<<

The Pennsylvania State University

The Graduate School

College of Earth and Mineral Sciences

EVALUATING CONTROLS ON CREVASSE-SPLAY GROWTH IN MODERN AND ANCIENT

FLUVIAL SYSTEMS

A Thesis in

Geosciences

by

Craig L. Millard

 2013 Craig L. Millard

Submitted in Partial Fulfillment of the Requirements for the Degree of

Master of Science

May 2013

The thesis of Craig L. Millard was reviewed and approved* by the following:

Elizabeth A. Hajek Assistant Professor of Geosciences Thesis Adviser

Eric Kirby Associate Professor of Geosciences

Rudy L. Slingerland Professor of Geology

Chris J. Marone Professor of Geosciences Associate Head of Graduate Program of the Department of Geosciences

*Signatures are on file in the Graduate School

ii

Abstract

Crevasse-splays and related facies are common in some avulsive fluvial systems, while being nearly absent in others. As such, the study of splay and other overbank deposits may be fundamental to understanding the processes of fluvial avulsion and basin-filling in terrestrial systems. Despite this importance, the primary controls on crevasse-splay growth and development are poorly understood and constrained. Within this study, the role of floodplain drainage and channel grain-size distribution on crevasse-splay growth is examined, as these variables influence the amount of sediment and water discharged through levee crevasses. To evaluate the influence of floodplain drainage and channel grain-size distributions on crevasse-splay growth, three approaches were used. First, crevasse-splay extent, plan form, and frequency were documented using aerial photography from Google Earth in four modern, avulsive river systems including the Ovens (Victoria, Australia), Sandover (Northern Territory, Australia), Saskatchewan (Saskatchewan, Canada), and upper Columbia (British Columbia, Canada). Second, field observations and measurements from the Paleocene Fort Union Formation (Bighorn Basin, Wyoming), Paleocene-Eocene Willwood Formation (Bighorn Basin, Wyoming), and Late -Paleocene Ferris Formation (Hanna Basin, Wyoming) documented floodplain drainage and grain-size characteristics in these ancient systems. Finally, numerical modeling using Delft3D-FLOW was conducted to explore splay growth under different floodplain drainage and channel grain-size conditions. Collectively, modern observations, ancient deposits, and modeling results show that systems producing wide-spread, basin-filling crevasse-splay deposits are associated with well- drained floodplains with steep cross-floodplain water-surface gradients and abundant intermediate grain-sizes (i.e. coarse silt to fine sand, which is readily suspended in channel flows and also settles from suspension quickly as through-crevasse flows expand and decelerate on the floodplain). In contrast, splay production was limited in systems with poorly-drained floodplains and/or coarse grain sizes. Some of the systems featured little or no crevasse-splay deposition (e.g. Ovens River and Ferris Formation) or numerous splays with relatively little depositional volume (e.g. upper Columbia and Sandover Rivers). In these systems, splay deposition is limited because standing water on the floodplain reduced cross-floodplain water-surface slopes,

iii

suppressing sediment advection away from the channel margin, and/or these systems lacked intermediate sediment sizes that supply crevasse-splay sediment to the floodplain. Ultimately, these results suggest that well-drained floodplains with steep, lateral water- surface gradients promote extensive crevasse-splay growth. The influence of channel grain-size distribution is less pronounced than floodplain-drainage conditions, although abundant "intermediate" sediment supply (i.e. coarse suspended sediment) promotes large crevasse-splay deposits.

iv

Table of Contents

List of Figures ...... vii List of Tables ...... x Acknowledgements ...... xi 1.0 Introduction ...... 1 2.0 Background ...... 14 2.1 Controls on Splay Development ...... 14 2.1.1 Primary Controls on Splay Area and Volume ...... 15 2.2 Relationship of Splays and Fluvial Avulsion ...... 17 2.2.1 Splays within Progradational Avulsion Systems ...... 18 2.2.2 Splays within Incisional Avulsion Systems ...... 19 2.3 Crevasse-splays in the Stratigraphic Record ...... 20 3.0 Hypotheses and Study Phases ...... 29 3.1 Hypotheses ...... 29 3.1.1 Hypothesis I ...... 29 3.1.2 Hypothesis II ...... 29 3.2 Study Components ...... 30 4.0 Observations in Modern Rivers ...... 33 4.1 Study Areas ...... 33 4.1.1 Upper Columbia River ...... 33 4.1.2 Saskatchewan River ...... 34 4.1.3 Sandover River ...... 35 4.1.4 Ovens River ...... 36 4.2 Google Earth Observations and Measurements ...... 37 4.3 Google Earth Study Results ...... 38 4.3.1 General Appearance of Splays ...... 38 4.3.2 Analysis of Splay Shape ...... 40 4.3.2.1 Splay Frequency and Basin Filling ...... 40 4.3.2.2 Splay Sizes and Shapes ...... 41 4.3.3 Splay Volume Estimates ...... 43 4.4 Discussion ...... 45 5.0 Observations in Ancient Systems ...... 71 v

5.1 Formations ...... 71 5.1.1 Ferris Formation ...... 71 5.1.2 Fort Union Formation ...... 72 5.1.3 Willwood Formation ...... 72 5.2 Field Observations ...... 73 5.3 Results ...... 74 5.3.1 Ferris Formation ...... 74 5.3.2 Fort Union Formation ...... 75 5.3.3 Willwood Formation ...... 76 5.4 Discussion ...... 77 6.0 Modeling Splay Development in Delft3D ...... 85 6.1 Methods...... 85 6.1.1 Model Grid and Initial Topography ...... 86 6.1.2 Initial Bed Sediments ...... 86 6.1.3 Initial and Boundary Conditions, Including Water and Sediment Discharge ...... 88 6.2 Results ...... 90 6.2.1 Run 1: Fine Channel Sediment with Dry Floodplain ...... 90 6.2.2 Run 2: Fine Channel Sediment with Wet Floodplain ...... 91 6.2.3 Run 3: Intermediate Channel Sediment with Dry Floodplain ...... 91 6.2.4 Run 4: Intermediate Channel Sediment with Wet Floodplain ...... 92 6.2.5 Run 5: Coarse Channel Sediment with Dry Floodplain ...... 92 6.2.6 Run 6: Coarse Channel Sediment with Wet Floodplain ...... 93 6.3 Discussion ...... 94 7.0 Synthesis and Discussion ...... 140 8.0 Conclusions ...... 142 References ...... 144 Appendix A: Splay Polygon Raw Values ...... 154 Appendix B: Flood-basin, Channel, and Island Measurements ...... 161 Appendix C: Values from Representative Ancient Grain-Size Distributions ...... 165 Appendix D: Summary Statistics for Studied Outcrops ...... 171 Appendix E: Location of Data Files ...... 174

vi

List of Figures

Figure 1.1: Splay Development Schematic...... 4 Figure 1.2: Splay Length vs. Channel Width...... 10 Figure 1.3: Splay Width vs. Channel Width...... 11 Figure 1.4: Splay-Deposit Thickness vs. Channel Width...... 12 Figure 1.5: Comparison of Saskatchewan and Amazon Splays...... 13 Figure 2.1: Floodplain Drainage Types (after Adams et al., 2004)...... 22 Figure 2.2: Progradational and Incisional Avulsion Diagrams and Stratigraphy (after Mohrig et al., 2000 and Hajek, personal file)...... 24 Figure 2.3: Progradational System Splay Evolution (after Perez-Arlucea and Smith, 1999)...... 25 Figure 2.4: Stratigraphically Transitional and Abrupt Representations (after Jones, 2007)...... 28 Figure 3.1: Location Map for Modern Systems...... 31 Figure 3.2: Location Map for Wyoming Field Sites (after Jones and Hajek, 2007)...... 32 Figure 4.1: Upper Columbia River Study Reach with Mapped Splays...... 47 Figure 4.2: Splays Mapped near Castledale...... 48 Figure 4.3: Comparison of Main Channel and Castledale Reaches...... 49 Figure 4.4: Saskatchewan River Study Reach with Mapped Splays...... 50 Figure 4.5: Sandover River and Ovens River Study Reaches with Mapped Splays...... 51 Figure 4.6: Splay Measurements Made within Google Earth...... 52 Figure 4.7. Splay Examples from the Upper Columbia River...... 53 Figure 4.8: Splay Examples from the Saskatchewan River...... 54 Figure 4.9: Splay Examples from the Sandover River...... 55 Figure 4.10: Channel Development and Floodplain Incisions along the Ovens River...... 56 Figure 4.11: Splay Areas within the Three Systems...... 58 Figure 4.12: Channel Widths within the Three Systems...... 59 Figure 4.13: Orthogonal Extent into Flood-basin within the Three Systems...... 60 Figure 4.14: Measured Splay Widths within the Three Systems...... 61 Figure 4.15: Splay Channel Path Lengths within the Three Systems...... 62 Figure 4.16: Normalized Orthogonal Extent (by Channel Width)...... 63 Figure 4.17: Aspect Ratio (Maximum Splay Width/Splay Channel Path Length)...... 64 Figure 4.18: Splay Area vs. Channel Width in the Upper Columbia System...... 66 Figure 4.19: Splay Area vs.Channel Width in the Sanodver and Saskatchewan Systems...... 67

vii

Figure 4.20: Orthogonal Extent vs. Channel Width in the Three Systems...... 68 Figure 4.21: Splay Channel Path Length vs. Maximum Splay Width in the Three Systems...... 69 Figure 5.1: Representation of Field Sampling...... 80 Figure 5.2: Range of Visually Estimated Grain-Sizes for Facies Types...... 81 Figure 5.3: Example Ferris Grain-Size Distribution...... 82 Figure 5.4: Example Fort Union Grain-Size Distribution...... 83 Figure 5.5: Example Willwood Grain-Size Distribution...... 84 Figure 6.1: Abbreviated modeling flow plan...... 96 Figure 6.2: Modeled grid and bathymetry...... 97 Figure 6.3: Proportions of Each Sediment Fraction in Model Runs...... 101 Figure 6.4: Cumulative Sedimentation in Channel with No Crevasse...... 103 Figure 6.5: Initial Flow Magnitudes for Dry Floodplain Models...... 104 Figure 6.6: Initial Flow Magnitudes for Wet Floodplain Models...... 105 Figure 6.7: Crevasse and Downstream Discharges for Run 1...... 106 Figure 6.8: Final Flow Magnitudes (Day 162) for Run 1: Fine Channel Sediment with Dry Floodplain. 107 Figure 6.9: Cumulative Erosion/Sedimentation (Day 162) for Run 1: Fine Channel Sediment with Dry Floodplain...... 108 Figure 6.10: Run 1: Cross-sections A, B, and C...... 109 Figure 6.11: Run 1: Cross-sections D, E, and F...... 110 Figure 6.12: Cumulative Total Sediment Transport Through Crevasse...... 111 Figure 6.13: Instantaneous Total Sediment Discharge Through Crevasse...... 112 Figure 6.14: Normalized Instantaneous Sediment Discharge Through Crevasse...... 113 Figure 6.15: Crevasse and Downstream Discharges for Run 2...... 115 Figure 6.16: Final Flow Magnitudes (Day 153.5) for Run 2: Fine Channel Sediment with Wet Floodplain...... 116 Figure 6.17: Cumulative Erosion/Sedimentation (Day 153.5) for Run 2: Fine Channel Sediment with Wet Floodplain...... 117 Figure 6.18: Run 2: Cross-sections A, B, and C...... 118 Figure 6.19: Run 2: Cross-sections D, E, and F...... 119 Figure 6.20: Crevasse and Downstream Discharges for Run 3...... 120 Figure 6.21: Final Flow Magnitudes (Day 198) for Run 3: Intermediate Channel Sediment with Dry Floodplain...... 121 Figure 6.22: Cumulative Erosion/Sedimentation (Day 198) for Run 3: Intermediate Channel Sediment with Dry Floodplain...... 122 viii

