondary channel, labelled MU-A in Figure 4.8. While thepath of bedload transport is influenced by channel shape and may be directed along topographiclows (Habersack et al., 2008), the bed elevation at the entrance of the secondary channel is 0.3m to 0.4 m higher than the thalweg elevation. Therefore, at this location, enough force isexerted on the bed to transport sediment along an upwards local channel slope of 0.9% intothe smaller channel. Once the sediment has entered this channel it becomes further dividedbetween secondary channels but remains largely within the northern half of the island complex(Figure 6.1).The preferential routing of sediment into secondary channel MU-A is likely caused by thepresence of strong secondary flow circulation near the bed. This type of flow circulation isintensified by upstream channel curvature and is responsible for generating the lateral sedimentsorting observed around meanders. In addition, given that flow velocity past the secondarychannel entrance is high, the observed sediment routing may occur due to a hydraulic effectproduced at channel bifurcations where most of the flow entering a secondary channel comesfrom the near-bed region (Bulle, 1926). This effect is a result of the vertical velocity distributionof the flow, where low velocity flow in the near-bed region has less inertia than high velocitysurface flow, causing a disproportionate amount of near-bed flow, and consequently sediment,to enter the secondary channel (Vasquez, 2005). Application of a three-dimensional flow andsediment transport model would be useful to fully resolve the local hydrodynamic patternsdriving the flow-sediment separation at this location.In addition to spatial components, a temporal component influences bedload transport rateswithin and downstream of the island complex. Sediment transport rates were high within themain secondary channel ML-B at the peak of the annual hydrograph (600-700 m3/s) (Table4.3), despite the low transport capacity at this location during high flows. This likely reflectsthe time needed for the sediment that was transported into the reach, past the US transect, totravel downstream to the ML-B transect.Sediment transport within the ML-B channel then decreased as flows receded to 400-500m3/s and upstream sediment input decreased. However, a second period of high transportwas observed later in the year on August 4th, during a discharge of 218 m3/s. This secondperiod of high sediment transport likely occurred as sediment deposits were re-mobilized by thelocally increasing transport capacity associated with receding flows. The temporal lag betweenmid-reach deposition during high flow and sediment re-mobilization during low flow is reflected47by progressively increasing transport rates downstream of the island complex throughout thetailing end of the receding hydrograph (Figure 4.7).Hysteresis in the discharge-bedload transport relation at the US transect (Figure 4.6) sug-gests the availability of bedload sediment within the active width of the channel became limitedduring the rising limb of the hydrograph. This may have occurred because the amount of sed-iment input from bank erosion during high flow was limited and confined to localized erosionof terrace scarps due to post-regulation vegetation encroachment and bank stabilization (Roodand Neill, 1987; NHC and McAdam, 2003a). Alternatively, the sediment that had been inputto the Nechako River by its tributaries during the spring freshet may have become depleted.Given the prolonged period of competent flow, sediment previously stored within the channelupstream of the spawning reach may have been transported considerable distances betweendepositional areas characterized by marked reductions in hydraulic gradient. This is supportedby the composition of the bedload because coarse sand can be easily entrained and transportedin saltation at high velocity, especially over a channel bed that has become armoured in responseto flow regulation (Church, 1995). It is plausible that the supply of readily available sedimentwithin the 35-45 km segment of river between the spawning reach and the next upstreamdepositional area (NHC, 2013) began to deplete after approximately 90 days of rising discharge(300 m3/s exceeded 60% of the time). Data collected in 2014, when daily maximum dischargereached only 325 m3/s, did not show any hysteresis and transport rates generally plotted alongthe rising limb trend of the 2015 hydrograph (NHC, 2016).The grain size of the bedload did not significantly coarsen with increasing discharge ortransport rate (Figure 4.5) and contained very little gravel large enough (>8 mm) to providesuitable larval habitat. The lack of a rapid increase in bedload grain size suggests that the bedarmor around the US transect did not become mobile during peak flow. The only fully mobilesize fraction (Wilcock and McArdell, 1993; Church and Hassan, 2002) at the US transect duringhigh flow was 2 mm sand. The ratio between the proportion of 2 mm sand within the bedloadto its proportion on the bed surface was over 7.5 for flows exceeding 597 m3/s, indicating thatupstream sediment sources are important in supplying bedload material. This ratio dropped tobelow 0.5 for the 2.8 mm size class and ranged between approximately 0.3 and 0.1 for all coarsergrain size intervals. During peak flow in 2015, the mean shear velocity within the upstreamarea of the reach was sufficient to partially suspend grains finer than 1 mm, as indicated by aRouse number 2 of 1.6, 1.9 and 2.8 for sediment sized 0.5 mm, 1.0 mm and 2.0 mm, respectively.Bedload transport at this location can therefore be characterized as sand overpassing a relativelycoarse, largely static bed surface.2The Rouse number is a non-dimensional number indicating the mode of sediment transport. It is expressedas the ratio of particle settling velocity to shear velocity multiplied by the von Karman constant.48Figure 6.1: Schematic representation of bedload sediment transport through the spawning reach (arrow width represents magnitudeof transport).49The mobility of coarse grains is limited by the low overall magnitude of shear stress withinthe reach. During the peak 2015 flow of 675 m3/s, hydrodynamic simulations indicate that thehighest shear stresses were 14-19 N/m2. The spatially averaged mean shear stress within theupstream high velocity area was 12 N/m2. Varying the Shields parameter from 0.06 to 0.03,the maximum mobile grain size would have been 14-39 mm, with a mobile grain size of 18-25mm using the commonly applied Shields value of 0.047 for gravel. This traditional approachsuggests that flow was not competent to mobilize the D50 (36 mm) grain size of the bed surfaceat the upstream location, which would have required a threshold shear stress of 27 N/m2. Byaccounting for the reduction in critical shear stress associated with a 12% sand content ofthe substrate (Wilcock and Crowe, 2003), the D50 grain size would have been mobilized at ashear stress of 9 N/m2. However, the higher estimate obtained using the traditional Shieldsapproach is considered a better representation of the system because sand is overpassing acoarse, structured bed, rather than constituting a significant proportion of the surficial grainsize distribution.The morphology of the spawning reach, containing numerous mid-channel bars and low-elevation islands, reflects the depositional legacy of the area. Even prior to flow regulation,it was a threshold-type channel (using classification from Church, 2006) with infrequent, low-intensity sediment transport due to its very mild channel gradient and rapid expansion in totalchannel width. Although the amount of coarse substrate within the area has always beenrelatively limited due to low shear stress and the high proportion of fines contained withinthe surrounding glaciolacustrine sediment, gravel transport and deposition was likely moreactive during the pre-regulation era due to more extensive bank erosion upstream, increasedmobilization of tributary fans and greater overall stream power within the Upper Nechakosystem. In addition, increased flow conveyance and floodplain storage prior to the constructionof the Burrard Ave. Bridge may have maintained higher velocity flow within the mainstemchannel. Thus, larval habitat within the spawning reach may have historically (i.e. prior tobridge construction and flow regulation) been maintained by freshet flows having sufficientstream power to mobilize the surface of unvegetated gravel deposits and spatially segregatecoarse and fine sediment in flow convergence zones and around meander bends.6.3 Implications for larval habitat and restorationBackwatering and sediment deposition during high flow is discordant with the conceptual modelof freshet-spawning sturgeon utilizing deep, high-velocity habitat over coarse heterogenous sub-strate devoid of fine sediment. Clearly, the functional relation between fluvial ecology andgeomorphology has been altered, evidenced by recruitment failure since 1967 (McAdam et al.,2005). The reduction in stream power from the regulated flow regime is problematic for restora-tion objectives because it significantly limits the maximum mobile grain size. Even during arelatively high flow of 775 m3/s, moderate shear stresses throughout the spawning reach sug-gest gravel substrates are rarely mobilized to release infilled fines. While the addition of coarse50substrate within the spawning area has been used as a restorative measure in the past (NWSRI,2012), selecting an appropriate grain size and location to place the substrate is limited by thereduced competence of the river and by the active channel width conveying sand as bedload.The development of backwater upstream of the Burrard Ave. Bridge poses another chal-lenge to habitat restoration because flows exceeding 275-325 m3/s do not significantly increaseshear stress or transport capacity over the downstream spawning pad. Historically, increasedfloodplain storage and greater side-channel conveyance past the current bridge location mayhave maintained higher flow velocity and shear stress within the mainstem channel during largerpre-regulation flows. However, additional modelling is needed to test this hypothesis. Underthe current dynamics, increasing discharge shifts areas with moderate transport capacity pro-gressively upstream except for in the area immediately upstream of the bridge where hydraulicgradient increases with backwater. This upstream shift