Figure 6.23: Run 3: Cross-sections A,B, and C...... 123 Figure 6.24: Run 3: Cross-sections D, E, and F...... 124 Figure 6.25: Crevasse and Downstream Discharges for Run 4...... 125 Figure 6.26: Final Flow Magnitudes (Day 148) for Run 4: Intermediate Channel Sediment with Wet Floodplain...... 126 Figure 6.27: Cumulative Erosion/Sedimentation (Day 148) for Run 4: Intermediate Channel Sediment with Wet Floodplain...... 127 Figure 6.28: Run 4: Cross-sections A, B, and C...... 128 Figure 6.29: Run 4: Cross-sections D, E, and F...... 129 Figure 6.30: Crevasse and Downstream Discharges for Run 5...... 130 Figure 6.31: Final Flow Magnitudes (Day 196) for Run 5: Coarse Channel Sediment with Dry Floodplain...... 131 Figure 6.32: Cumulative Erosion/Sedimentation (Day 196) for Run 5: Coarse Channel Sediment with Dry Floodplain...... 132 Figure 6.33: Run 5: Cross-sections A, B, and C...... 133 Figure 6.34: Run 5: Cross-sections D, E, and F...... 134 Figure 6.35: Crevasse and Downstream Discharges for Run 6...... 135 Figure 6.36: Final Flow Magnitudes (Day 141.5) for Run 6: Coarse Channel Sediment with Wet Floodplain...... 136 Figure 6.37: Cumulative Erosion/Sedimentation (Day 75) for Run 6: Coarse Channel Sediment with Wet Floodplain...... 137 Figure 6.38: Run 6: Cross-sections A, B, and C...... 138 Figure 6.39: Run 6: Cross-sections D, E, and F...... 139

ix

List of Tables

Table 1.1: Splays Described within Literature...... 5 Table 2.1: Sediment Transport Predicted by Rouse Number...... 23 Table 2.2: Well Described Incisionally Avulsive Systems...... 26 Table 4.1: Measured Splay Quantities in Modern Systems...... 57 Table 4.2: Summary Statistics and T-test Comparison Results...... 65 Table 4.3. Estimated Splay Volumes and Normalized Splay Discharge...... 70 Table 6.1: Grid and Bathymetry Conditions...... 98 Table 6.2: Modeled Sediment Fractions...... 99 Table 6.3: Vertical Sediment Thicknesses of Modeled Grain-size Distributions...... 100 Table 6.4: Final Model Parameters, Initial Conditions, and Boundary Conditions...... 102 Table 6.5: Surface Areas and Volumes for Splays Produced within Models...... 114

x

Acknowledgements

This thesis would not have been possible without the support of various grants to me and my adviser, Elizabeth “Liz” A. Hajek. I received graduate student grants from the Geological Society of America and the American Association of Petroleum Geologists, both in the amount of $2000. Additionally, this research is part of a larger grant from the National Science Foundation (Award #1124167) provided to Liz Hajek and Douglas Edmonds (Indiana University) for the investigation of incisional avulsion processes. Within the Penn State community, I am also personally indebted for the financial support provided by the Department of Geosciences, by my adviser, and by Chris Marone. The Department of Geosciences provided financial support and tuition for four semesters of coursework and research. Funding for summer research was provided through Chris Marone (summer 2011) and Liz Hajek (summer 2012). I am especially grateful for the support of Chris Marone, given that I am not his student and the only thing he asked for in return was that I “do a good job for Liz.” Beyond financial support, I appreciate the advice, suggestions, and mentoring provided by Liz Hajek. I further appreciate advice and suggestions provided by the members of my committee: Rudy Slingerland, Eric Kirby, and Doug Edmonds (an unofficial member). I am especially grateful for the access to equipment and the Delft3D license provided by Rudy Slingerland, without which, significant portions of this thesis would not have been possible. Finally, I’d like to thank my family and friends for their words of encouragement and for providing the necessary amount of levity to complete this thesis. Whew, I’ve made it.

xi

Dedication

During my time at Penn State, I lost several people close to my family and me. These include a family friend, Paul Kahler (1927 - 2010), my granduncle and last surviving relative of my paternal grandfather, Paul Millard (1916 – 2010), my grandaunt and last surviving sibling of my maternal grandfather, Mary Barzanti (1918 – 2010), and my uncle, Edward Millard (1944 – 2012). This thesis is dedicated to their memory.

xii

1.0 Introduction

In fluvial systems, crevasse-splay deposits develop through breaks in levees and channel banks as flow expansion outside of the channel results in decreased competence for further transport of sediment, which leads to overbank deposition (Bridge and Demicco, 2008). Crevasse splays play an important role in levee development, floodplain sedimentation, and channel avulsion processes. Crevasse-splay deposits generally consist of clay, silt, and sand, with coarser sediments deposited proximal to the channel and finer sediments deposited farther from the channel (e.g. Smith et al., 1989) (Figure 1.1); consequently, crevasse-splay deposition also has important consequences for the size and inter-connectedness of hydrocarbon reservoirs and groundwater aquifers located within buried fluvial systems. Crevasse-splays have been studied extensively in several modern and sub-surface Holocene systems (e.g. Smith et al., 1998; Bristow et al., 1999; Perez-Arlucea and Smith, 1999; Stouthamer, 2001), and in ancient systems through outcrop (e.g. Eberth and Miall, 1991; Mjos et al., 1993; Sarti et al., 2001; Pranter et al., 2009) and subsurface studies (Reynolds, 1999), as presented in Table 1.1. Nevertheless, it remains difficult to predict crevasse-splay characteristics in a fluvial system with given boundary conditions. In this study, I aim to determine which variables exert primary control on crevasse-splay size for similarly sized fluvial systems. While several variables influence crevasse-splay development and growth, I hypothesize that factors controlling sediment supply and sediment-dispersal patterns should control the characteristic size and volume of crevasse splays produced by a given fluvial system. Sediment supply is important because very fine sediments may not be deposited proximal to channel margins and very coarse sediments transported as bed load may be unlikely to leave the channel through crevasses. Additionally, excessive supply may promote quick crevasse healing and limited supply may promote further splay growth. On the floodplain, flow patterns determine where and how sediment is dispersed, which influences the extent of crevasse-splay deposits. For these reasons, I hypothesize that splay size and volume is most dependent on variables that influence floodplain sediment supply and deposition patterns. Within fluvial systems there are three factors—ceteris paribus—that impact both sediment supply to and deposition patterns on the floodplain: (1) river discharge, (2) floodplain water-surface slope, and (3) channel sediment grain-size distribution. River discharge is the most 1

direct control on crevasse-splay growth, as higher discharges lead to a higher sediment supply and greater potential for extensive overbank sediment dispersal. Thus, river discharge should inherently set the maximum potential crevasse-splay scale within a given system, as high water and sediment discharges lead to larger crevasse-splay deposits. Floodplain water-surface slope influences spatial deposition rates away from the crevasse opening (e.g. Pizzuto, 1987; Adams et al., 2004; Hajek and Wolinsky, 2012) and different water-surface gradients can lead to different water (and sediment) discharges through crevasses. Channel sediment grain-size distribution influences both the availability of material to the splay (e.g. Slingerland and Smith, 1998) and downstream sediment extraction (i.e. deposition) profiles (e.g. Parker, 1991; Cazanacli and Smith, 1998; Fedele and Paola, 2007). Review of previously studied crevasse-splays (Table 1.1, Figures 1.2 - 1.4) shows that, although large rivers can produce large crevasse splays, it is not necessarily the dominant control on crevasse-splay size in different systems. For example, large crevasse splays are formed via both large-scale rivers (e.g. glacial outwash Mississippi River, Blum et al., 2000) and relatively small rivers (e.g. Texas’s Colorado River, Aslan and Blum, 1999), and there is no predictable relationship between splay dimensions (i.e. length, width, and thickness) and main channel width (a rough proxy for discharge), as shown in Figures 1.2 – 1.4. The relationship between channel and splay sizes may be particularly weak because, in this data set, channel width is weakly correlated to discharge (e.g. the Niobrara River has a channel width of ~250 meters, but an average discharge of only 48 m3/s, Bristow et al., 1999). Nonetheless, based on these data, channel scale cannot be used as a direct predictor of crevasse-splay scale. Figure 1.5 provides an example of this, where, although the Amazon River is substantially larger than the Saskatchewan River, the scale of crevasse splays in the Saskatchewan system are large relative to the size of the main river, while Amazon splays are small relative to the scale of the main channel. Thus, for a channel with a given discharge, crevasse-splay size may be largely influenced by factors that affect the downstream mass-balance within the crevasse-splay system: floodplain water-surface slopes and channel grain-sizes. These factors specifically control the spatial deposition and downstream fining rates from the crevasse mouth to the distal splay edge. To evaluate how floodplain water-surface slope and channel grain size affect splay development, I tested two main hypotheses: (1) crevasse-splay size and volume is larger in systems featuring steep cross-floodplain water-surface gradients that allow for the development 2

of advective currents to draw sediment away from the channel margin and into the floodbasin (e.g. Adams et al., 2004), and (2) crevasse-splay size and volume is greater in systems featuring intermediate grain-size distributions (i.e. coarse silt to fine sand in sand-bed rivers) that is both easily suspended in channels and readily deposited in unconfined overbank flows. To test these hypotheses, I employed a multifaceted approach that included aerial-image analysis of modern rivers using Google Earth, field measurements in ancient systems, and morphodynamic modeling of splay development using Delft3D software. The following sections detail background and the results for each study. Combined results of these studies suggests that floodplain water-surface slope and channel grain-size are useful predictors of crevasse-splay abundance and size in modern and ancient systems, and that, ultimately, floodplain water-surface slope may be the primary control on whether large splays form in alluvial systems.

3

Figure 1.1: Splay Development Schematic. (A) A currently active splay with sands deposited proximally and mud deposited distally. In this system, numerous abandoned splays are evident and have contributed to floodplain and levee aggradation along the channel. (B) When beyond threshold conditions (e.g. Slingerland and Smith, 1998), splay evolution may lead to avulsion for a channel system.

4

Table 1.1: Splays Described within Literature. Location/Formation Classification Type Splay Splay Splay Splay Grain-Sizes Channel Notes (Source) Length Width Height Width (meters) (meters) (meters) (meters) Umbum Creek terminal Modern Single Splay 8600 13500 0.6 - 1 Mostly fine- 200 Dimensions from splay, (dry) Lake Eyre, (Surface) grained sand, some Google Earth, central Australia (Lang et al., coarse sand near channel depth 2 -3 2004) mouth, some silts meters and muds distally Southern splay, Wild Cow Modern Single Splay 7300 600 – 1 – 2 Very Fine – Fine, 125 Channel width Island, Colorado River, north (Surface) 3000 silt, based on measured in of Matagorda, Texas (Aslan similar subsurface Google Earth and Blum, 1999) deposits (see below) Northern splay, Colorado Modern Single Splay 4500 400 – 1 – 2 Very Fine – Fine, 100 Channel width River, north of Matagorda, (Surface) 1300 silt, based on measured in Texas (Aslan and Blum, similar subsurface Google Earth 1999) deposits Windy Lake Splay (ca. 1915 Modern Single Splay 3500 2000 0.5 – 2.5 Very Fine – Fine 150 – 1953 AD), Mossy River, (Surface) (mean ~2) Sand, silt, clay Cumberland Marshes, Saskatchewan, Canada (Smith and Perez-Arlucea, 1994) Splay off Brant Bayou (1975- Modern Single Splay 3000 4000 2 - 3 Up to Very Fine – 150 Channel width 1993 AD), Mississippi River (Surface) Medium sand from Google delta, LA (Roberts, 1997) Earth Douglas Creek terminal Modern Single Splay 1500 2300 0.1 – 1.7 Medium – Coarse 150 Channel depth 1 – splay, (dry) Lake Eyre, (Surface) sand proximally, 1.7 meters central Australia (Fisher et Very Fine sand, al., 2008) silt, and clay East Splay, Niobrara River Modern Single Splay 1000 200 2 – 3 Medium sand with 250 Channel discharge near confluence with (Surface) fine-grained sands is approximately Missouri, Nebraska, May and silts downdip 48 m3/s 1993 - June 1994 (Bristow et al., 1999)