probably occurred historically as welldue to backwater development downstream of the bridge location, but backwatering may haveaffected less of the spawning reach since the control was further downstream. Either way, themid-reach deposition observed during high flow in 2015 is an issue because it becomes a signifi-cant source of sediment that supplies the relatively constant, moderate transport rates infillingthe substrate at the downstream spawning location.The current state of the Nechako system is the outcome of channel adjustment to nearly 65years of imposed flow and sediment regimes. It is unlikely that historical fluvial dynamics canbe restored on a large scale due to community flooding risk and to the underlying geomorphicchange that has already occurred. Consequently, habitat restoration within the spawning reachis constrained by the present fluvial context. An effective restoration strategy will need toaccount for, and work with the present day sediment transport processes within the reach.Site selection may benefit from the fact that a few secondary channels convey most ofthe bedload trough the anabranching reach with very little sediment being transported in thedeeper mainstem channel. Areas that have a high flow to bedload conveyance ratio indicatelocations that may be able to maintain a coarse substrate with minimal infilling. Consistency inthe location of the active width within the mainstem channel also provides guidance regardinglocations that are, or are not suitable for restorative measures like gravel addition. Substraterestoration may also benefit from the wide range of hydraulic conditions available within thereach during a given discharge and from the inversing relation between discharge and bedloadtransport from upstream to downstream. An appropriate design discharge may therefore beused to site restoration in locations that combine appropriate local hydraulics with a spatialavoidance of bedload transport.6.4 LimitationsSampling bedload transport must ideally account for the inherent variability of sediment trans-port rates that fluctuate through space and time (Gomez, 1991; Habersack et al., 2008). Theintention of this study was to characterize geomorphic processes over the largest spatial and51temporal scales possible. Consequently, the number of sample replicates and the length of timeduring which each sample was collected were constrained. In this regard, the accuracy of bed-load data presented herein is limited by relatively short sampling durations of 300 seconds perlocation. However, this was considered acceptable given the high sampling frequency through-out the study period. The selected sampling schedule generated robust data as indicated by thestrong relation at the US transect location (Figure 4.6) and the agreement between predictedand observed rates (NHC, 2016).A second limitation with collecting bedload data is sampling bias. Helley-Smith samplerscan over- or under-sample different size fractions depending on the ratio between the size of themaximum mobile grain and the opening of the sampler. Sampling efficiency greatly decreasesfor ratios above 0.1 (Sterling and Church, 2002) to 0.2 (Emmett, 1980). The largest threeclasts collected in 2015 were 45 mm, sampled at the US transect during peak flow. To reducebias during this period (Vericat et al., 2006), a sampler with a 203.2 mm opening was used,translating to a ratio of 0.22. More commonly, however, the maximum mobile grain size withinthe reach was 11.2 mm and was collected with a 76.2 mm opening sampler corresponding toa ratio of 0.15. This suggests that coarse bedload was likely underrepresented in the samplesand that sediment finer than medium sand may have been overrepresented due to the collectionof suspended material (Sterling and Church, 2002). Additional bias may have occurred due totilting and perching of the sampler on coarse substrate (Vericat et al., 2006), especially whilesampling on the downstream spawning pad.Numerical modelling requires the explicit specification of several, often unknown, parametersthat can contribute to errors or biases. For example, channel roughness affects modelled shearstress both directly through its calculation and indirectly through its influence on flow velocity(Lane et al., 1999). In this study, the lack of recent WSE data collected over different flowsprevented any assessment of how the influence of boundary roughness changes with discharge.Consequently, a single roughness value calibrated to the water surface at 525 m3/s was used forall simulations. Although this method unrealistically assumes roughness does not spatially andtemporally change with discharge, it does increase the comparability of model output betweendifferent flows and reduces the effect that varying roughness has on estimated shear stress. Useof a single roughness value was also considered acceptable due to the hydraulic smoothness ofthe channel and to the low grain size to depth ratio, or relative roughness, for most flows.Additional error can result from the spatial interpolation required to generate input dataover the entire model domain (Pasternack et al., 2006). To produce the reach-scale DEM, inter-pretation and manual digitization of the thalweg was required in several locations due to channelcomplexity and spacing interval between bathymetry transects. Significant interpolation wasalso necessary to specify the grain size distribution and sand content of the bed surface using30 underwater images collected at six locations. The photo-sieving method itself was limitedby water turbidity, image resolution (1.76 pixels per 1.0 mm) and the area of substrate coveredper image (627.3 cm2).52Other limitations in this study stem from the difficulty in capturing the influence of bedformson local hydrodynamics and sediment transport rates. Bedform migration limits the accuracyof bedload sampling because of the temporal variation in sediment transport that occurs asripples, sand sheets and dunes travel downstream. The development of bedforms is also an issuefor modelling because it introduces greater form drag within the channel that could result inslower than average flows in some locations, and faster than average flows in others. Lastly, theinfluence of bedforms on transport capacity is not captured by sediment transport functions.Given the large proportion of sand within the island complex, these limitation may be ofparticular relevance within secondary channels and within the downstream portion of the studyreach.Finally, the analysis related to sediment transport capacity was limited by several keyfactors. Firstly, the modelled shear stress that was used to calculate the capacity represented thetotal shear stress, rather than the skin drag. The skin drag is the proportion of the total shearstress responsible for grain mobility, which is the proportion that should be applied in sedimenttransport functions. Secondly, the sparseness of data on substrate composition throughout thereach significantly limited the accuracy of capacity estimates because the Wilcock and Crowe(2003) transport function is strongly influenced by the grain size distribution of the bed surface.The estimated transport capacity in this study is intended as a reference to identify downstreamtrends, rather than as a predictive value, due to the combination of these limiting factors.53Chapter 7ConclusionFlow regulation has altered the fluvial processes that link flow and sediment transport to eco-logical integrity within the Nechako River. Geomorphic change within a critical white sturgeonspawning reach has decreased the quality and availability of early rearing habitat. Efforts torestore the habitat continue to be negatively impacted by progressive sedimentation of the re-stored spawning substrate. This study, conducted within the framework of the Nechako WhiteSturgeon Recovery Initiative, was intended to advance our understanding of reach-scale fluvialdynamics to contribute to the knowledge base supporting future habitat restoration design.High flows in 2015 presented an opportunity to sample bedload sediment transport overthe course of the flood hydrograph. Results from bedload sampling indicate that the rateat which sediment was transported into the reach past the upstream-most sampling transectwas positively correlated with discharge. This relation was non-linear and transport ratesincreased rapidly once flows exceeded about 400 m3/s. Data collected at this location alsoshowed hysteresis in the transport rates, suggesting the availability of bedload sediment withinthe channel upstream of the US transect became limited during the period of high flow.The relation between discharge and bedload transport weakened with downstream distanceuntil no relation was observed at the LP transect. The timing of maximum sediment transportpast each sampling location also had a downstream trend, where peak sediment transportoccurred progressively later throughout the year from upstream to downstream. Maximumsediment transport past the LP transect occurred during a discharge of 81 m3/s at the tailingend of the receding hydrograph limb in late August. However, the maximum transport rateswithin the downstream portion of the reach remained low and relatively constant compared tothose sampled upstream.Sediment was primarily transported through the island complex by a subset of secondarychannels with only a minimal amount of bedload transported by the mainstem channel. Theseactive secondary channels conveyed disproportionately large amounts of sediment compared toflow. Upstream and downstream of the anastomosed reach, where the channel has a single-thread morphology, the cross-channel location having the highest transport rates remainedspatially consistent throughout the year. Bedload was conveyed past the US and LP transects54within a 50-m portion of the total channel width and the highest transport rates generallyoccurred within a 20-m subsection near the left bank.Results from hydrodynamic modelling used to supplement data analysis indicate that flowthrough the spawning reach becomes non-uniform once discharge exceeds 225-275 m3/s. Thenon-uniform water surface profile develops upstream of the Burrard Ave. Bridge because thebridge sufficiently constricts the channel to reduce conveyance of moderate to high flows. Ve-locity, shear stress and transport capacity within the downstream portion of the spawning reachdo not increase with discharge due to this backwater