5

Location/Formation Classification Type Splay Splay Splay Splay Grain-Sizes Channel Notes (Source) Length Width Height Width (meters) (meters) (meters) (meters) West Splay, Niobrara River Modern Single Splay 250 150 1 – 1.5 Medium sand with 250 Channel discharge near confluence with (Surface) fine-grained sands is approximately Missouri, Nebraska, autumn and silts downdip 48 m3/s 1995 - August 1996 (Bristow et al., 1999) Colville River Delta, North Modern Composite 240 – 170 – Not reported Very Fine to Fine 270 – Channel depths up Slope, Alaska (Tye, 2004) (Surface) (n = 14) 1380 670 sand (based on 780 to 13 meters, (mean = (mean = Nanuk Lake Splay) (40 discharge of 492 630) 380) channels) m3/s (probably combined) Nanuk Lake splay, began Modern Single Splay 730 220 Not reported Very Fine to Fine 200 Channel depths up pre-1949, Colville River (Surface) sand to 12 meters Delta, North Slope, Alaska (Tye, 2004) Recent (1953-1977 AD) Modern Single Splay 600 300 2 - 3 Silt to Fine sand, 100 Channel depth 10 splay north of Centre (Surface) with Very Fine – – 12 meters, Angling Channel, Medium sand in discharge is Cumberland Marshes, splay-channel approximately Saskatchewan, (Farrell, 450 m3/s 2001) Corps Breach intentional Modern Single Splay 428 195 0.1 – 0.8 70 – 95% is 15 23.5 – 25.5 m3/s levee breach (1997 – 2000 (Surface) (max) (max) Medium and to cause flooding, AD), Cosumnes River, Coarse sand Channel width California (Florsheim and from Google Mount, 2002) Earth Accidental Forest intentional Modern Single Splay 387 207 0.1 – 0.7 70 – 95% is 16 23.5 – 25.5 m3/s levee breach (1997 – 2000 (Surface) (max) (max) Medium and to cause flooding, AD), Cosumnes River, Coarse sand > 100 m3/s to California (Florsheim and cause major Mount, 2002) extraction, Channel width from Google Earth

6

Location/Formation Classification Type Splay Splay Splay Splay Grain-Sizes Channel Notes (Source) Length Width Height Width (meters) (meters) (meters) (meters) 1782 AD flood of Thinhope Modern Single Splay 80 50 < 1 Cobbles and 10 Burn, Northern Pennines, UK (Surface) boulders with (Macklin et al., 1992) sandy matrix, some bedded sand Ephemeral tributary of Modern Single Splay 40 15 0.3 – 0.4 Medium – Coarse 5 – 6 Channel depth Clarence River, northeast (Surface) moderately sorted approximately 1 New South Wales, Australia sand meter (O’Brien and Wells, 1986) Inland deltaic splay (1421 - Recent (Sub- Single Splay 15000 5000 0 – 4 Fine – Medium Not 1685+ AD), Rhine-Meuse surface) fining upward, silt reported delta, Netherlands (Kleinhans and clay et al., 2010) Holocene, lower Mississippi Recent (Sub- Single Splay 14000 8000 4 – 8 Medium – Coarse Not River Valley, Missouri - surface) grading to silt and reported Arkansas (Blum et al., 2000) clayey silt down valley Sub-surface splays, Colorado Recent (Sub- Composite 4000 Not 1 – 2 Very Fine – Fine, Not River, near Matagorda, Texas surface) reported silt reported (Aslan and Blum, 1999) Holocene (ca. 3000-2600 Recent (Sub- Single Splay 3000+ 2000 < 1.5 Fine sand and silt, 300 - 500 ybp), Lower Mississippi surface) coarsening upward River Valley, Tensas Basin, Nolan Site, Indian Mound Crevasse, northeast Louisiana (Arco et al., 2006) Lopik splay complex (~4205 Recent (Sub- Single Splay 3000 – 1325 – 3.2 – 5 Silty clay, sandy 200 Channel depths – 3925 years before present surface) 4000 1875 clay, Fine – Coarse (channel from (ybp)), Rhine-Meuse Delta, sand belt) approximately 5 – Netherlands (Stouthamer, 15 meters 2001) Zuid-Stuivenberg splay Recent (Sub- Single Splay 2000 – 500 – 0.5 – 3 (4-5 Silty clay, Sandy 350 – Channel depths complex (ceased ~3660 ybp), surface) 4000 1000 in some clay, F-M sand 500 from Rhine-Meuse Delta, splay (mostly silty clay (channel approximately 5 – Netherlands (Stouthamer, channels) downstream) belt) 15 meters 2001)

7

Location/Formation Classification Type Splay Splay Splay Splay Grain-Sizes Channel Notes (Source) Length Width Height Width (meters) (meters) (meters) (meters) Western Hallandse Ijssel Recent (Sub- Single Splay 2500 1000 - 0.5 - 2 Silty clay, Sandy 200 – Channel depths Splay (channel active 1805- surface) 1500 clay, F-M sand 250 from 665 ybp), Rhine-Meuse (mostly silty clay (channel approximately 5 – Delta, Netherlands downstream) belt) 15 meters (Stouthamer, 2001) Eastern Hallandse Ijssel Recent (Sub- Single Splay 2000 1000 0.5 - 2 Silty clay, Sandy 200 – Channel depths Splay (channel active 1805- surface) clay, F-M sand 250 from 665 ybp), Rhine-Meuse (mostly silty clay (channel approximately 5 – Delta, Netherlands downstream) belt) 15 meters (Stouthamer, 2001) Breached dyke at Diefdijk, Recent (Sub- Single Splay 1500 – 2500 0.1 – 1.4 Sandy to silty clay, 300 Netherlands, 1571 AD surface) 2000 sand (105-2000 (Hesselink et al., 2003) um) in splay- channels Westernmost Holocene sub- Recent (Sub- Single Splay 1500 1500 Up to 4 Fine sand, sandy 200 - 250 surface splay, Cabauw area, surface) clay, silty clay Rhine-Meuse delta, Netherlands (Weerts and Bierkens, 1993) 2nd easternmost Holocene Recent (Sub- Single Splay 1500 1500 3 - 4 Fine sand, sandy 200 - 250 splay, Cabauw area, Rhine- surface) clay, silty clay Meuse delta, Netherlands (Weerts and Bierkens, 1993) 2nd westernmost Holocene Recent (Sub- Single Splay 1000 1000 Up to 3 Fine sand, sandy 200 - 250 splay, Cabauw area, Rhine- surface) clay, silty clay Meuse delta, Netherlands (Weerts and Bierkens, 1993) Easternmost Holocene splay, Recent (Sub- Single Splay 1000 1000 3 - 4 Fine sand, sandy 200 - 250 Cabauw area, Rhine-Meuse surface) clay, silty clay delta, Netherlands (Weerts and Bierkens, 1993) Holocene (<5000 ybp), Recent (Sub- Composite 500 – 500 – 1 – 3 Very Fine to 500 – Channel depths 20 Lower Mississippi River surface) 1000 1000 Medium sand, clay 1000 -30+ meters Valley, near Ferriday, and silt (channel Louisiana (Aslan and Autin, belt) 1999) 8

Location/Formation Classification Type Splay Splay Splay Splay Grain-Sizes Channel Notes (Source) Length Width Height Width (meters) (meters) (meters) (meters) Various paralic sandstones, Ancient Composite 160 – 18 – 0.3 – 12 Not reported 20 - 5900 Channel depths no explicit locations given (n = 84) 11700 7700 (mean = 1.4 from 1 - 40 meters (Reynolds, 1999) from (mean = (mean = m) multiple 5577) 787) locations Upper Tangal Coal Ancient Composite 5000 Not 20 – 30 Fine – Medium in 1000 Channel depth ~ Measures, Bowen Basin, reported (presumably splay-channel (5.2 15 meters for Australia (Michaelson et al., multiple m thick), fine single story 2000) sequences) sandstone and siltstone Upper , Callide Seam Ancient Composite 4000 2500 Less than 4 Fine – Coarse 2800 Channel depths Member, Queensland, (max) sandstones and (channel less than 10 Australia (Jorgensen and sandy siltstones belt) meters Fielding, 2008) with coaly traces Late Pennsylvanian to Early Ancient Composite 100s – < 1000 < 10 Silty to pebbly < 1000 Channel depths 2 Permian, Cutler Formation, 1000s (stacked sandstone, fines – 5 meters north-central New Mexico deposits) upward (Eberth and Miall, 1991) Late Permian, Beaufort Ancient Composite 150 50 < 2 Fine-grained Not Group, Karoo Sequence, sandstones and reported South Africa (Smith, 1993) mudstones Upper Cretaceous, lower Ancient Composite Not 45.9 – 0.5 – 15 feet Very Fine to Fine 45.2 – Channel body Williams Fork Formation, (n = 279) reported 869.4 (median = sand 695.7 thicknesses of 3.9 Coal Canyon, Colorado feet 4.7 feet) feet – 29.9 feet (Pranter et al., 2009) (median (median (median 11.8 feet) = 214.9 202.7 feet) feet) Middle Ancient Single Splay 10 – 25 10 0.2 – 2 Very Fine to Not Formation, Baltic Basin, Medium sand reported southern Estonia to southern (bimodal), grades Lithuania (Ponten and Plink- into mudstone Bjorklund, 2007)

9

Figure 1.2: Splay Length vs. Channel Width. This figure consists of values compiled from literature and provided within Table 1.1. This chart excludes large composite data sets and systems where only the channel-belt width is presented. The values for the Amazon splays shown in Figure 1.5 are provided. Saskatchewan splays are highlighted (not specifically shown in Figure 1.5). 10

Figure 1.3: Splay Width vs. Channel Width. This figure consists of values compiled from literature and provided within Table 1.1. This chart excludes large composite data sets and systems where only the channel-belt width is presented. The values for the Amazon splays shown in Figure 1.5 are provided. Saskatchewan splays are highlighted (not specifically shown in Figure 1.5). 11

Figure 1.4: Splay-Deposit Thickness vs. Channel Width. This figure consists of values compiled from literature and provided within Table 1.1. This chart excludes large composite data sets and systems where only the channel-belt width is presented. Saskatchewan splays are highlighted (not specifically shown in Figure 1.5).

12

Figure 1.5: Comparison of Saskatchewan and Amazon Splays. (A) The Saskatchewan River system features splays that are large in size relative to channel width. Additionally, these splays feature a more distributary planform with multiple bifurcating channels. (B) The Amazon River system features splays that are small in relation to the channel scale and tend to form a single dominant finger-like channel prograding into a lake. (Images are from Google Earth and copyrights are held by Cnes/Spot Image (A), GeoEye (B), U.S. Geological Survey (B), and Digital Globe (B).)

13

2.0 Background

Crevasse splays have been described in a range of modern, Holocene, and ancient fluvial systems (Table 1.1). Splays are found in a wide variety of fluvial systems ranging from small tributaries (e.g. Cosumnes River, California, Florsheim and Mount, 2002) to the glacial outwash Mississippi (Blum et al., 2000), humid (e.g. Texas’s Colorado River, Blum et al., 2000) to arid (e.g. terminal splays in Lake Eyre, Australia, Lang et al., 2004) environments, and with grain-sizes ranging from clays (e.g. Rhine-Meuse Delta, Netherlands, Stouthamer, 2001; Hesselink et al., 2003) to cobbles (e.g. Northern Pennines, United Kingdom, Macklin et al., 1992). Ultimately, it is the interplay of these and other variables that controls splay development. Here, I review how different variables influence crevasse-splay development.

2.1 Controls on Splay Development

As overbank depositional products of fluvial systems, crevasse-splay development and growth is controlled by in-channel inputs, which supply the splay, and floodplain conditions, which influence how and where splay deposits accumulate. These boundary conditions will collectively determine the volume, area, and planview shape of a splay. As discussed in the Introduction, this study is primarily concerned with conditions, ceteris paribus, that cause changes in the volume and extent of splays in different systems. Thus, conditions which primarily exert control over planview shape, but have little influence on sediment supply through levee crevasses, are not considered primary controls on splay volume and mass balance. Many factors can influence the shape and specific planform geometry of crevasse-splay deposits. For example, vegetation and floodplain cohesion may play a pivotal role in some systems, but these variables will primarily manifest as modifications in sediment-routing and dispersal patterns across floodplains. For example, vegetation may increase the effective cohesiveness of floodplain deposits (e.g. Saskatchewan River, Smith et al., 1989) and/or act as a baffle creating low overbank-flow velocities and potentially low water-surface gradients (e.g. Okavango Fan, McCarthy et al., 1992; Smith et al., 1997). In some systems, the presence of large animals, such as hippopotamus or elk, create floodplain pathways and/or depressions, which leads to increased crevasse discharge and overbank-flow routing (e.g. Okavango Fan, McCarthy et al., 1992; Narew River,

14

Gradzinski et al., 2003; Klip River, Tooth et al., 2007). While these conditions are important for overall splay development, they primarily affect the shape of crevasse-splay deposits; this study focuses specifically on splay size rather than form.

2.1.1 Primary Controls on Splay Area and Volume

Here, I consider three conditions as primary controls on the sediment supply available for splay growth: (1) channel (and associated crevasse) discharge, (2) floodplain water-surface slope, and (3) channel grain size. While channel discharge is important in determining overall splay volume and area, within a given system, or between systems of similar scale, differences in splay size may be controlled largely by cross-floodplain water-surface gradient and the channel grain-size distribution. The importance of floodplain water-surface gradients lies in the balance between advective and diffusive transport processes, which determine the competence for floodplain sediment transport (Figure 2.1) (e.g. Pizzuto, 1987; Adams et al., 2004; Hajek and Wolinsky, 2012). Advective transport results from cross-floodplain currents capable of transporting sediment as suspended load or bedload (Adams et al., 2004), while diffusive transport results from turbulent processes that arise at the free shear boundary between rapid channelized flow and stagnant (or much slower) floodplain flow (Rajaratnam and Ahmadi, 1979). These turbulent processes produce eddies that propagate into the floodbasin and are capable of “diffusing” suspended sediment across the floodplain (Adams et al., 2004). In systems where the floodplain water-surface elevation is below that of the channel, the difference in these water-surface elevations results in a cross-floodplain water-surface gradient (Figure 2.1A). Steeper gradients lead to higher velocity advective currents (e.g. Adams et al., 2004; Hajek and Wolinsky, 2012). Thus, cross-floodplain advective-dominated transport is most common on wide and well-drained (“dry”) floodplains (Adams et al., 2004; Hajek and Wolinsky, 2012). Where strong enough, these advective currents can transport even fine sand onto the floodplain (Adams et al., 2004). While turbulent diffusive processes still occur on these floodplains, the importance of diffusive eddies is overshadowed by these cross-floodplain advective flows (Adams et al., 2004).

15

In laterally constricted and poorly drained (“wet”) settings (e.g. Hajek and Wolinsky, 2012), water surface on the floodplain rises in concert with channel stage, which inhibits the development of a water-surface slope away from the channel and strong currents onto the floodplain (e.g. Pizzuto, 1987; Adams et al., 2004; Hajek and Wolinsky, 2012). Instead, floodplain waters may be relatively stagnant, while channelized flow exiting a levee crevasse is much more rapid, resulting in the development of turbulent eddies at the boundary between the channelized and stagnant water, as shown in Figure 2.1B (e.g. Adams et al., 2004). These eddies can transport some transferred suspended sediment from the main channel, but the abrupt decrease in turbulence across the floodplain causes the coarser sediment to fall out of suspension within a short distance of the channel margin (Adams et al., 2004). Hajek and Wolinsky (2012) predict that this diffusive settling length is potentially around 1% of advective settling length. Given the various settings and processes (e.g. vegetation) for fluvial systems, it is likely that different floodplains exhibit different characteristic mixes of diffusive and advective transport processes (Adams et al., 2004). As proximal-overbank features, crevasse-splay deposition requires an appropriate sediment load and grain-size distribution that enables both (1) sediment transport beyond the channel’s banks and (2) sediment deposition proximal to the channel during flood events. Coarse sediments, relative to a channel’s competence, are transported as bedload and may never be suspended high enough in the water column to escape from the channel. Conversely, transport of excessively fine sediments as washload may continue overland until re-joining a channel or until the water reaches a ponded topographic low—perhaps on the distal floodplain where fine sediments may settle out in still water. Thus, crevasse-splay growth seems to require abundant, relatively coarse suspended grains, (e.g. particles ranging from coarse silt to fine sand for sand- bed rivers, i.e. slope O[10-4]). The Rouse (1937) number is a ratio of grain-settling velocity (a function of particle size, where larger grains generally settle faster) to shear velocity within a flow, and is determined by the equation 휔 푃 = 푠 [1] 퐾푢∗ where P is the non-dimensional Rouse number, ωs is the particle settling velocity, κ is von

Karman’s constant (approximately 0.41), and u* is the shear velocity, which is determined at the bed of the channel through the equation

16

휏푏 푢∗ = √ ⁄휌 [2] where τb is the shear stress at the lower boundary (i.e. bed/ground), and ρ is the density of water. The Rouse number is useful for describing when and how sediment of a given grain size will be transported in a given flow. When the Rouse number is large (e.g., P > 12.5), grains are too large to be transported. When Rouse number is very small (e.g., P < 0.5), particles will be transported entirely as washload (Table 2.1). As water escapes a levee it generally undergoes a sudden lateral expansion on the open floodplain that causes a drop in shear velocity and shear stress. As this happens, sediment can be transferred to the bed and ultimately deposited on the floodplain. The Rouse number is useful for considering how particle size and flow deceleration (influenced by floodplain drainage conditions) can both influence crevasse-splay deposition.

2.2 Relationship of Splays and Fluvial Avulsion

Given the morphodynamic formation of splays through crevasses, these deposits are closely associated with fluvial avulsions. In three of the most thoroughly studied Holocene and modern systems, the Mississippi River (e.g. Aslan and Autin, 1999), the Saskatchewan River (e.g. Smith et al., 1989), and the Rhine-Meuse Delta (e.g. Stouthamer, 2001), crevasse-splay development is closely linked to avulsion processes. In each of these systems, a significant portion of overbank deposition is attributable to the avulsion process and the building of coeval splay deposits (e.g. Aslan and Autin, 1999; Smith and Perez-Arlucea, 1994; Makaske, 1998). These splays consist of fine-grained sediments interspersed with coarser sand beds correlating with either more proximal deposits or crevasse- channels (Bridge and Demicco, 2008). Slingerland and Smith (2004) categorized fluvial avulsions into three types with two distinct end members, as originally described by Mohrig et al. (2000). The two end members are progradational and incisional avulsions (Figure 2.2). In progradational avulsions, the development of a new channel occurs when flow is diverted through a crevasse into a floodbasin, resulting in the loss of competence for sediment-laden flows and deposition of coarser suspended sediment as a splay. Progressive building of the channel leads to further progradation of this splay wedge across the floodplain in the downstream direction (Slingerland and Smith, 2004). In incisional avulsions, the development of new channel pathways progress upstream from 17

knickpoints through headward erosion of the floodplain. When this erosional pathway intersects the existing channel, it captures the channel’s discharge resulting in an avulsion (Mohrig et al., 2000; Slingerland and Smith, 2004). The third type—annexation avulsions—is a subset where avulsion occurs through re-occupation of a pre-existing channel or side channel, which may result in crevasse-splay deposits when the annexed channel is insufficient to carry the full discharge (Slingerland and Smith, 2004). Annexation avulsion is documented in the progradational Saskatchewan (Smith et al., 1998; Perez-Arlucea and Smith, 1999; Slingerland and Smith, 2004) and Rhine-Meuse systems (Stouthamer, 2001), incisional Baghmati River system (Jain and Sinha, 2004), and mixed progradational/incisional lower Rio Pastaza system (Bernal et al., in press). Slingerland and Smith (2004) conclude that annexation avulsions likely make up a portion of most major avulsion sequences.

2.2.1 Splays within Progradational Avulsion Systems

Splays produced during progradational avulsions take on a variety of shapes--including lobate, elliptical, and elongate--and stages of development (Smith et al., 1989; Slingerland and Smith, 2004). Within the Saskatchewan system, Smith et al. (1989) delineated the progression of development and maturity in splays into three stages. Stage I splays are small (< 1 km2) lobate splays deposited by sheet flow or poorly defined channels. Stage I splays progressively develop into Stage II splays, which are larger (up to 8 km2), have mature channels, and become interconnected with adjoining splays. Stage III splays are even larger (up to 20 km2), generally transport finer sediments than Stages I and II, and make up the distal, elongate extensions of Stage I and II splays that have combined to form a single large anastomosed unit (Figure 2.3). Smith and Perez-Arlucea (1994) estimated that 50-70% of the Saskatchewan avulsion belt consists of these Stage III splays. Similar development of avulsion-related crevasse-splays is observed in post-glacial Rhine-Meuse Delta deposits (Makaske, 1998; Stouthamer, 2001), in Holocene Mississippi River deposits (Aslan and Autin, 1999), and in Holocene Colorado River (Texas Gulf Coast) deposits (Aslan and Blum, 1999). In general, recent and modern progradational systems show extensive development of avulsion-related splays that fill a significant portion of the flood-basin within these systems (e.g. Smith and Perez-Arlucea, 1994; Makaske, 1998).

18

2.2.2 Splays within Incisional Avulsion Systems

Overbank deposition—including splays—is not an innate process of initial channel development during incisional avulsion, which contrasts with progradational avulsions (Mohrig et al., 2000; Slingerland and Smith, 2004). Rather, incisional avulsions likely occur in systems where floodplain erosion outpaces channel super-elevation and/or increases in cross-valley gradient. Given the close relationship between aggradation and the potential to develop splays (e.g. Bridge and Demicco, 2008), incisional avulsions are often presumed common in systems with little or no net aggradation (Slingerland and Smith, 2004). Notably, incisional avulsions are not restricted to degradational settings (Table 2.2), and net-aggradational systems avulsing by incision often lack crevasse-splay deposits. For example, the Narew River (Poland) is rapidly aggrading (~1- 1.5 mm/y), but splays and associated overbank deposits (e.g. levees and alluvial ridges) are not observed (Gradzinski et al, 2003). In contrast, splays are present within the rapidly aggrading Baghmati River system (India) (Jain and Sinha, 2004) and the Magela Creek system (Australia) (Tooth et al., 2008), both of which avulse by incision. The slowly aggrading “Channel Country” (Australia) features no along-channel splays, but does feature splay-like deposits at “waterholes” representing downstream terminal fans of channelized overland floodways (Gibling et al., 1998). However, splay-like features are observed near the initiation points of overland flow paths in the slowly aggrading, fine-grained (primarily clay and silt) Red Creek system (Wyoming, USA) (Schumann, 1989). In other low or non-aggradational settings, related overbank deposits—such as, levees and/or alluvial ridges—are present in the Ovens and Kings Rivers (Australia) (Schumm et al., 1996), Fitzroy River (Australia) (Taylor, 1999), and Klip River (South Africa) (Tooth et al., 2007), although splays are not specifically described in any of these systems and are probably absent from (at least) the latter two. Within incisional systems, the best documentation of splay development and overbank sedimentation is within the Baghmati River system—and this system differs markedly from progradational systems described in the previous section (2.2.1). One Baghmati splay deposited within a single flood season measured 200 meters in length and 100 meters in width for a channel of approximate bankfull width of 100 meters (Sinha et al., 2005), which is small in comparison to splays produced in systems of similar size (e.g. Table 1.1). Despite frequent crevassing, there is little evidence of progradational splay complexes (Sinha et al., 2005), and

19

“distal” floodplain deposits consisting of interbedded fine-grained sand and shelly clay were found only 200 meters from the main active channel (Sinha et al., 2005). Based on the small scale of splays and rapid accumulation of distal floodplain deposits, Sinha et al. (2005) argue that basin-filling in the system is primarily through "repeated flood deposits" rather than avulsion- associated deposition.