effect and to the rapid expansion in to-tal channel width that occurs once secondary channels become wetted. Rather, the areas ofmaximum shear stress and transport capacity shift from mid-reach to upstream locations withincreasing discharge. Over the range of simulated flows, however, the magnitude of maximumshear stress within the reach remained below 23 N/m2.This study identified several challenges for successful habitat restoration posed by the cur-rent fluvial dynamics within the spawning reach. Firstly, the development of backwater isproblematic because high flows do not increase velocity or shear stress within the downstreamportion of the reach. Given that discharge and sediment transport are positively correlatedupstream, high flow years including 2015 can input a large amount of sediment that becomesdeposited mid-reach due to the drop in shear stress and transport capacity caused by the back-water. The deposited sediment is then available to be moved at a constant, relatively moderatebedload transport rate over the downstream spawning pad. The low magnitude of shear stresseswithin the reach is also problematic because it does not get high enough to move coarse particlesand coarse grain mobility is a key mechanism needed to release infilled fines from interstitialvoids. Low stream competence is an issue for restorative measures like gravel addition becauseit constrains the size of placed substrate to a relatively narrow range and makes the substrateprone to infilling.Although it is unlikely that historical fluvial dynamics can be restored on a large scale, itmay be possible to locally improve the quality of sturgeon habitat within the Nechako criticalspawning reach by optimizing restoration design based on advantageous fluvial and sedimen-tological dynamics. Site selection may benefit from the disproportionate amount of flow andsediment conveyed through different channels and from the spatially consistent locations withhighest transport. Site selection may also take advantage of the wide range of hydraulic con-ditions available within the reach during a given flow and of the contrasting relations betweenbedload transport, velocity and discharge from upstream to downstream. Successful restora-tion of larval habitat will likely result from siting locations that avoid preferential pathwaysof sediment transport while maintaining appropriate local hydrodynamics during the designdischarge.55BibliographyAshmore, P. Channel morphology and bed load pulses in braided, gravel-bed streams.Geografiska Annaler. Series A. Physical Geography, pages 37–52, 1991. → pages 2Ashmore, P. and Parker, G. Confluence scour in coarse braided streams. Water ResourcesResearch, 19(2):392, 1983. → pages 3Ashmore, P., Bertoldi, W., and Tobias Gardner, J. Active width of gravel-bed braided rivers.Earth Surface Processes and Landforms, 36(11):1510–1521, 2011. → pages 3, 4Bennett, W., Edmondson, G., Williamson, K., and Gelley, J. An investigation of the substratepreference of white sturgeon (Acipenser transmontanus) eleutheroembryos. Journal ofApplied Ichthyology, 23(5):539–542, 2007. → pages 2Bertoldi, W., Zanoni, L., and Tubino, M. Assessment of morphological changes induced byflow and flood pulses in a gravel bed braided river: The Tagliamento River (Italy).Geomorphology, 114(3):348–360, 2010. → pages 4Beschta, R. L. and Jackson, W. L. The intrusion of fine sediments into a stable gravel bed.Journal of the Fisheries Board of Canada, 36(2):204–210, 1979. → pages 3Best, J. L. Sediment transport and bed morphology at river channel confluences.Sedimentology, 35(3):481–498, 1988. → pages 3Biron, P., Carver, R., and Carre, D. Sediment transport and flow dynamics around a restoredpool in a fish habitat rehabilitation project: Field and 3D numerical modelling experiments.River Research and Applications, 28(7):926–939, 2012. → pages 5Boavida, I., Santos, J., Katopodis, C., Ferreira, M., and Pinheiro, A. Uncertainty inpredicting the fish-response to two-dimensional habitat modeling using field data. RiverResearch and Applications, 29(9):1164–1174, 2013. → pages 5Brown, R. A. and Pasternack, G. B. Comparison of methods for analysing salmon habitatrehabilitation designs for regulated rivers. River Research and Applications, 25(6):745–772,2009. → pages 5Buffington, J. M. and Montgomery, D. R. Effects of sediment supply on surface textures ofgravel-bed rivers. Water Resources Research, 35(11):3523–3530, 1999. → pages 2Bulle, H. Untersuchungen uber die Geschiebeableitung bei der Spaltung von Wasserlaufen.VDI Verlag, Berlin, 1926. → pages 47Church, M. Geomorphic response to river flow regulation: Case studies and time-scales.Regulated Rivers: Research & Management, 11(1):3–22, 1995. → pages 1, 3, 4856Church, M. Bed material transport and the morphology of alluvial river channels. Annu. Rev.Earth Planet. Sci., 34:325–354, 2006. → pages 50Church, M. and Hassan, M. A. Mobility of bed material in Harris Creek. Water ResourcesResearch, 38(11), 2002. → pages 3, 48Coutant, C. C. A riparian habitat hypothesis for successful reproduction of white sturgeon.Reviews in Fisheries Science, 2010. → pages 2Crossman, J. and Hildebrand, L. Evaluation of spawning substrate enhancement for whitesturgeon in a regulated river: Effects on larval retention and dispersal. River Research andApplications, 30(1):1–10, 2014. → pages 1, 4Crowder, D. and Diplas, P. Using two-dimensional hydrodynamic models at scales ofecological importance. Journal of hydrology, 230(3):172–191, 2000. → pages 5, 156DFO. Recovery strategy for white sturgeon (Acipenser transmontanus) in Canada [Final].Species at Risk Act Recovery Strategy Series, pages 1–252, 2014. → pages 1, 4Dietrich, W. E. and Smith, J. D. Bed load transport in a river meander. Water ResourcesResearch, 20(10):1355–1380, 1984. → pages 3Dietrich, W. E. and Whiting, P. Boundary shear stress and sediment transport in rivermeanders of sand and gravel. River meandering, pages 1–50, 1989. → pages 3Dumont, P., DAmours, J., Thibodeau, S., Dubuc, N., Verdon, R., Garceau, S., Bilodeau, P.,Mailhot, Y., and Fortin, R. Effects of the development of a newly created spawning groundin the Des Prairies River (Quebec, Canada) on the reproductive success of lake sturgeon(Acipenser fulvescens). Journal of Applied Ichthyology, 27(2):394–404, 2011. → pages 4Eaton, B. and Lapointe, M. Effects of large floods on sediment transport and reachmorphology in the cobble-bed sainte marguerite river. Geomorphology, 40(3):291–309, 2001.→ pages 2Emmett, W. W. A field calibration of the sediment-trapping characteristics of theHelley-Smith bed-load sampler. Technical report, US Govt. Print. Off. (No. 1139), 1980. →pages 52Fissel, D. B. and Jiang, J. 3D numerical modeling of flows at the confluence of the Columbiaand Pend d’Oreille rivers. In Estuarine and Coastal Modeling: proceedings of the eighthinternational conference, pages 928–941, 2008. → pages 4, 6Gard, M. Comparison of spawning habitat predictions of PHABSIM and RIVER2D models.International Journal of River Basin Management, 7(1):55–71, 2009. → pages 5Gendron, M., Lafrance, P., and Lahaye, M. Assessment of spawning activity on the man-madespawning ground downstream from the Beauharnois Power Dam. Technical report,Environnement Illimit Inc. prepared for Hydro-Qubec, Division Production, 2002. → pages4, 5Gibson, S., Abraham, D., Heath, R., and Schoellhamer, D. Vertical gradational variability offines deposited in a gravel framework. Sedimentology, 56(3):661–676, 2009. → pages 357Golder. White sturgeon spawning at Waneta, 2005 investigations. Technical report, GolderAssociates Ltd. prepared for Teck Cominco Metals and BC Hydro (No. 05-1480-030F),2006. → pages 4, 6Gomez, B. Bedload transport. Earth-Science Reviews, 31(2):89–132, 1991. → pages 3, 51Grant, G. E., Schmidt, J. C., and Lewis, S. L. A geological framework for interpretingdownstream effects of dams on rivers. A peculiar river, pages 203–219, 2003. → pages 1Greig, S., Sear, D., and Carling, P. The impact of fine sediment accumulation on the survivalof incubating salmon progeny: implications for sediment management. Science of the totalenvironment, 344(1):241–258, 2005. → pages 2Habersack, H., Seitz, H., and Laronne, J. B. Spatio-temporal variability of bedload transportrate: Analysis and 2D modelling approach. Geodinamica Acta, 21(1-2):67–79, 2008. →pages 3, 47, 51Hassan, M. A. and Church, M. Experiments on surface structure and partial sedimenttransport on a gravel bed. Water Resources Research, 36(7):1885–1895, 2000. → pages 2Hassan, M. A., Egozi, R., and Parker, G. Experiments on the effect of hydrographcharacteristics on vertical grain sorting in gravel bed rivers. Water Resources Research, 42(9), 2006. → pages 2, 3Hassan, M. A., Smith, B. J., Hogan, D. L., Luzi, D. S., Zimmermann, A. E., and Eaton, B. C.18 Sediment storage and transport in coarse bed streams: scale considerations.Developments in Earth Surface Processes, 11:473–496, 2007. → pages 2Hatfield, T., Coopper, T., and McAdam, S. Scientific information in support of identifyingcritical habitat for SARA listed white sturgeon populations in Canada: Nechako, Columbia,Kootenay and Upper Fraser (2009). Fisheries and Oceans Canada, Science Branch, 2013.→ pages 2Hildebrand, L., McLeod, C., and McKenzie, S. Status and management of white sturgeon inthe Columbia River in British Columbia, Canada: An overview. Journal of AppliedIchthyology, 15(4-5):164–172, 1999. → pages 2Hildebrand, L., Drauch Schreier, A., Lepla, K., McAdam, S., McLellan, J., Parsley, M.,Paragamian, V., and Young, S. Status of white sturgeon (acipenser transmontanusrichardson, 1863) throughout the species range, threats to survival, and prognosis for thefuture. Journal of Applied Ichthyology, 32(S1):261–312, 2016. → pages 1, 2Hoey, T. Temporal variations in bedload transport rates and sediment storage in gravel-bedrivers. Progress in physical geography, 16(3):319–338, 1992. → pages 2Johnson, J., LaPan, S., Klindt, R., and Schiavone, A. Lake sturgeon spawning on artificialhabitat in the St Lawrence River. Journal of Applied Ichthyology, 22(6):465–470, 2006. →pages 1, 5Kerr, S. J., Davison, M. J., and Funnell, E. A review of lake sturgeon habitat requirements andstrategies to protect and enhance sturgeon habitat. Ontario Ministry of Natural Resources,2011. → pages 4, 558Kock, T. J., Congleton, J. L., and Anders, P. J. Effects of sediment cover on survival anddevelopment of white sturgeon embryos. North American Journal of FisheriesManagement, 26(1):134–141, 2006. → pages 2KTOI. Kootenai River habitat restoration project master plan: A conceptual feasibilityanalysis and design framework. Kootenai Tribe of Idaho, Bonners Ferry, ID, 2009. → pages1, 4, 5LaHaye, M., Branchaud, A., Gendron, M., Verdon, R., and Fortin, R. Reproduction, early lifehistory, and characteristics of the spawning grounds of the lake sturgeon (Acipenserfulvescens) in Des Prairies and L’Assomption rivers, near Montreal, Quebec. CanadianJournal of Zoology, 70(9):1681–1689, 1992. → pages 4Lane, S., Bradbrook, K., Richards, K., Biron, P., and Roy, A. The application ofcomputational fluid dynamics to natural river channels: Three-dimensional versustwo-dimensional approaches. Geomorphology, 29(1):1–20, 1999. → pages 5, 6, 52Lapointe, M., Eaton, B., Driscoll, S., and Latulippe, C. Modelling the probability of salmonidegg pocket scour due to floods. Canadian Journal of Fisheries and Aquatic Sciences, 57(6):1120–1130, 2000. → pages 2Lisle, T. E. Sediment transport and resulting deposition in spawning gravels, north coastalCalifornia. Water resources research, 25(6):1303–1319, 1989. → pages 2, 3Lisle, T. E., Nelson, J. M., Pitlick, J., Madej, M. A., and Barkett, B. L. Variability of bedmobility in natural, gravel-bed channels and adjustments to sediment load at local andreach scales. Water Resources Research, 36(12):3743–3755, 2000. → pages 3Logan, B., McDonald, R., Nelson, J., Kinzel, P., and Barton, G. Use of multidimensionalmodeling to evaluate a channel restoration design for the Kootenai River, 2011. → pages 6Lytle, D. A. and Poff, N. L. Adaptation to natural flow regimes. Trends in ecology &evolution, 19(2):94–100, 2004. → pages 1McAdam, D. S. O. Retrospective weight-of-evidence analysis identifies substrate change asthe apparent cause of recruitment failure in the upper Columbia River white sturgeon(Acipenser transmontanus). Canadian Journal of Fisheries and Aquatic Sciences, 72(8):1208–1220, 2015. → pages 2, 4, 6McAdam, S. O. Effects of substrate condition on habitat use and survival by white sturgeon(Acipenser transmontanus) larvae and potential implications for recruitment. CanadianJournal of Fisheries and Aquatic Sciences, 68(5):812–822, 2011. → pages 1, 2McAdam, S. O., Walters, C. J., and Nistor, C. Linkages between white sturgeon recruitmentand altered bed substrates in the Nechako River, Canada. Transactions of the AmericanFisheries Society, 134(6):1448–1456, 2005. → pages 1, 2, 4, 50McDonald, R., Nelson, J., Paragamian, V., and Barton, G. Modeling the effect of flow andsediment transport on white sturgeon spawning habitat in the Kootenai River, Idaho.Journal of Hydraulic Engineering, 136(12):1077–1092, 2010. → pages 3, 4, 5, 6Nanson, G. C. and Knighton, A. D. Anabranching rivers: Their cause, character andclassification. Earth surface processes and landforms, 21(3):217–239, 1996. → pages 859NHC. Preliminary substrate investigation: Nechako River at Vanderhoof. Technical report,NHC Ltd. prepared for BC Ministry of Environment (No. 34203), 2006. → pages 9, 37NHC. Nechako River at Vanderhoof hydrodynamic model upgrade: Final Report. Technicalreport, NHC Ltd. prepared for BC Ministry of Environment (No. 34623), 2008. → pages 6,37NHC. Nechako sturgeon spawning gravel september: 2011 substrate assessment. Technicalreport, NHC Ltd. prepared for Ministry of Forests, Lands and Natural ResourceOperations, 2012. → pages 1, 4, 5, 9, 10NHC. Nechako sturgeon spawning gravel: September 2012 substrate assessment. Technicalreport, NHC Ltd. prepared for Ministry of Forests, Lands and Natural ResourceOperations, 2013. → pages 9, 48NHC. Nechako River 2013 sediment transport investigations. Technical report, NHC Ltd.prepared for Ministry of Forests, Lands and Natural Resource Operations, 2014. → pages 7,9, 10, 46, 101, 151NHC. 2014 sediment transport investigation on the Vanderhoof reach of the Nechako River.Technical report, NHC Ltd. prepared for Ministry of Forests, Lands and Natural ResourceOperations, 2015. → pages 7, 10, 14, 46, 101, 151NHC. 2015 sediment transport investigation on the Vanderhoof reach of the Nechako River.Technical report, NHC Ltd. prepared for Ministry of Forests, Lands and Natural ResourceOperations, 2016. → pages 10, 14, 48, 52, 151NHC and McAdam, S. Nechako River geomorphic assessment Phase I: Historical analysis ofLower Nechako River. Technical report, NHC Ltd. prepared for BC Ministry of Water,Land and Air Protection, 2003a. → pages 9, 10, 48NHC and McAdam, S. Nechako River geomorphic assessment Phase II: Detailed analyses ofpotential white sturgeon habitat sites. Technical report, NHC Ltd. prepared for BCMinistry of Water, Land and Air Protection, 2003b. → pages 10Nicholas, A., Ashworth, P., Kirkby, M., Macklin, M., and Murray, T. Sediment slugs:Large-scale fluctuations in fluvial sediment transport rates and storage volumes. Progress inPhysical Geography, 19(4):500–519, 1995. → pages 2NWSRI. Nechako White Sturgeon Recovery Initiative 2011-2012 Annual Report. Accessedfrom http://nechakowhitesturgeon.org, 2012. → pages 1, 4, 10, 51Paragamian, V. Kootenai River white sturgeon: Synthesis of two decades of research.Endangered Species Research, 17(2):157–167, 2012. → pages 2Paragamian, V., McDonald, R., Nelson, G. J., and Barton, G. Kootenai River velocities,depth, and white sturgeon spawning site selection-a mystery unraveled? Journal of AppliedIchthyology, 25(6):640–646, 2009. → pages 2, 4Parsley, M. J. and Beckman, L. G. White sturgeon spawning and rearing habitat in the lowerColumbia River. North American Journal of Fisheries Management, 14(4):812–827, 1994.→ pages 260Pasternack, G. B., Wang, C. L., and Merz, J. E. Application of a 2D hydrodynamic model todesign of reach-scale spawning gravel replenishment on the Mokelumne River, California.River Research and Applications, 20(2):205–225, 2004. → pages 5, 156Pasternack, G. B., Gilbert, A. T., Wheaton, J. M., and Buckland, E. M. Error propagationfor velocity and shear stress prediction using 2D models for environmental management.Journal of Hydrology, 328(1):227–241, 2006. → pages 5, 52, 156Perrin, C. J., Rempel, L. L., and Rosenau, M. L. White sturgeon spawning habitat in anunregulated river: Fraser River, Canada. Transactions of the American Fisheries Society,132(1):154–165, 2003. → pages 2, 4Plouffe, A. and Levson, V. Late Quaternary glacial and interglacial environments of theNechako River-Cheslatta Lake area, central British Columbia. Canadian Journal of EarthSciences, 38(4):719–731, 2001. → pages 9Powell, D. M. Patterns and processes of sediment sorting in gravel-bed rivers. Progress inPhysical Geography, 22(1):1–32, 1998. → pages 3Rhoads, B. L. and Kenworthy, S. T. Flow structure at an asymmetrical stream confluence.Geomorphology, 11(4):273–293, 1995. → pages 3Rood, K. Nechako River substrate quality and composition. Technical report, NechakoFisheries Conservation Program (No. M89-7), 1998. → pages 10Rood, K. Identification and ranking of sources contributing sediment to the upper NechakoRiver. Technical report, Nechako Fisheries Conservation Program (No. RM89-7), 1999. →pages 9, 10Rood, K. and Neill, C. A study of some aspects of the geomorphology of the Nechako River.Technical report, Reid Crowther and Partners Ltd and NHC Ltd. prepared for Fisheriesand Oceans Canada, 1987. → pages 9, 48Roy, A. and Bergeron, N. Flow and particle paths at a natural river confluence with coarsebed material. Geomorphology, 3(2):99–112, 1990. → pages 3Shen, Y. and Diplas, P. Application of two-and three-dimensional computational fluiddynamics models to complex ecological stream flows. Journal of Hydrology, 348(1):195–214,2008. → pages 5, 6Sterling, S. M. and Church, M. Sediment trapping characteristics of a pit trap and theHelley-Smith sampler in a cobble gravel bed river. Water Resources Research, 38(8), 2002.→ pages 52Trencia, G. and Collin, P.-Y. Rapport d’ame´nagement d’une fraye`re pour le poisson a` laRivie`re Chaudie`re. Ministe`re des Ressources Naturelles et de la Faune, Direction del’ame´nagement de la faune Chaudie`re-Appalaches, Charny, Que´bec, 2006. → pages 4Vasquez, J. Two-dimensional numerical simulation of flow diversions. In 17th CanadianHydrotechnical Conference: Hydrotechnical Engineering: Cornerstone Of A SustainableEnvironment, Edmonton, Alberta, 2005. → pages 4761Vericat, D., Church, M., and Batalla, R. J. Bed load bias: Comparison of measurementsobtained using two (76 and 152 mm) Helley-Smith samplers in a gravel bed river. WaterResources Research, 42(1), 2006. → pages 15, 52Wheaton, J. M., Pasternack, G. B., and Merz, J. E. Spawning habitat rehabilitation-I.Conceptual approach and methods. International Journal of River Basin Management, 2(1):3–20, 2004a. → pages 5, 6Wheaton, J. M., Pasternack, G. B., and Merz, J. E. Spawning habitat rehabilitation-II. Usinghypothesis development and testing in design, Mokelumne River, California, USA.International Journal of River Basin Management, 2(1):21–37, 2004b. → pages 5Wilcock, P. R. and Crowe, J. C. Surface-based transport model for mixed-size sediment.Journal of Hydraulic Engineering, 129(2):120–128, 2003. → pages 33, 50, 53, 136Wilcock, P. R. and McArdell, B. W. Surface-based fractional transport rates: Mobilizationthresholds and partial transport of a sand-gravel sediment. Water Resources Research, 29(4):1297–1312, 1993. → pages 48Winemiller, K. O. Life history strategies, population regulation, and implications for fisheriesmanagement. Canadian Journal of Fisheries and Aquatic Sciences, 62(4):872–885, 2005. →pages 1, 2Zimmermann, A. E. and Lapointe, M. Intergranular flow velocity through salmonid redds:sensitivity to fines infiltration from low intensity sediment transport events. River Researchand Applications, 21(8):865–881, 2005. → pages 262Appendix AReal-Time Kinematic (RTK) Surveyon the Nechako RiverA.1 MethodsThis section provides supplementary information to section 3.1 Bathymetry, topographyand water surface elevation in the thesis; previously described aspects have been omitted.A.1.1 ConfigurationsTable A.1: Configurations used for RTK surveying..Parameter Selected ConfigurationDatum NAD83 (Canadian Spatial Reference System)Version (Epoch) V4.0.0 (2002.0)Geoid HTv2.0Coordinate system UTM 10 NSurvey style RTK Fixed (resolved ambiguity)Base RTK initialization Known Point (Keyed In)Rover RTK initialization On-The-FlyEstablishing control Static Point observation off GCMSNR mask 7Elevation mask 10 degreesLogging interval - Control point 600 second (Control)Logging interval - Topographic point 10 second (TOPO)Logging interval - Survey point 3 second (Fast-Static)A.1.2 Data collectionGeodetic Control Monument (GCM) 653659, accessed through MASCOT government database,was located approximately 8 km southeast of the spawning reach. This benchmark was therefore63used to position a spike on the property of the Nechako White Sturgeon Conservation Center(NWSCI) using a 600-second Control Point logging interval (Figure A.1). This spike remainedfixed in the ground for the duration of the survey, allowing for rapid and precise reinstallationof the base-station (Figure A.2).Channel bathymetry was surveyed between May 12-15th, 2015, a period during which dis-charge increased from 510 m3/s to 533 m3/s. Two longitudinal profiles of bed and water surfaceelevation were collected on May 14th at a discharge of 523 m3/s; one within the mainstem andthe other within a secondary channel to the north. Figure A.3 shows the installation used toconduct the survey, with the RTK GPS received mounted on top of the survey-grade Sonar-Mite echo sounder. Banklines and bar topographies were surveyed on foot from August 25th toSeptember 6th, 2015, during a discharge of 64 m3/s to 118 m3/s (Figure A.4). These surveyswere done during low flow to mesh the topographic and bathymetric data and delineate shape-files for bottom-of-bank (BOB), top-of-bank (TOB) and bar contours, which were submergedat high flow. Overall, 64,594 points were surveyed, consisting of 83 bar elevation points, 3,042bank elevations points and 61,469 bed elevation points.Figure A.1: Establishing local control using GCM 653659, study reach outlined in yellow(Google Earth image).64Figure A.2: RTK base-station receiver and radio positioned over the spike in front of theNWSCI.Figure A.3: Installation used for bathymetric survey, showing the RTK rover receivermounted on top of the survey-grade SonarMite echo sounder.65Figure A.4: Surveying top-of-bank topography.A.1.3 Data processingAnalysis was required to verify and correct the base-station location that was obtained usingthe baseline observation from GCM 653659. To do so, these coordinates were compared to PPPcoordinates obtained using the CSRS-PPP service offered by the Geodetic Survey Division ofNatural Resources Canada (NRCAN, 2013). The CSRS-PPP process requires that raw GNSSdata logged by the base-station be converted to Receiver Independent Exchange (RINEX)Format prior to submission to NRCAN. Once the data is submitted, along with the height ofthe GPS head and the antenna type, the PPP data is processed and a summary report will bereturned to the user containing a computed position, standard deviation and accuracy of thepositioning.PPP reports were generated for the four longest duration base-station observations, whichranged from about 4 hours to over 11 hours (Figure A.5 - Figure A.8). Only long observationswere used because PPP accuracy increases with the duration of the observation, which mustexceed four hours to achieve centimeter-scale accuracy (NRCAN, 2013). Results of the PPPreports are shown and compared to the spike location obtained using the GCM baseline inTable A.2. The survey data were post-processed by shifting the spike coordinates, and hencethe entire survey, by 0.359 m to the south and 0.786 m to the east to match the Northing andEasting obtained from the PPP results averaged over 33 hours of data logging. The PPP resultswere considered more accurate than the GCM baseline location because GCM 653659 had notbeen maintained recently and was located near a highway ditch, leading to the possibility ofit being disturbed during roadworks, snow plowing, etc. The GCM baseline and PPP resultswere in good agreement regarding the elevation of the base-station, but for consistency, it toowas corrected to match the PPP results by decreasing it 0.007 m.66Table A.2: Comparison of control point location obtained using PPP and GCM baselinemethods..Method Date Duration Northing (m) Easting (m) Elevation (m)PPP 5/12/2015 8h 56m 5986529.484 433953.101 639.041PPP 5/13/2015 11h 24m 5986529.484 433953.123 639.012PPP 5/14/2015 4h 10m 5986529.472 433953.105 639.046PPP 5/15/2015 9h 31m 5986529.478 433953.116 639.032PPP Average 5986529.480 433953.111 639.033GCM 5986529.839 433952.325 639.040Difference 0.359 - 0.786 0.00767Figure A.5: PPP report for observation 1 (8 hours and 56 minutes).68Figure A.6: PPP report for observation 2 (11 hours and 24 minutes).69Figure A.7: PPP report for observation 3 (4 hours and 10 minutes).70Figure A.8: PPP report for observation 4 (9 hours and 31 minutes).71A.2 LimitationsOverall, RTK surveying was an efficient technique to collect data on the morphology of thestudy reach. However, operational issues related to signal and communication interferencebecame problematic in vegetated areas. This resulted in low survey point density and limitedaccuracy on islands and along banklines within the upstream portion of the study reach.A.3 ReferencesNatural Resources Canada, Surveyor General Branch. 2013. Guidelines for RTK/RTN GNSSsurveying in Canada (Version 1.1). Accessed from www.nrcan.gc.ca.Natural Resources Canada. 2016. Height reference system modernization. Accessed fromwww.nrcan.gc.ca.Trimble. 2003. Real-time kinematic surveying; Training Guide. (Revision D). Part Number33142-40. Accessed from www.gisresources.com.Trimble. 2004. Trimble 5700/5800 GPS receiver; User Guide. (Revision A), Version 2.21.Accessed from www.ngs.noaa.gov.United States Geological Survey. 2014. Hurricane storm-surge GNSS surveying campaigns;Surveying methods. Accessed from http://water.usgs.gov.72Appendix BVelocity and DischargeMeasurement in the Nechako Riverusing an Acoustic Doppler CurrentProfiler (ADCP)B.1 MethodsThis section provides supplementary information to section 3.2 Flow velocity and dischargein the thesis; previously described aspects have been omitted.B.1.1 Data collectionData were collected from May 12-15th, 2015, during a rising discharge of approximately 510m3/s to 533 m3/s. To collect the data, a Teledyne RDI RiverRay Acoustic Doppler CurrentProfiler (ADCP) raft was tethered to a wooden boom off the side of a motorized boat (FigureB.1). The boat was operated at the slowest and most constant speed possible across all transects(Figure B.2 - Figure B.10). However, it proved difficult to maintain speeds inferior to the flowvelocity along near-bank vegetated areas and across transect TRH, where flow velocity was wellbelow 1.0 m/s over most the transect (Figure B.9).Compass calibrations were performed on each day prior to data collection, except for thefirst day when two calibrations were done. Calibrations were performed within a low-velocitybay located along the right bank in the upstream part of the reach (432125 E, 5986020 N).Compass error for each of these calibrations was 1.2◦, 1.8◦, 0.6◦ and 0.9◦.Each transect began and ended as close to the vegetated bankline as possible. It was thennecessary to estimate the remaining wetted channel distance because significant portions ofbankline and low-lying islands were submerged due to the high-flow conditions at the time73(Figure B.11). These distances were estimated by eye and subsequently input to the ADCPsoftware to estimate overbank discharge.On May 13th, two moving bed tests were performed at near transect TRA within theupstream, high-velocity end of the reach. These tests included one stationary test and onecross-channel loop test; neither test detected a moving bottom.Figure B.1: Teledyne RDI RiverRay ADCP raft tethered to a wooden boom to collectvelocity profiles.Legend for Figures B.2 to B.10;• Red: Water velocity referenced to the Bottom Track• Blue: Water velocity referenced to the GGA GPS string• Green: Water velocity referenced to the VTG GPS string• Orange: Boat velocity referenced to the Bottom Track• Purple: Boat velocity referenced to the GGA GPS string• Black: Boat velocity referenced to the VTG GPS string74Figure B.2: Boat speed versus flow velocity across transect TRA.Figure B.3: Boat speed versus flow velocity across transect TRB.Figure B.4: Boat speed versus flow velocity across transect TRC.75Figure B.5: Boat speed versus flow velocity across transect TRD.Figure B.6: Boat speed versus flow velocity across transect TRE.Figure B.7: Boat speed versus flow velocity across transect TRF.76Figure B.8: Boat speed versus flow velocity across transect TRG.Figure B.9: Boat speed versus flow velocity across transect TRH.Figure B.10: Boat speed versus flow velocity across transect TRI.77Figure B.11: Banklines and low-lying islands were submerged during data collection,making it difficult to estimate wetted channel width.B.1.2 Data processingUsing the Teledyne RDI WinRiver II software, the raw data were first processed by setting themagnetic variation to 17.3◦ for all transects based on alignment of the GPS ship track with theADCP Bottom Track (BT) referenced path at transect TRH (Figure B.12). This transect wasselected because sampled bedload transport rates were very low, suggesting that the bed at thislocation was very unlikely to be mobile. Good agreement between the ship tracks at transectTRC indicates that the magnetic variation applied was appropriate for the upstream extent ofthe reach as well.Discharge and velocity were referenced to the BT for all transects except TRA and TRB,which were referenced to the GGA sentence from the GPS receiver because the bed was suffi-ciently mobile in local areas to offset the two ship tracks (Figure B.13 - Figure B.14). However,bed mobility was relatively minor and had a minimal influence on total estimated discharge,with the BT-referenced and GPS-referenced discharge estimates varying by 1.4% and 2.0% atthe TRA and TRB transects, respectively.To estimate total discharge, WinRiver II was configured to fit a power function to the near-surface and near-bed data to estimate the flux through these regions. No data is availablewithin these regions due to the required blanking distance from the transducer and to bedand side-lobe interference (Mueller and Wagner, 2009). All transects were then cropped to78exclude poor quality data in shallow, low-velocity and vegetated near-bank areas. To estimateflux through the near-shore areas, bank geometry was assumed to have a triangular slope.Additional configurations regarding discharge estimation and data screening are presented inFigure B.15.Each transect was replicated a minimum of four times during data collection. If the