2.3 Crevasse-splays in the Stratigraphic Record

As a primary product of many fluvial avulsions, crevasse-splay abundance and characteristics may be key to understanding flood-basin filling processes within ancient fluvial successions. Jones and Hajek (2007) identified two types of fluvial stratigraphy, which they termed “stratigraphically transitional” and “stratigraphically abrupt.” Transitional successions consist of abundant splays, and other proximal overbank deposits beneath and lateral to paleochannel deposits, while abrupt successions feature paleochannel bodies directly overlying and encased by fine-grained “distal” floodplain deposits (Figure 2.4). Jones and Hajek (2007) also note that ancient systems dominated by transitional stratigraphy have abundant and extensive crevasse- splay deposits, while similar deposits are rare-to-absent in ancient deposits dominated by abrupt stratigraphy. These differences in splay proneness and extent suggest the primary controls for splay growth and development may be determined from ancient formations. Stratigraphically-transitional deposits feature paleochannel deposits underlain by coarsening-upward successions that include splay deposits (Jones and Hajek, 2007). Examples are found in the Paleocene Fort Union and Paleocene-Eocene Willwood formations of the Bighorn Basin, Wyoming (Jones and Hajek, 2007). In these Formations, Kraus and Aslan (1993), Kraus (1996), and Kraus (1998) identified several-meter thick, several-kilometer wide lithologically heterogeneous horizons with little or no paleosol development, which suggests rapid floodplain deposition. Ultimately, these authors linked these heterolithic horizons with the Stage III splays and avulsion sequences of Smith et al. (1989). Kraus and Wells (1999) further examined this relationship and determined these “heterolithic avulsion deposits” consist of 60% clay on average and feature no or limited erosion by or into the encasing well-developed, fine- grained paleosol horizons. Heterolithic horizons make up 59 and 65% of Fort Union Formation and Willwood Formation stratigraphic sections, respectively, and feature coarsening-upward

20

successions of claystone/mudstone to siltstone/sandstone sheets, and are always overlain by a major sheet sandstone paleochannel body (Kraus and Wells, 1999). In comparison to the Saskatchewan and Rhine-Meuse systems, the ancient formations have similar coarsening-upward successions overlain by channels (e.g. Smith et al., 1989; Perez-Arlucea and Smith, 1999), ribbon channel (i.e. Stage III) anastomosed patterns (e.g. Smith et al., 1989; Stouthamer, 2001), predominance of very fine-grained facies (e.g. Smith et al., 1989; Stouthamer, 2001) and avulsion-related deposits (e.g. Smith and Perez-Arlucea, 1994; Makaske, 1998). Kraus and Wells (1999) also discuss the similarity of the ancient deposits with other systems including progradationally linked Holocene deposits of the Colorado River in Texas (Aslan and Blum, 1999) and the Miocene Chinji Formation of Pakistan (Willis and Behrensmeyer, 1994). In contrast, stratigraphically abrupt formations feature coarse-grained paleochannel deposits positioned directly above floodplain deposits with no or little evidence of avulsion- related coarsening-upward successions (Jones and Hajek, 2007). Jones and Hajek (2007) identified an example in the Upper Cretaceous-Paleocene Ferris Formation (Hanna Basin, Wyoming), and similar successions were described by Mohrig et al. (2000) in the Oligocene Guadalope-Matarranya system (Spain) and the Eocene Shire Member of the Wasatch Formation (western Colorado). The lack of avulsion-related deposits means these formations either fully reworked these deposits or did not feature extensive crevasse-splay deposition during avulsion (Jones and Hajek, 2007). The latter possibility suggests these deposits formed in systems that primarily aggrade via overbank flood sedimentation (Jones and Hajek, 2007), and possibly in incisional avulsion systems (Mohrig et al., 2000; Jones and Hajek, 2007).

21

Figure 2.1: Floodplain Drainage Types (after Adams et al., 2004). (A) On well-drained floodplains, advective-sediment transport dominates as significant water-surface gradients (β) develop due to differences in channel and overbank water-surface levels, leading to advective currents away from the channel (arrows) capable of transporting sediment. This is common on laterally “wide and dry” floodplains (Hajek and Wolinsky, 2012). (B) On floodplains with standing water, diffusive-sediment transport occurs as nearly equal water surface levels between the channel and distal floodplain prevent the development of advective currents away from the channel. Instead, diffusive eddies arise parallel to the channel (as depicted by the spirals) due to differences in flow velocity between the channel and floodplain, which allows for limited transport of sediment through turbulence. This is more common on laterally “narrow and wet” floodplains (Hajek and Wolinsky, 2012).

22

Table 2.1: Sediment Transport Predicted by Rouse Number. The Rouse (1937) number is useful in predicting the sediment transport mode for a given grain- size and shear-velocity (Values from Shah-Fairbank et al., 2011). Rouse Number Range Sediment Particle Transport P ≥ 12.5 Little or no movement 12.5 ≥ P ≥ 5.0 Bedload 5.0 ≥ P ≥ 1.25 Suspended—significant contact with bed (mixed load) 1.25 ≥ P ≥ 0.5 Suspended—Less than 20% in contact with bed 0.5 ≥ P Suspended—100% wash-load

23

Figure 2.2: Progradational and Incisional Avulsion Diagrams and Stratigraphy (after Mohrig et al., 2000 and Hajek, personal file). (A) Progradational avulsions feature sediment-laden flow through crevasse(s) building prograding splay wedges (top, in brown) via anastomosed pathways (top). Eventually, these pathways coalesce into a new primary channel (middle) that captures the main channel’s flow. The nascent channel may rejoin this channel downstream (as shown via tributary) or cut an entirely different pathway and abandon the previous course. Stratigraphically, this produces interbedded and coarsening-upward sequences of sediment from the splay wedge (tan in bottom panel) correlating with the “heterolithic avulsion deposits” of Kraus and Wells (1999). (B) Early stage incisional avulsions (top) consist of overbank flows eroding floodplain pathways, and may follow previously developed topographic lows (brown). These flows may re-enter the main channel farther downstream at an erosional knickpoint (as shown) or cut a new pathway via topographic low. Over time overbank flow will cause headward erosion from the knickpoint/topographic low that eventually intersects the preceding trunk channel causing an avulsion. In stratigraphic cross-section this channel may consist of abrupt deposits (Jones and Hajek, 2007) where the channel sandstone and levee wings (yellow) would overlay floodplain deposits (purple and red).

24

Figure 2.3: Progradational System Splay Evolution (after Perez-Arlucea and Smith, 1999). (A) Crevasse(s) in main channel levee leads to nascent development of Stage I splay sand sheet via numerous distributary channels. (B) Primary conduits develop to transport bulk of overbank flows and these conduits begin to develop levees and an anastomosing pattern representative of Stage II splays. (C) Further anastomosis and elongate extension of distal mouth bars is representative of Stage III splays.

25

Table 2.2: Well Described Incisionally Avulsive Systems. System Avulsion Splays and other Grain Sizes Floodplain Aggradation Rates Mechanism Deposition Character Red Creek, Loss of channel capacity Small splaylike deposits near 62-99% silt and clay with Arid climate with grasses Not specifically mentioned, Wyoming, USA due to in-channel benches overbank flow initiation point median of 3 microns (p. and sagebrush (p. 277- but aggrading within (Schumann, 1989) (p. 283-284) (p.284) 281-282) 279), gradient 0.001- channel and on floodplain 0.005 (p. 286) (p. 277) Small tributaries Loss of channel capacity Apparently none and overbank Gravel alluvium with Occurs within hardwood Degradation (p. 222) of the Ohio River, due to in-channel flows are devoid of coarse grains fine to coarse sand forest (p. 222), Silt to Indiana, USA accretion (p. 226) (p. 226) matrix in channels (p. coarse sandy loam in (Miller, 1991) 223) floodplain (p. 223) Okavango Fan, Loss of channel capacity Splays and related deposition is Fine sand for bedload (p. Semi-arid, but swampy McCarthy et al. (1993) –all Botswana due to in-channel absent (p. 781), no levees form due 786-787) and very little with emergent aquatic sediment entering the fan is (McCarthy et al., accretion via excessive to lack of silt and clay (p. 791), suspended load, bedload vegetation (p. 779), 43% deposited and localized 1992) bedload and vegetation (p. rather form from vegetation and > suspended load (p.787- of channel flow is lost to calcite precipitation leads to 781, 791) peat (p. 793) 788) swamps (p. 786) island development and aggradation. Nearly 300 meters in total. (Smith et al., 1997) Loss of channel capacity Lacks splays, levees, banks, and Fine-medium sand with a Swampy, consists of See above due to in-channel mud-rich flood-basins (p. 51) paucity of silt and clay— papyrus (sedge) and reed accretion resulting from only 15% of load is grasses rooted in extensional faulting (p. suspended (p. 50-51) submerged peat (p. 51- 62-63) 52) Ovens and King Loss of channel No mention of splays, alluvial Gravel-bedded in Floodplain inundated Not Mentioned Rivers, Victoria, competence due to ridges and discontinuous levees are Brookfield Reach (~50 frequently (p. 1216), Australia (Schumm increased sinuosity and in- present (p. 1217), channels aggrade mm) and coarse sand and dense bank vegetation et al., 1996) channel accretion (p. for short periods of time (p. 1223) small gravel bedded in along mature channels 1221-1222) Pioneer Reach (p. 1215) (p. 1217) “Channel Loss of channel capacity Splays not observed, although Clay to fine sand, with Arid to semi-arid climate 8 - 12 meters of sediment Country,” east- due to in-channel benches small splaylike structures found at some medium to very prone to monsoonal over last 300 ka or ~0.03 central Australia restricting flow capacity waterholes (p. 602), small and coarse “sand” consisting weather (p. 597), mm/y (p. 598), ~40 meters (Gibling et al., (p. 610) wide levees (p. 611-612), pelleted of aggregated mud vegetation is sparse (p. of sediment in total (p. 599) muds breakdown and are deposited transported as bedload 598), fully covered 1998) on floodplain (p. 604) (p. 602-603) during floods except for aeolian dunes (p. 602), low water table (p. 609) Fitzroy River, Channel insufficient to Splays are likely rare or absent (p. Not specifically Inundation up to 3 meters Likely very low as Western Australia, carry high discharge 85), levees are intermittent or mentioned, but large in high discharge floods tectonically stable (p. 79), 3 Australia (Taylor, floods (tropical cyclones absent and “topographically sand-sized bedload and (~10000 m /s) on floodplain consists of up to 1999) and large monsoonal subtle” (p. 81) clayey suspended load floodplain approximately 8 meters of sediment above storms) and high velocity (p. 81) 6 km wide (p. 79), pedogenically modified overbank flows (p. 85) alluvium (p. 81)

26

System Avulsion Splays and other Grain Sizes Floodplain Aggradation Rates Mechanism Deposition Character Narew River, Loss of channel capacity No splays observed in area of study Mostly medium-coarse Peaty floodplain (p. 264) Regional subsidence ~2 Poland (Gradzinski due to in-channel (p. 262), levees and alluvial ridges sand in channels (p. 263) that is laterally narrow, 1- mm/y and net accumulation et al., 2003) aggradation (p. 271) also not present (p. 273) and peat-rich floodplain 4 km (p. 255) of peat is 1-1.5 mm/y (p. 264) mostly over 1-3 ka (p. 255) Baghmati River, Loss of channel Splays are observed and a high Very-Fine to Fine sand in Seasonally wet/dry with 0.7 -1.2 mm/y of India (Jain and competence due to in- ratio of sediment to discharge (p. channels, intercalated floodplain inundated on accumulation over ~1 ka (p. Sinha, 2004) channel aggradation and 166) with sandy silts and clays average 35 days at depth 166, after Sinha et al., tectonic adjustments (p. in overbank deposits (p. of 1.48 meters (p. 351 in 1996) 166-168) 164-168 in Sinha et al., Jain and Sinha, 2005) 2005) Klip River, eastern Loss of channel capacity Splays are not developed in study Medium sand to fine Seasonally wet/dry Locally aggrading, but South Africa (Tooth due to in-channel area (p. 455), small, discontinuous gravel as bedload, and wetland in region of system is net degradational et al., 2007) aggradation resulting from levees and alluvial ridges (p. 455) mud and fine sand as moderate rainfall, but over 100 ka (p. 459) increased sinuosity (p. suspended load (p. 459) high evaporation (p. 456) 454), less than 4 meters thick (p. 459) Magela Creek, Loss of channel capacity Splays are present, but no specific Fine and Medium sand, “Savannah-like” 1-2 mm/y for most of Northern Territory, due to in-channel description given (p. 1028) less than 15% fines (p. vegetation (p. 1024) and Holocene, but probably Australia (Tooth et vegetation growth leading 1023, 1029) wet/dry monsoonal lower now (p. 1023-1024) al., 2008) to deposition (p.1027) precipitation (p. 1023)

27

Figure 2.4: Stratigraphically Transitional and Abrupt Representations (after Jones, 2007). (A) Transitional stratigraphy features the coarsening-upward successions of “heterolithic avulsion deposits” (Kraus and Wells, 1999) consisting of finer—predominately silt—layers (greyish) and coarser—predominately very fine and fine sand—layers (tannish brown) overlain by channel deposits (tannish). Lateral splays similar to the coarse underlying splay sheets are also in abundance. This deposition style would feature extensive reservoir connectivity. (B) In abrupt stratigraphy, the channels are effectively incised directly into floodplain deposits (milky brown) due to the lack of heterolithic avulsion deposits, and lateral splays are either absent or rare with limited extent. Due to the predominance of floodplain mudstones, there is little reservoir connectivity.

28

3.0 Hypotheses and Study Phases

3.1 Hypotheses

3.1.1 Hypothesis I

Hypothesis I: Well-drained (dry and advective-transport dominated) floodplains will feature more extensive splay growth and deposition than poorly drained (wet and diffusive-transport dominated) floodplains.

Adams et al. (2004) observed that the poorly drained and diffusive-transport dominated upper Columbia River floodplain featured narrower and steeper levees than the well-drained and advective-transport dominated Saskatchewan floodplain and traced these differences to the dominant mechanism of sediment transport. Given similarities in the development of levees and splays, this difference would be expected to extend to splay deposition, as well. In well-drained systems, steeper cross-floodplain water-surface slopes drive suspended sediments away from the channel margin and across the flood basin, resulting in large splay areas for a given channel discharge. In contrast, floodplains with flood-water levels commensurate with channel stage will not have the cross-floodplain water-surface slopes necessary to transport sediment far away from the channel margin. These systems will be associated with crevasse-splay deposits with relatively small planform areas for a given channel discharge.

3.1.2 Hypothesis II

Hypothesis II: Systems featuring easily suspended and deposited “intermediate” grain-size distributions (generally coarse silt to fine sand for sand-bed rivers) will show more extensive splay development than systems with fine-grained (wash-load) or coarse-grained (bedload) dominated distributions.

High discharge of splay-forming sediments through a crevasse should lead to larger crevasse- splay areas for rivers of a given size. Intermediate grain-sizes are likely to both escape the channel flow and be deposited on the floodplain as flow expands away from levee crevasses. In contrast, while fine-grained sediments readily escape channel flow during floods, deposition of these sediments is a slow process requiring nearly stagnant flow, which is most common in distal

29

floodplain settings. On the other hand, coarse grain sizes typically occur as bedload and are not easily transported through crevasses onto the floodplain.

3.2 Study Components

Three study components contributed to testing these hypotheses. The first involved Google Earth mapping of splay deposits in several modern systems including the Saskatchewan River system of Saskatchewan, Canada, a portion of the upper Columbia River system near Castledale, British Columbia, Canada, the Ovens River near Wangaratta, Victoria, Australia, and the Sandover- Bundey River system in the Northern Territory, Australia (Figure 3.1). In the second component, field observations of channel, splay, and floodplain deposits in the Fort Union Formation (Paleocene, Bighorn Basin, Wyoming), Willwood Formation (Paleocene - Eocene, Bighorn Basin, Wyoming), and Ferris Formation (Late Cretaceous - Paleocene, Hanna Basin, Wyoming) are compared (Figure 3.2). The third component focused on modeling varying grain-size distributions and two end-member floodplain-drainage conditions (“dry” vs. “wet”) using Delft3D-FLOW software from Deltares Systems. The following sections detail each of these studies.

30

Figure 3.1: Location Map for Modern Systems. (A) Locations of the upper Columbia River (C) and Saskatchewan River (S) reaches within Canada. (B) Locations of the Sandover River (S) and Ovens River (O) reaches within Australia. (Images are from Google Earth and copyrights are held by Google (A), TerraMetrics (A), Cnes/Spot Image (B), Whereis Sensis Party Limited (B), US Department of State Geographer (both), Data SIO (both), NOAA (both), US Navy (both), NGA (both), and GEBCO (both).)

31

Figure 3.2: Location Map for Wyoming Field Sites (after Jones and Hajek, 2007). The Paleocene Fort Union and Paleocene-Eocene Willwood Formations are in the Bighorn Basin, Wyoming, USA, while the Late Cretaceous-Paleocene Ferris Formation is in the Hanna Basin, Wyoming, USA.

32

4.0 Observations in Modern Rivers

In order to evaluate how channel and floodplain conditions influence crevasse-splay development, I measured crevasse-splay dimensions and system characteristics in a range of modern avulsive systems using aerial photography within Google Earth. Specific measurements included length, width, shape, and orthogonal (to the channel margin) extent into the flood-basin for individual splays, and the length and area of channel reaches and floodplains. Collectively, these allow splay frequency and basin-filling properties to be characterized, which are important for determining how systems build levees and floodplains through overbank sedimentation, and, more broadly, may be key to understanding fluvial avulsions. To test my hypotheses, systems with varying grain-size distributions and floodplain drainage conditions were selected.

4.1 Study Areas

Four fluvial systems with extensively studied reaches and varied channel, splay, and floodplain characteristics were identified from literature. These systems include the anastomosed reach of the upper Columbia River in British Columbia, Canada, the Cumberland Marshes region of the Saskatchewan River in Saskatchewan, Canada, a floodout zone of the Sandover River in Northern Territory, Australia, and the Deep Creek reach and vicinity of the Ovens River in Victoria, Australia (Figure 3.1). All of these systems have undergone recent avulsions, and so are useful for evaluating the role of crevasse-splay deposition in channel avulsion.

4.1.1 Upper Columbia River

Makaske (1998) and Makaske et al. (2002) studied an anastomosed reach of the aggrading upper Columbia River near Castledale, British Columbia, Canada and mapped around 20 crevasse- splay deposits. This reach’s alluvial plain resides within an approximately 1.3 – 1.8 kilometer wide depression between the slopes of the Rocky Mountains (northeast) and Purcell Mountains (southwest) and consists of numerous channels and lakes (Figures 4.1 – 4.3). Given the laterally constricted floodplain and relatively wet environment, this system is poorly drained and standing water often accumulates across the floodplain during floods (Adams et al., 2004). Annual water discharge is around 108 m3/s (Water Survey of Canada, 1991, as referenced in Adams et al,

33

2004), but conditions are highly variable with high early summer flows due to snowmelt and rain (maximum discharge around June) and low winter discharges, when the channel contains ice (minimum discharge around February) (Makaske, 1998; Tabata and Hickin, 2003). The main channel mostly carries coarse and very coarse sand as bedload and medium sand as suspended load (Makaske, 1998; Abbado et al., 2005). Significant sediment extraction occurs within the anastomosed reach, as evidenced by bedload transport dropping from 7.1 kg/s to 0.6 kg/s in the 50 kilometer reach between Spillmacheen and Nicholson for measurements at near bankfull flood conditions (~220 m3/s) (Locking, 1983, as referenced in Makaske et al., 2009). This sediment extraction leads to in-channel aggradation of 0.9 – 20.9 mm/year and levee aggradation of 1.65 – 5 mm/year, with both rates decreasing downstream (Makaske et al., 2009). The measured bedload transport only accounts for 11% of total transport with the rest consisting of suspended and wash-load (Locking, 1983, as referenced in Abbado et al., 2005 and Makaske et al., 2009). Flood-basin aggradation estimates range from 1.7 mm/year (Makaske, 1998; Makaske et al., 2002) to 2.2 mm/year (Adams et al., 2004). The average downstream surface slope for the reach is approximately 15 cm/km (Adams et al., 2004). Makaske et al. (2002) estimated an avulsion frequency of approximately 3 per 1000 years, and, presently, there is an ongoing avulsion occurring within the downstream half of the anastomosed reach (Abbado et al., 2005).

4.1.2 Saskatchewan River

Studies by Smith et al. (1998) and Perez-Arlucea and Smith (1999) identified and characterized splays along the New and Centre Angling Channels of the Saskatchewan River (Figure 4.4). In this area (Cumberland Marshes), the floodplain for the entire system is up to 12 kilometers wide (Perez-Arlucea and Smith, 1999), and Adams et al. (2004) noted that water-surface elevations on the floodplain are below those of the channel, so they concluded this system is dominated by advective transport processes as seen on well-drained floodplains. Average annual water discharge is estimated at 450 – 500 m3/s with annual maxima between 1000 – 1400 m3/s occurring in July, following the spring thaw (Adams et al., 2004). Deposition within the Centre Angling reach was ongoing in 1945 and mostly complete by 1977 based on aerial photography (Perez-Arlucea and Smith, 1999). Grain sizes within channels and proximal splays are generally very fine to medium sand, but the overall most abundant grain size is medium to coarse silt

34

encountered in levees and most splays, with finer silt, clay, and peat formation observed in more distal and isolated settings (e.g. Smith et al., 1989). A Water Survey of Canada gauging station between Tobin Lake (Squaw Rapids Dam) and the study area recorded an annual suspended/wash-load transport rate of 2.9 kg/s, although this value is roughly 1% of the pre-dam (i.e. pre-1962) transport rate at Cumberland Marshes (Ashmore and Day, 1988). Smith et al. (1989) recorded a radiocarbon date of 2270 years before present for a peat layer beneath a 6.6 meter thick sequence of crevasse splay and levee deposits, which suggests an average aggradation rate of 2.9 mm/year. The average downstream surface slope within the Cumberland Marshes is approximately 10 cm/km (Adams et al., 2004). The Saskatchewan River is undergoing an avulsion that began in the early-mid 1870’s (e.g. Smith et al., 1989; Smith et al., 1998) and these New Channel (upstream of Steamboat Channel bifurcation) and Centre Angling Channel (downstream of the bifurcation) segments are emerging as the predominate thread following a heavily anastomosed phase (Smith and Perez-Arlucea, 2008) (Figure 4.4).