percentdifference of estimated discharge between consecutive passes was over 5%, the pass would beflagged as an outlier, discarded and repeated. Only one replicate from each transect is shownin Figure B.16 - Figure B.24. Discharge summary statistics for each transect are presented inFigure B.25.Figure B.12: TRH used to set magnetic variation to 17.3◦ by aligning GPS and BT shiptracks.79Figure B.13: Moving bed offsetting GPS and BT ship tracks at transect TRA.Figure B.14: Moving bed offsetting GPS and BT ship tracks at transect TRB.80Figure B.15: Additional configurations used in WinRiver II for data processing.Figure B.16: Cross-channel velocity profile at transect TRA.81Figure B.17: Cross-channel velocity profile at transect TRB.Figure B.18: Cross-channel velocity profile at transect TRC.82Figure B.19: Cross-channel velocity profile at transect TRD.Figure B.20: Cross-channel velocity profile at transect TRE.83Figure B.21: Cross-channel velocity profile at transect TRF.Figure B.22: Cross-channel velocity profile at transect TRG.84Figure B.23: Cross-channel velocity profile at transect TRH.Figure B.24: Cross-channel velocity profile at transect TRI.85Figure B.25: Summary statistics for ADCP transects.86Data were subsequently exported from WinRiverII and imported into the USGS VelocityMapping Toolbox (VMT), which is a Matlab-based software for ADCP data processing andvisualization (available at https://hydroacoustics.usgs.gov). This software was used to read theASCII output file from WinRiver II and convert it to CSV file formats. To avoid averaging theraw data at this stage, the grid node spacing within the VMT working environment was set tothe typical average bin size of the ADCP data. The specified grid size had a 0.3 m horizontalnode spacing and a 0.1 m vertical node spacing. The smoothing window was set to 1 in bothhorizontal and vertical directions. Once the data were refit to a common grid, depth-averagedvelocities were exported.Files exported from VMT were then imported to the R software environment for furtherdata processing. Firstly, the raw data were filtered to remove any bad points that had beenassigned error values. Then, the depth-averaged data were binned into 5-m distance intervalsacross each transect. These 5-m distance intervals were determined using a straight cross-channel distance, and therefore may include over 5 m of data if the ship track was particularlycurvilinear. Within each bin, the mean coordinates, depth-averaged velocity, depth and specificdischarges value were calculated and exported. The binned depth-averaged velocities, includingthe standard deviation within each bin, are plotted in Figure B.26 - Figure B.34.Figure B.26: Mean and standard deviation of depth-averaged velocity across-transectTRA.87Figure B.27: Mean and standard deviation of depth-averaged velocity across-transectTRB.Figure B.28: Mean and standard deviation of depth-averaged velocity across-transectTRC.88Figure B.29: Mean and standard deviation of depth-averaged velocity across-transectTRD.Figure B.30: Mean and standard deviation of depth-averaged velocity across-transectTRE.89Figure B.31: Mean and standard deviation of depth-averaged velocity across-transectTRF.Figure B.32: Mean and standard deviation of depth-averaged velocity across-transectTRG.90Figure B.33: Mean and standard deviation of depth-averaged velocity across-transectTRH.Figure B.34: Mean and standard deviation of depth-averaged velocity across-transectTRI.91B.2 LimitationsMaintaining a slow and constant boat speed was not trivial during data collection because flowvelocities reached over 2.5 m/s. A jet-boat was used to collect data on May 12-13th, sometimesmaking it difficult to transition from high-velocity flow to near-bank eddies at constant speedsdue to the momentum of the large boat. However, the boat operator was highly proficient andachieved well. The other potential complication when using a jet-boat is magnetic interfer-ence with the ADCP compass (Mueller et al., 2007). This was addressed with regular, dailycompass calibrations. Lastly, the accuracy of total discharge estimates were limited due to thedifficultly in specifying the total wetted width of the channel. Total channel width was difficultto assess because high flow conditions were causing extensive overbank flow. Given that theflow in overbank areas was generally low velocity, due to significant vegetation, this limitationis considered to have had a relatively minimal impact.B.3 ReferencesMueller, D.S., and Wagner, C.R. 2009. Measuring discharge with acoustic Doppler currentprofilers from a moving boat: U.S. Geological Survey Techniques and Methods 3A22, 72 p.Accessed from http://pubs.water.usgs.gov.Mueller, D.S., Wagner, C.R., and Winkler, M.F. 2007. Best practices for measuring dis-charge with Acoustic Doppler Current Profilers. USGS Publications.92Appendix CBedload Sampling ProtocolThe following excerpt is presented with permission from NHC and MFLNRO andwas obtained from:NHC. 2015. 2014 Sediment Transport Investigation on the Vanderhoof Reach of the NechakoRiver. Prepared for Ministry of Forests, Lands and Natural Resource Operations. March 2,2015.C.1 MethodologyThe samples will be collected with a Helley-Smith bedload sampler. A Helley-Smith samplerwith a 76.2 mm wide opening and 0.125 mm mesh bag is to be used to monitor sedimenttransport rates. In general, one sample is to be collected from each vertical over a durationof 300 seconds (5 minutes). If transport rates are exceptionally high and the bag over-fills,additional samples are to be collected for shorter duration periods.• Begin sampling at the first location where the water depth is sufficient for boat access.• Take a photo of the site that shows conditions of the river. Take photos facing upstream,downstream and of each bank.• Navigate the boat as close as possible to the GPS point (within 5 m of the point) anddrop the anchor.• Allow the boat to come to rest on the anchor. The amount of rope required will varydepending on flow but should be around 15 m. Record an estimate of how much rode islet out in the field notes.• Attach a clean sample bag onto the Helley-Smith.• Slowly lower the sampler to the river bed. The tail of the sampler should make contactwith the bed first, followed by the nozzle.93• As soon as the sampler is resting flat on the bed, start the timer.• Collect sediment for 300 seconds (5 minutes).• During sediment collection, it is essential that the cable to the sampler remains slack andthe boat does not pull on the sampler causing it to dredge up material. If this occurs, thesampler should be brought to surface, the sample bag flushed clean, and the collectionstarted again.• As soon as 300 seconds has been reached, the sampler should be raised back to the surface.• Check if the sample has an unexpected amount of sediment in it. This could indicate thatthe sampler nose-dived into the bottom or was dragged along the bed. If this is suspected,flag the sample and collect an additional one. If you are confident this happened, discardthe sample and collect another sample.• Bedload transport is highly variable in space and time so adjacent locations may collectvery different amounts. Make notes on any observed sheets or streams of mobile bedmaterial.• Check that the sample bag is not over-filled (over 40% full of sediment).• Using a squeezable water bottle, wash any sediment that is stuck in the opening or upperparts of the sample bag into the back of the bag.• Carefully transfer the sediment from the sample bag to a Ziploc. Use the water bottleto wash the sediment stuck to the collection bag, and then carefully drain off the excesswater from the Ziploc.• Label the Ziploc with the site and station number, the length of the sample, and the dateand time. If multiple samples were taken for one station, this should also be included.• If the sample bag is more than 40% full, it is likely that the hydraulic efficiency of thesampler has been reduced and a biased sample has been collected. The sample must bediscarded and a new sample collected.• If the bag is only slightly over-filled, attempt to collect two samples of 150 seconds orthree samples for 100 seconds.• If there is a tiny bit or no sediment, the sample from the next vertical can be included.In this case, do not replace the bag but just go to the next location. This is likely onlysuitable when sampling during low river discharge.94Appendix DGeographic Information System(GIS) Data Processing and AnalysisD.1 Data processingA speed of sound correction was applied to all surveyed depths using the temperature of thewater at the time of data collection. The corrected depth value and the distance from thetransducer to the RTK GPS head were subtracted from the RTK elevation to obtain bedelevation. The distance from the transducer to the water surface was added to the correcteddepth to obtain the water surface elevation (WSE). The processed bathymetric data werecombined with the topographic data and imported to ArcGIS. Data from the bankline surveywas imported separately to ArcGIS from AutoCAD as polylines.A triangulated irregular network (TIN) of bed elevation and a TIN of WSE were createdusing the banklines as hardlines and the survey extent as a hardclip. Both TINs were manuallyadjusted by connecting nodes on large triangulated wedges. The bed elevation TIN also requiredthe connection of TIN nodes along the thalweg, especially in areas with low surveyed pointdensity. The TINs were then converted to 1-m resolution rasters with common processingextents. Finally, the rasters were differenced to produce a depth map, which was clipped toexclude vegetated islands and overbank areas. The DEM of bed elevation was coarsened to aresolution of 2 m and exported to be used for hydrodynamic modelling, along with shapefilesof banklines and bar contours.D.2 MapsFigure D.1 presents the TIN of bed elevation that was used to generate the reach-scale DEM(Figure D.2). The interpolated water surface during data collection (525 m3/s) in Figure D.3was differenced with the DEM to produce the depth map presented in Figure D.4.95Figure D.1: TIN generated from surveyed elevations.96Figure D.2: 2015 DEM.97Figure D.3: Water surface elevation during a discharge of 525 m3/s.98Figure D.4: Water depth during a discharge of 525 m3/s.99Appendix EAnalysis of Annual Sediment Loadusing a Rating Curve ApproachThe following excerpt is presented with permission from NHC and MFLNRO andwas obtained from:NHC. 2016. 