4.1.3 Sandover River

The Sandover River near Ammaroo, Northern Territory, Australia is a slowly aggrading, ephemeral, dryland system with documented splays (Tooth, 1999a; 2005) and avulsion (Tooth 1999a; 1999b; 2000a; 2000b; 2005; Tooth and Nanson, 1999; 2000). According to Tooth (1999a, p. 95), “floodout” is a common Australian term to designate an ephemeral stream reach where channelized flows end, but periodic floods still transport water downstream across the alluvial plain. In the floodout zone (the focus area for this and previous studies), the Sandover features a relatively wide (~6 kilometers), silty sand (30-40% silt-clay) floodplain and only periodic high discharge floods (every 3-4 years) (Tooth, 1999a) (Figure 4.5A). The active channel is generally 1 - 3 meters deep with steep banks, and transports medium to very coarse sand with gravel (Tooth, 1999a; 2005). Splays in the floodout zone are generally less than 0.5 km2, and feature deeply erosive channels (up to 2 meters deep) in the proximal-medial reaches and limited inter- channel deposition (Tooth, 2005). The distal ends of many splays consist of up to 0.4 meter high sediment lobes (Tooth, 2005). Similar depositional patterns are observed at the channel floodouts, although these features are generally larger in areal extent, and may contain finer sediments (Tooth, 1999a; 2005). Holocene alluvium is up to 6 meters thick in the study area

35

(Tooth, 1999a), suggesting aggradation rates of ~0.4 mm/year or less. Calculations using Google Earth yielded a downstream surface slope of 0.0011 (110 cm/km) for the studied reach, which falls in the range of 0.0005 – 0.002 for the arid Northern Plains provided by Tooth (1999a). In this area, there are two observed, recently active floodouts due to a 1974 avulsion that resulted in the infilling of the previously active channel reach with silty sand and very fine to medium sand (Tooth, 1999a).

4.1.4 Ovens River

Unlike the other studied rivers, no previous studies document the presence of splays on the avulsive Ovens River (e.g. Binnie and Partners, 1984; Rippin and Sheehan, 1986; Schumm et al., 1996; Cottingham et al., 2001; Judd et al., 2007), although the presence of alluvial ridges, natural levees, and channel margin crevasses is documented (Binnie and Partners, 1984; Schumm et al., 1996; Judd et al., 2007). While portions of the approximately 5 kilometer wide Ovens floodplain are anthropogenically modified through flood abatement strategies and agriculture (Binnie and Partners, 1984; Cottingham et al., 2001), an approximate 1.5 kilometer reach abandoned in 1974 (Judd et al., 2007) is evident in Google Earth, suggesting that the floodplain is not modified enough to obscure recent depositional and/or erosive features (Figure 4.5B). Channel discharge for the study reach is not specifically known, although stream gauges approximately 15 kilometers upstream (Rocky Point) and downstream (Wangaratta) have average instantaneous water discharges of 42.9 m3/s and 75.7 m3/s, respectively (Victorian Water Resources Data Warehouse, online data). This study examines the Pioneer, Deep Creek, and the Tarrawingee Reaches downstream of the Pioneer-Deep Creek bifurcation, because this area features a reach being abandoned (Pioneer), a reach developing through headward erosion (Deep Creek), and an active, mature reach (Tarrawingee) (Binnie and Partners, 1984; Schumm et al, 1996; Judd et al., 2007). Specific sediment load values are not known, although the river is classified as mixed load and floodwaters are relatively deprived of sediment (promoting headward incision of new channels) (Schumm et al., 1996). Vertical accretion is a slow process along the studied reach of the Ovens River and no estimates are provided (e.g. Judd et al., 2007). Downstream surface slope for the studied reach is approximately 0.0018 (180 cm/km) based on measurements within Google Earth, which is equal to the bedslope of the relatively straight and young Deep Creek

36

channel (Schumm et al., 1996). In this area, the channel beds consist of coarse sand and gravel and channel banks and floodplains consist of clay through gravel (Binnie and Partners, 1984; Schumm et al., 1996; Judd et al., 2007). Despite the wide apparent floodplain, the Pioneer Reach is flooded multiple times per year (Judd et al., 2007), suggesting poor drainage.

4.2 Google Earth Observations and Measurements

For all rivers, reaches from published studies were located in Google Earth, and previously mapped splays and channel reaches were identified (predominately from: Makaske, 1998; Makaske et al., 2002; Smith et al., 1998; Perez-Arlucea and Smith, 1999; Tooth, 1999a; Tooth, 2005; Schumm et al., 1996). The character and surface presentation of previously identified splays (where present) was used a basis for splay identification in unmapped areas adjacent to mapped splays. The previously mapped splays and reaches are presented in Figures 4.1 – 4.5. Measurements for each splay included length (along the primary crevasse-channel where apparent), width (generally orthogonal to crevasse-channel), parent channel width, orthogonal extent of the splay (perpendicular distance from the parent channel margin to the edge of the splay), and the latitude and longitude (WGS 84 coordinates) of the origin and terminus. Following mapping, each splay’s polygon was saved in the .kml format provided within Google Earth. A representation of these measurements is presented as Figure 4.6. Once splays were mapped several other parameters were calculated to determine splay frequency and the percentage of the flood-basin filled by splays. These additional parameters include the length of the channel reach(es), and the area of channel(s), island(s), and flood-basins or floodout zones (i.e. the maximum extent of flooding for an ephemeral channel). Categorizing channel reaches required some interpretation on the classification of bifurcations around islands. This determination was made based on the daughter channel’s width (w) and length (l). Where the w/l ratio is greater than 10, the daughter channel is considered a separate reach and is mapped and measured separately. Such determination was most necessary in the Columbia and Saskatchewan systems. Flood-basin boundaries (which also define study areas) are identified as the apparent orthogonal extent of floodwaters for the studied channel reaches, and this varies among systems. The Sandover is based on the limits of the floodout zone for the currently and recently active

37

channels of Tooth (2005). Defining flood-basins for channel reaches is harder in the other three systems because of the scale and complexity of the systems. Measurements in the Saskatchewan system focused on a partial reach of the New Channel/Centre Angling Channel (from the confluence with the Torch River to the confluence with the South Angling Channel) where the extent of the flood-basin was defined by the levees of the nearest adjacent primary channels (primarily the North Angling channel to the north and the South Angling or Gun Creek channels to the south). For the Ovens River, flood-basin extent was defined by the approximate floodplain limit to the north and the levee of the nearest abandoned channel (i.e. the method used for the Saskatchewan River) to the south. For the Columbia River, the floodbasin of the main channel (in the proximity of Castledale) was defined using the methods for the Saskatchewan and Ovens rivers. Since portions of the defined floodbasin are closer to adjacent, unstudied channels within these three systems, calculation of basin filling utilizing this definition might be seen as minimum estimate. With this in mind, an additional set of measurements was made for the upper Columbia utilizing the full width of the floodplain near Castledale, which is the area studied in detail by Makaske (1998) and Makaske et al (2002) (Figures 4.2 and 4.3). This floodplain width reach serves as a check on the values obtained using the method outlined for the Saskatchewan, Ovens, and Columbia’s main channel reaches. Mapped lengths (i.e. splay channels, splay widths, parent channel widths, cross-sectional extent of splays, and channel reaches) and areas (i.e. splays, channels, islands, and flood-basins) were determined using the “Calculate Polygon Area” calculator at www.earthpoint.us. Further calculations to determine coverage percentages used simple arithmetic and statistical hypothesis testing to compare the studied fluvial systems utilized the Student t-test.

4.3 Google Earth Study Results

4.3.1 General Appearance of Splays

In the upper Columbia, larger splays show an elongate, “finger-like” appearance while smaller splays may show either an elongate form or a lobate form (Figure 4.7). In general, splays within the Columbia are widest and have the most channels near their origin and splays narrow and become single thread closer to their terminus. Splays that are poorly vegetated and appear sandy

38

are generally smaller with a more lobate appearance. Some splays show interconnectedness with adjoining splays, but most are isolated enough that wide flood-basin areas separate splays. Splays along the Centre Angling Channel of the Saskatchewan River follow the descriptions of Smith et al. (1989) (Figure 4.8). Smaller splays are normally isolated bodies with a lobate appearance and possibly anastomosed internal channel pattern, while larger splays show this internal anastomosed pattern and are connected to adjoining splays, including merging of primary crevasse-channels. Despite large-scale splay interconnectedness, individual splays are relatively easy to identify within the group. Unlike the Columbia system, the splay channels show a distributary pattern with a single or few channels near the origin, multiple bifurcations along the splay’s length, and frequent merging of these distributaries to produce a downstream anastomosed pattern. Sandover River splays share some similarities with splays in the Saskatchewan system, but also some key differences. Splays are generally lobate, contain numerous, bifurcating channels, and are frequently connected to adjoining splays. However, single splays are hard to identify unless isolated (Figure 4.9). This is due to multiple bank breaches within short distances along channels. Thus, the more appropriate term for cataloged splays may be “splay complexes,” which consist of multiple small, highly interconnected splays in close proximity. For these reasons, a different categorization could lead to a vastly different splay count. The two documented floodouts of Tooth (1999a; 2005) are the features most similar to splays within the Saskatchewan system. These floodouts start as a few large distributaries that undergo many bifurcations and re-connections to show an anastomosed pattern downstream. Finally, portions of the floodout zone consist of pockets of banded reddish coloration similar to portions of active and recent splays (Figure 4.9), and these features might be representative of older weathered splay deposits. The Ovens River system contained no identified splays along any of the three studied reaches (developing Deep Creek anabranch, moderately mature Tarrawingee reach, and abandoning Pioneer reach) despite having an obvious pathway for the 1974 local avulsion and showing evidence for floodplain incisions. Figure 4.10 includes aerial documentation of these features.

39

4.3.2 Analysis of Splay Shape

For each of the systems the number of identified splays, length of channel reach(es), and areal coverage of splays, channels, and other related features was determined and selected properties are provided in Table 4.1.

4.3.2.1 Splay Frequency and Basin Filling

Of the three systems with identified splays, there are significant differences in both the system’s propensity to develop crevasse splays and the basin-filling properties of these splays. Despite having a very small study area, the upper Columbia system creates the most splays per kilometer of channel length (2.10 km-1) and the Saskatchewan the least (0.878 km-1), as shown by the splay frequency statistic in Table 4.1. However, despite the upper Columbia’s increased splay creation propensity, Saskatchewan splays fill much more of the floodbasin, covering 51.2% of the basin (as defined above). A middle value of 35.2% was obtained for the Sandover system, although this system’s splays feature significant erosion and only limited deposition, primarily concentrated near the distal splay edge (Tooth, 2005). The upper Columbia—despite a propensity to create many splays—features splays that collectively only fill 17.6% of the studied basin. Additionally, similar values for splay frequency (1.82 km-1) and basin-filling (15.1%) in the upper Columbia were determined utilizing the entire floodplain width (i.e. Castledale Reach, as depicted in Figure 4.2), suggesting these results are not the product of sampling method. Collectively, these results corroborate qualitative observations made regarding the proximity and interconnectedness of splays in the preceding section, and may have implications for the predominate style of basin-filling within each system (splay development vs. flood-basin aggradation). Ultimately, the key results are that the Saskatchewan is producing expansive basin- filling splays, the Sandover is primarily producing erosive splays within the floodout zone, and the upper Columbia is producing many small splays that resemble nascent channels. Previous studies also examined the basin-filling characteristics of splay-related deposition in the Saskatchewan (Smith and Perez-Arlucea, 1994) and upper Columbia (Lavooi, 2010) systems. In the Saskatchewan, Smith and Perez-Arlucea (1994) estimated that large scale splay complexes comprised approximately 50 - 70% of the avulsion belt based on aerial photos and borings, which is in line with this study’s result of 51%. Additionally, given this study’s 40

conservative flood-basin definition and the New Channel’s peripheral location within the avulsion belt, it’s possible that splay basin-filling is under-represented in this study. For the upper Columbia system, Lavooi (2010) used borings along three profiles and aerial photography to determine facies proportions. She found that crevasse-splays made up approximately 4% - 20% of profiles, and that crevasse-splay deposition decreased downstream from Harrogate to Parson, which is also corroborated by Abbado et al. (2005). This study’s estimate of approximately 18% is within the range of her estimates.