2015 Sediment Transport Investigation on the Vanderhoof Reach of the NechakoRiver. Prepared for Ministry of Forests, Lands and Natural Resource Operations. January 31,2016.E.1 Bedload sediment transportData collected at the Upper Site suggests that the supply of bedload sediment became limitedduring the period of peak annual flow in June. This supply limitation is shown by the hysteresisin Figure E.1, where sediment transport declines at a greater rate per unit discharge during thereceding limb of the hydrograph. Two separate bedload-discharge rating curves were developedfor the Upper Site in order to accurately represent these different transport rates. The risingand falling limb rating curves were applied to the 2015 hydrograph, resulting in an estimatedannual sediment load of 9,250 m3. Predicted daily bedload transport at the Upper Site is insurprisingly good agreement with measured values given the inherent variability associated withsediment transport processes (Figure E.2).Bedload transport at the Lower Patch showed no clear relation with discharge and thereforeno rating curve could be used to derive the 2015 annual sediment load (Figure E.3). Rather,the estimated load of 3,050 m3 was obtained by interpolating daily transport rates betweensampled days. The Lower Patch bedload rate remained relatively constant between 100-300g/s/transect for the majority of 2015, until the greatest transport rate of 1,384 g/s/transectwas sampled on August 31st at a discharge of approximately 80 m3/s.100Figure E.1: Bedload rating curves developed for the Upper Site showing hysteresis (risinglimb shown in red, receding limb in blue).Figure E.2: Predicted versus observed bedload transport rate at Upper Site in 2015.The maximum daily bedload transport rate is predicted to have been roughly 2.5 timesgreater at the Upper Site than Lower Patch, translating to values of 190 m3/day and 75 m3/dayrespectively. The daily bedload rate at the Upper Site exceeded 75 m3/day for 61 consecutivedays between April 25th and June 24th, 2015. The significant difference in bedload transportbetween the upstream and downstream extent of the reach suggests 6,200 m3 of sediment hasbeen stored within the reach in this year. This net storage is interesting because previous yearshave observed the opposite trend with more sediment being output from the reach than input(NHC, 2014; NHC, 2015).101Figure E.3: Bedload transport at the Lower Patch showing no clear relation with dis-charge in 2015. Samples taken in 2015 are yellow, 2014 are black and 2013 areblue. The rating curve used in 2014 is shown in red.102Appendix FBedload Sediment Transportthrough the Nechako SpawningReachF.1 Data processingSampled transport rates were binned into 100 m3/s discharge intervals. If a location wassampled more than once during the discharge interval, the mean transport rate was calcu-lated. Figures F.1 to F.7 plot the bedload transport rate through different channels, sampledthroughout the 2015 flood hydrograph.F.2 Maps103Figure F.1: Sampled transport rates (discharge below 100 m3/s).Figure F.2: Sampled transport rates (100-200 m3/s).104Figure F.3: Sampled transport rates (200-300 m3/s).Figure F.4: Sampled transport rates (300-400 m3/s).105Figure F.5: Sampled transport rates (400-500 m3/s).Figure F.6: Sampled transport rates (500-600 m3/s).106Figure F.7: Sampled transport rates (600-700 m3/s).107Appendix GTwo-Dimensional Flow Modelling ofthe Nechako River using Nays2DHG.1 MethodsThis section provides supplementary information to section 5 Modelling in the thesis; previ-ously described aspects have been omitted.G.1.1 ConfigurationsFigure G.1: Solver type calculation conditions.108Figure G.2: Boundary conditions (for simulated discharge of 523 m3/s).The mesh was created from a polygonal centerline and defined domain width of 900 m. Thecurvilinear centerline was drawn down the middle of the reach, rather than along the mainsouthern channel, to avoid pinching of the mesh in meanders. The length of the modellingdomain is 3,300 m, resulting in a total of 118,440 cells for a 5-m resolution mesh. Elevation wasimported to the model from the DEM created in ArcGIS and overbank areas with no data wereassigned a high elevation of 638 m (Figure G.3). Shapefiles of bar contours and banklines fromthe RTK survey were imported and assigned different vegetation densities to locally increasedrag (Figure G.4). The model domain was separated into five regions (Figure G.5), each with aunique grain size distribution that was assigned based on the results of photo-sieving. Althoughit was preferable to use only one value of Manning’s roughness for the entire channel to reduceits influence on modelled shear stress, three polygons were drawn and assigned slightly higherroughness values within a hydrodynamically complex area to increase model stability, especiallyduring high-flow simulations (Figure G.6).A stage-discharge rating curve was developed to specify the water surface elevation at thedownstream extent of the model domain. This was necessary because flow is non-uniform duringmoderate to high discharge and therefore the model could not use uniform flow calculationsto set the downstream boundary condition. The rating curve (Figure G.7) was developedby iteratively adjusting the input WSE until good agreement was achieved between modelledoutput and measured stage at WSC Gauge 08JC001, located approximately 1 km upstream ofthe model boundary (Figure G.8).109Figure G.3: Entire modelling domain showing elevation in meters.Figure G.4: Vegetation density specified using shapefiles of bar contours and banklines,bars (green) were assigned a vegetation density of 0.1 stems/m2 and overbankareas (red) were assigned a vegetation density of 2 stems/m2.110Figure G.5: Specific grain size distributions were input to the model for each region basedon the results of photo-sieving underwater photos.Figure G.6: Manning’s roughness used for the entire domain in blue (n = 0.0215) withadditional polygons added for numerical stability during high flow simulations ingreen (n = 0.022) and in red (n = 0.024).111Figure G.7: Rating curve developed to specify WSE at the downstream extent of themodel domain.Figure G.8: Modelled vs measured stage at the WSC gauge used to develop the down-stream rating curve.112G.1.2 Processing the grain size dataThe grain size distributions for Regions 1-5 were obtained by photo-sieving a series of 30 un-derwater images taken across transects US, MU-A, MU-B, MU-D, M-A and LP. Figures G.9to G.13 provide examples of one photo from each transect. Images were photo-sieved using aWolman Pebble Count approach, where 100 grains were measured within the image frame atfixed intervals. The finest size class used for classification was sand, so all grains sized 2 mm orfiner were included within the 2 mm fraction. The resulting grain size distribution for Region1 through Region 5 are presented in Figures G.14 to G.18Figure G.9: Region 1 grain size distribution obtained by photo-sieving images at the UStransect.113Figure G.10: Region 2 grain size distribution obtained by photo-sieving images at theMU-A transect.Figure G.11: Region 3 grain size distribution obtained by photo-sieving images at theMU- B and MU-D transects.114Figure G.12: Region 4 grain size distribution obtained by photo-sieving images at theM-A transect.Figure G.13: Region 5 grain size distribution obtained by photo-sieving images at theLP transect.115Figure G.14: Region 1 grain size distribution.Figure G.15: Region 2 grain size distribution.116Figure G.16: Region 3 grain size distribution.Figure G.17: Region 4 grain size distribution.117Figure G.18: Region 5 grain size distribution.G.1.3 Processing the model outputThe first step in processing the model output was averaging the last 1,000 seconds of eachsimulation. A discharge of 523 m3/s was used to validate the model against WSE and velocitydata collected with the ADCP. Modelled WSE profiles (presented in section 5.2 Calibrationand Validation in the thesis) were validated by comparing each data point to the value withinthe nearest 5 m grid cell from the model output. This same approach was used to validate themodel against cross-channel binned velocity and depth data. For each of the 9 ADCP transects,absolute error and percent error was calculated for velocity, depth and specific discharge withineach cross-channel bin. Results from the model validation are presented as a series of plots;simulated versus observed velocity is presented in Figures G.19 to G.27, and simulated versusobserved depth in Figures G.28 to G.36. The cross-channel errors for each transect were thenaveraged into a mean absolute and mean absolute percent error (Table 5.1 within the thesis).118Figure G.19: Model velocity validation at transect TRA.119Figure G.20: Model velocity validation at transect TRB.120Figure G.21: Model velocity validation at transect TRC.121Figure G.22: Model velocity validation at transect TRD.122Figure G.23: Model velocity validation at transect TRE.123Figure G.24: Model velocity validation at transect TRF.124Figure G.25: Model velocity validation at transect TRG.125Figure G.26: Model velocity validation at transect TRH.126Figure G.27: Model velocity validation at transect TRI.127Figure G.28: Model depth validation at transect TRA.128Figure G.29: Model depth validation at transect TRB.129Figure G.30: Model depth validation at transect TRC.130Figure G.31: Model depth validation at transect TRD.131Figure G.32: Model depth validation at transect TRE.132Figure G.33: Model depth validation at transect TRF.133Figure G.34: Model depth validation at transect TRG.134Figure G.35: Model depth validation at transect TRH.135Figure G.36: Model depth validation at transect TRI.For each simulated discharge, modelled shear stress was used to calculate sediment transportcapacity using the Wilcock and Crowe (2003) transport function. First, the model output wasfiltered to remove areas with less than 10 cm depth. Then, a shear stress raster with a 10-mgrid resolution was interpolated to the dimensions of the wetted channel (i.e. > 10 cm depth).To assign a grain size distribution to each cell within the raster, size fractions and substratecharacteristics (percent sand and geometric mean grain size) were interpolated between Regions1-5 and mapped to a common grid. Each raster was then cropped to remove areas outside of thesurveyed bottom-of-bank bankline and shear velocity was calculated for each cell. The resultof this process is a single gridded data frame, where each cell contains the necessary attributesto apply the sediment transport function.136The calculated unit transport rate of each size fraction was then multiplied by the cellsize to obtain volumetric fractional transport rates. These fractional rates were summed toobtain a total sediment transport capacity for each cell. The subsequent step in the analysiswas to obtain the total cross-sectional transport capacity with downstream distance. Cross-sections were established at 30-m downstream intervals using the modelling mesh, excluding100-m channel lengths at the upstream and downstream extent of the model domain to avoidboundary effects. Transport capacity was extracted from each grid cell along the channel cross-sections and summed to obtain a total cross-channel capacity.G.2 Model resultsShear stress rasters are presented in Figures G.37 to G.45, transport capacity rasters in FiguresG.46 to G.54 and downstream capacity profiles in Figures G.55 to G.63. Although this analysiswas conducted for every 50 m3/s discharge interval between 45 m3/s and 775 m3/s, figures areonly presented per 100 m3/s simulated discharge interval.Figure G.37: Simulated shear stress during a discharge of 45 m3/s.137Figure G.38: Simulated shear stress during a discharge of 75 m3/s.Figure G.39: Simulated shear stress during a discharge of 175 m3/s.138Figure G.40: Simulated shear stress during a discharge of 275 m3/s.Figure G.41: Simulated shear stress during a discharge of 375 m3/s.139Figure G.42: Simulated shear stress during a discharge of 475 m3/s.Figure G.43: Simulated shear stress during a discharge of 575 m3/s.140Figure G.44: Simulated shear stress during a discharge of 675 m3/s.Figure G.45: Simulated shear stress during a discharge of 775 m3/s.141Figure G.46: Estimated sediment transport capacity during a discharge of 45 m3/s.Figure G.47: Estimated sediment transport capacity during a discharge of 75 m3/s.142Figure G.48: Estimated sediment transport capacity during a discharge of 175 m3/s.Figure G.49: Estimated sediment transport capacity during a discharge of 275 m3/s.143Figure G.50: Estimated sediment transport capacity during a discharge of 375 m3/s.Figure G.51: Estimated sediment transport capacity during a discharge of 475 m3/s.144Figure G.52: Estimated sediment transport capacity during a discharge of 575 m3/s.Figure G.53: Estimated sediment transport capacity during a discharge of 675 m3/s.145Figure G.54: Estimated sediment transport capacity during a discharge of 775 m3/s.Figure G.55: Profile of downstream transport capacity during a discharge of 45 m3/s.146Figure G.56: Profile of downstream transport capacity during a discharge of 75 m3/s.Figure G.57: Profile of downstream transport capacity during a discharge of 175 m3/s.147Figure G.58: Profile of downstream transport capacity during a discharge of 275 m3/s.Figure G.59: Profile of downstream transport capacity during a discharge of 375 m3/s.148Figure G.60: Profile of downstream transport capacity during a discharge of 475 m3/s.Figure G.61: Profile of downstream transport capacity during a discharge of 575 m3/s.149Figure G.62: Profile of downstream transport capacity during a discharge of 675 m3/s.Figure G.63: Profile of downstream transport capacity during a discharge of 775 m3/s.150To explore possible relations between flow and sediment transport through the spawningreach, the cumulative transport capacity was plotted for three downstream locations over a seriesof hydrographs. This was motivated by previous investigations (NHC, 2014; NHC, 2015; NHC,2016) which observed a discrepancy between the amount of sediment transported into and outof the reach depending on the annual hydrograph. The three locations correspond to upstream(US bedload transect), mid-reach and downstream (LP bedload transect) locations. At eachlocation, transport capacity was averaged over a 90-m downstream distance to reduce the effectof local variations. This was repeated for all simulated discharge intervals. Daily flow datawere downloaded from WSC gauge 08JC001 and binned into 50 m3/s intervals, correspondingto the simulated discharge intervals. The transport capacity at each location was multiplied bythe flow duration within each bin to obtain the cumulative transport capacity. The analysiswas done for three hydrograph sequences of three years each. The sequences included threelow-flow years (1954-1956) which occurred during the period of reservoir infilling (Figure G.64),three typical hydrograph years (2002-2004) (Figure G.65) and the past three years (2014-2016)(Figure G.66) which included the high flow in 2015. The 2015 flood was the only year wherecapacity was greater within the upstream part of the reach than downstream (Figure G.67).To remove the effect that varying the grain size distribution has on estimated capacity, thesame analysis was done with a single grain size distribution assigned to the entire reach. Thesingle grain size distribution was obtained by taking the mean of all GSD data collected at alllocations, and had a D50 of 13.6 mm and 17% sand. Results from this analysis are presentedin the same order as was previously shown; 1954-1957 (Figure G.68), 2002-2004 (Figure G.69),2014-2016 (Figure G.70) and 2015 (Figure G.71). By using the single GSD, transport ratesat the US transect greatly increase because the mean GSD is finer than observed substratecomposition.It is important to note that this analysis is not intended to predict future sediment loadsor estimate historic loads with any degree of accuracy. Rather, it is intended as an interpretivetool to better understand reach-scale sediment dynamics.151Figure G.64: Cumulative transport capacity for a sequence of low-flow hydrographs.Figure G.65: Cumulative transport capacity for a sequence of typical hydrographs.152Figure G.66: Cumulative transport capacity for a sequence containing a high flow hy-drograph.Figure G.67: Cumulative transport capacity for the 2015 flood.153Figure G.68: Cumulative transport capacity (uniform GSD) for a sequence of low-flowhydrographs.Figure G.69: Cumulative transport capacity (uniform GSD) for a sequence of typicalhydrographs.154Figure G.70: Cumulative transport capacity (uniform GSD) for a sequence containing ahigh flow hydrograph.Figure G.71: Cumulative transport capacity (uniform GSD) for the 2015 flood.155G.3 LimitationsThe accuracy of model performance is limited in areas that have relatively low surveyed pointdensities. These areas required considerable interpolation between data points to produce theDEM. Spatial interpolation is likely to misrepresent complex channel geometry and may gen-erate artefacts that affect simulation output. The error produced by DEM creation and surveypoint density has been documented elsewhere and remains a limiting factor for 2-dimensionalmodelling projects (Pasternack et al., 2004; Pasternack et al., 2006). Possible simulation errorcaused by DEM interpolation can be seen as locally high shear stresses within the middle of thereach (at approximately 432500 m E, 5986400 m N) for flows between 125-275 m3/s (FiguresG.37 to G.40).Mesh resolution is another factor that limits model performance (Crowder and Diplas, 2000),especially within some of the smaller secondary channels. The 5-m resolution used herein waschosen a compromise between spatial detail and reasonable computation times. Given thatthe purpose of the model was to explore reach-scale dynamics, rather than to evaluate smallto meso-scale restorative measures or local habitat conditions, the resolution of the analysis isconsidered acceptable.The accuracy of sediment transport capacity is limited by numerous factors. Firstly, calcu-lation results are very sensitive to the grain size distribution of the bed surface and the percentsand content. The data used to define these characteristics was not very robust because un-derwater images were only taken at a few downstream locations, only a few images were takenper location, the area covered per image was relatively small (especially for coarse substrate)and visibility was relatively poor due to water turbidity. Attributing a grain size distributionto all cells within the model domain required extensive interpolation and is only representativeof the very general trend of downstream fining.In addition, the grain size of the substrate may not be representative of the sediment beingtransported as bedload. This was the case at the upstream end of the reach, where sand istransported over a coarse static bed rather than constituting a fraction of it. In contrast, thehigh sand content within the bed at the downstream end of the reach may cause sediment tobe transported as migrating bedforms. The development of bedforms increases the importanceof form drag and decreases the proportion of the total shear stress exerted as skin drag; theproportion responsible for grain mobility within sediment transport functions (note that totalshear stress was used to calculate capacity in this study, which is another significant limitation).Neither the model nor the sediment transport function capture the effect of bedforms on localhydrodynamics and sediment transport.G.4 ReferencesCrowder, D.W., and Diplas, P. 2000. Using two-dimensional hydrodynamics models at scalesof ecological importance. Journal of Hydrology, 230, 172-191.156NHC. 2014. Nechako River 2013 Sediment Transport Investigations. Prepared by North-west Hydraulic Consultants for the Ministry of Forests, Lands and Natural Resource Operations,North Vancouver BC. 48 pp.NHC. 2015. 2014 Sediment Transport Investigation on the Vanderhoof Reach of the NechakoRiver. Prepared for Ministry of Forests, Lands and Natural Resource Operations, North Van-couver BC. 29 pp.NHC. 2016. 2015 Sediment Transport Investigation on the Vanderhoof Reach of the NechakoRiver. Prepared for Ministry of Forests, Lands and Natural Resource Operations. North Van-couver BC. 38 pp.Pasternack, G. B., Wang, C. L., and Merz, J. E. 2003. Application of a 2D hydrodynamicmodel to design of reach scale spawning gravel replenishment on the Mokelumne River, Cali-fornia. River Research and Applications, 19, 1-21.Pasternack, G., Gilbert, A., Wheaton, J. and Buckland, E. 2006. Error propagation forvelocity and shear stress prediction using 2D models for environmental management. Journalof Hydrology. 328, 227-241.157

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0342969/manifest

Comment

Related Items