4.3.2.2 Splay Sizes and Shapes

Measurements and calculations also documented individual crevasse-splay parameters including splay areas (Figure 4.11), parent channel widths (Figure 4.12), orthogonal extents into the flood- basin (Figure 4.13), maximum splay widths (Figure 4.14), and splay channel path lengths (Figure 4.15), with normalization of orthogonal extent by channel width (Figure 4.16) and determination of aspect ratio (path length divided by maximum width) (Figure 4.17). Summary statistics (mean, median, and standard deviations) for all of these parameters are presented in Table 4.2. Additionally, statistical hypothesis testing utilized Student t-tests (Table 4.2), and tested the null hypothesis that systems have the same parameter means. In each of the three systems, the relationships between splay area and parent channel width (Figures 4.18 and 4.19), orthogonal extent and parent channel width (Figure 4.20), and between splay channel path length and maximum splay width (Figure 4.21) were plotted. Collectively, these sets of measurements not only serve to document splay characteristics within the three fluvial systems, but also to provide a means of comparison among these systems. Review of Figures 4.11 – 4.15 and Table 4.2 reveals the differences in crevasse-splay and system scale between the upper Columbia and the other two splay-producing systems. Splay areas in the Columbia are generally O[103] to O[104] meters2 while they are primarily O[105] to O[106] meters2 in the Sandover and Saskatchewan. Similar relative discrepancies in raw values generally hold for the other measured parameters, where high-end values in the Columbia are comparable to low-end values in the Sandover and Saskatchewan. Two additional observations from these plots and Table 4.2 include (1) the Sandover and Saskatchewan feature significant overlap in their raw value ranges, and (2) the plotted data are nearly log-normally distributed.

41

Using a t-test of the logarithm of splay area, it is calculated that there is greater than 95% confidence that mean splay area in the Columbia is different from that of either the Sandover or Saskatchewan. However, there is no statistically significant difference between the Sandover and Saskatchewan. Almost identical results are obtained when comparing the logarithms of parent channel widths. Thus, there is statistical evidence at the 95% level corroborating a difference in splay area and physical system scale for the Columbia when compared to the other two systems, but little evidence that there is statistical difference between the Sandover and Saskatchewan. Given these differences in system scale, it is appropriate to compare normalized values within the systems. Normalized orthogonal extent (by channel width) is presented in Figure 4.16, and it shows that most splays perpendicularly extend between 1 and 10 channel widths into their respective flood-basins in all systems—a rather wide range of values. Taking the logarithm of normalized orthogonal extent and performing t-tests suggests there is no statistically significant difference between the three systems for this measure (Table 4.2). The spread of splay aspect ratios is a bit tighter than for normalized orthogonal extent, and most values are within the 1 to 6 range (Figure 4.17). T-tests performed on the logarithm of aspect ratio indicate Columbia and Sandover have statistically equivalent shapes (aspect ratios), but the difference between these two systems and the Saskatchewan is significant at the 95% confidence interval (Table 4.2). The Saskatchewan River is more likely to produce long and narrow crevasse-splays (i.e. higher aspect ratio). This result may be attributable to this system’s tendency to produce compound splays featuring numerous smaller individual splays. Figures 4.18 – 4.21 document relationships between measured parameters for individual splays. Figures 4.18 – 4.19 show that there is virtually no correlation between splay area and channel width in any of the three systems measured. Likewise, orthogonal splay extent and channel width are unrelated in the Columbia and Sandover rivers, and shows only a weak positive relationship in the Saskatchewan River (Figures 4.20 – 4.21). In summary, channel width is a relatively poor predictor of splay area and extent for the Columbia, Saskatchewan, and Sandover Rivers.

42

4.3.3 Splay Volume Estimates

Using crevasse-splay thickness values presented in previous studies and the results presented in Sections 4.3.1 and 4.3.2, it is possible to project the sediment volume deposited within splays for these three systems and more accurately evaluate basin-filling adjusted to system scale (i.e. discharge). Within the upper Columbia, Machusick (2000) documented the planform and sediment thicknesses of two modern splays, and other authors (e.g. Makaske, 1998; Makaske et al., 2002; Makaske et al., 2009; Lavooi, 2010) prepared basin-wide cross-sections that document and interpret crevasse-splay thicknesses for the modern and sub-surface. Machusick’s maps indicate the tops of the Soles and Shalla Galla splays are mostly between 0.6 – 1.2 and 0.4 – 1.0 meters above the surrounding flood-basin, respectively, and eight cores within the Shalla Galla splay indicated thicknesses between 25 and 103 centimeters, suggesting similarity between measured surface levels and thicknesses. The Shalla Galla splay is within the “Main” Channel flood-basin studied in Section 4.3.1. Trenches at Soles splay identified underlying flood-basin sediments within 65 centimeters of the surface. Mapped crevasse-splay thicknesses within other studies (Makaske, 1998; Makaske et al., 2002; 2009; Lavooi, 2010) are up to ~2.5 meters thick, but are mostly between 0.5 and 1.5 meters, with the thickest deposits possibly representing multi-storied deposits. Machusick (2000) and Makaske et al. (2002) reviewed aerial photos that indicate the splays developed on ~30 year timescales. Smith and Perez-Arlucea (1994), Perez-Arlucea and Smith (1999), and Farrell (2001) studied crevasse-splays of various sizes, ranging from Stage I – Stage III development, within the Saskatchewan River system. The Stage III Windy Lake splay formed within a 35-year period and is mostly between 1.5 and 3 meters of thickness with an average thickness of 1.8 meters (Smith and Perez-Arlucea, 1994). The Stage III Cadotte complex is generally 1.5 – 3 meters thick with an average thickness of 1.9 meters (Perez-Arlucea and Smith, 1999). This splay’s development time is poorly constrained, but it is fully developed on the first aerial photos of the region taken approximately 65 years after the avulsion’s start (Perez-Arlucea and Smith, 1999). Farrell (2001) mapped a small, Stage I splay off the Centre Angling Channel that is mostly 2 – 3 meters thick and fully developed within 25 years. This splay is within the studied reach analyzed in Section 4.3.1.

43

Tooth (2005) studied splay development within several dryland systems in central Australia, including the Sandover system. These dryland splays typically feature splay channels incised up to 1.5 meters into the floodplain, with gradual decrease in incision depth towards the distal splay margins where net deposition is up to approximately 0.4 meters. In areas between these incised splay channels, net deposition ranges from 0 – 0.3 meters with some splay channels being flanked by thin levees. Average deposition is not mentioned by Tooth (2005), but based on his descriptions and the low aggradation rate in this system (Section 4.1.3), it is likely that average splay thicknesses are no more than approximately 0.1 meters. Splay development may occur within single or few flood events, and the full timescale of splays evolution is likely a few decades as some splays observed in 1950’s aerial photography are no longer apparent (Tooth, 2005). Based on estimated average-splay thicknesses of 0.8 meters, 1.8 meters, and 0.1 meters, in the Columbia, Saskatchewan, and Sandover systems, respectively, the volume of crevasse- splay sediment in the studied reaches (Section 4.3.1) is estimated and presented in Table 4.3. Additionally, a normalized splay volume ratio is presented for the Columbia and Saskatchewan. This ratio is calculated as

(Average Splay Area [푚2] ∗ Average Thickness [푚] / Average Development Time [푦푒푎푟푠]) [3] Sediment Discharge [푚3/푦푒푎푟] where the development time is estimated as 30 years for both systems. Estimated sediment discharges are 0.041 m3/s and 0.181 m3/s for the Columbia and Saskatchewan, respectively, based on estimated sediment fluxes of 65 kg/s (estimated from Locking, 1983, as referenced in Abbado et al., 2005 and Makaske et al., 2009) and 290 kg/s (estimated pre-dam construction from Ashmore and Day, 1988), respectively, and sediment bulk densities of 1,600 kg/m3. These calculations show that the area of an average sized Columbia splay is two orders of magnitude less than in the Saskatchewan, and that this difference is not fully attributable to system scale, as normalized sedimentation in the Columbia is an order of magnitude less than in the Saskatchewan (Table 4.2). In other words, the relative diversion of Columbia River sediments into crevasse-splay deposits is an order of magnitude less than the relative diversion of river sediments into crevasse-splay deposits in the Saskatchewan system.

44

4.4 Discussion

Study of modern avulsive systems makes it apparent that there are significant differences in crevasse-splay shapes, scales, and basin-filling patterns even among similarly sized systems. The preceding results demonstrate that there are clear differences in splay propensity and basin coverage in avulsive fluvial systems. The upper Columbia system produces many small lobate and elongate finger-like splays, but the splays collectively cover less than 20% of the basin and feature extensive inter-splay areas of overbank deposition. On the other hand, despite producing fewer splays per given reach length, the Saskatchewan system features large, distributary-like splays that cover more than half of the flood-basin. Within the Sandover system, splay development is common and basin-coverage characteristics are intermediate between the Columbia and Saskatchewan, although splay channels themselves are highly erosive and splays generally feature only limited deposition at their margins and between splay channels. There is high interconnectivity between the erosive channels. Splays are not observed in the Ovens River system. The upper Columbia and Saskatchewan systems have similar net aggradation rates, longitudinal bedslopes, and have comparable sediment loads adjusted for system scale. Despite these similarities, splays within the Saskatchewan are much larger in planform and volume, and fill much more of the floodbasin than splays in the upper Columbia system. Adjusting for estimated sediment discharge in each system, the Saskatchewan splay volume is an order of magnitude larger than in upper Columbia splays, which means the discrepancy in areal size is not attributable to poor dispersal of overbank sediments leading to overly thick splays in the upper Columbia. Instead, the discrepancy between splay size (area and volume) in the Saskatchewan and upper Columbia may be attributable to processes that limit sediment supply from the channel to the floodplain. The upper Columbia features coarser grain sizes than the Saskatchewan (medium – very coarse sand versus very fine – medium sand), which may limit the amount of suspended material supplied to the floodplain through levee crevasses. Additionally, the upper Columbia features a poorly drained, laterally constricted floodplain, whereas, the Saskatchewan features a well-drained, wider floodplain (Adams et al., 2004). This difference would limit the potential for advective transport of overbank sediment within the Columbia, and may promote crevasse healing in the upper Columbia. Collectively, the differences in splay-deposit volume

45

relative to estimated sediment discharge between the upper Columbia and Saskatchewan systems implies that overbank sediment-supply to floodplains in the Columbia River is suppressed relative to the Saskatchewan. This may be the result of grain-size or floodplain-drainage differences between the two systems, or a combination of both.

46

Figure 4.1: Upper Columbia River Study Reach with Mapped Splays. Flow is from southeast to northwest (right to left). (A) The “main” channel’s upstream reach is from near Harrogate to Castledale and features numerous splays, although splays are generally small and the reach features large intra-splay flood-basins. Splays mapped by Makaske (1998, his Figure 3.4) and repeated as Makaske et al. (2002, their Figure 4) are shown in yellow outline. Also note that this channel is not the predominate channel in the most upstream portion of the entire system reach (i.e. far southeast or far right in figure). (B) The downstream reach is similar to that of A, but fewer channels are observed in this downstream portion of the entire system reach. (Images are from Google Earth and copyrights are held by Province of British Columbia (A and B) and Image Parks Canada (A).)

47

Figure 4.2: Splays Mapped near Castledale. This figure identifies splays across the entire floodplain for the reach near Castledale, BC. This figure corresponds almost exactly with the area study of Makaske (1998, his Figure 3.4) and repeated in Makaske et al. (2002, their Figure 4), and documents splays identified within those studies. Notably, this study identified 43 splays within this reach (23.6 kilometers combined for identified channels), which is many more than those studies; however, this study identified several deposits as splays that were previously identified as levee extensions within those studies. In some cases, these bodies obviously grew between their identification and the date of this photo (circa 2005). Splay frequencies and splay basin-filling proportion are similar for this reach and the Main Channel Reach documented in the text, Table 4.1, and Figure 4.1. (Image is from Google Earth and copyrights are held by the Province of British Columbia.) 48

Figure 4.3: Comparison of Main Channel and Castledale Reaches. This figure shows the location of the Main Channel (Figure 4.1) and Castledale Reaches (Figure 4.2) as a means of comparison. (Image is from Google Earth and copyrights are held by the Province of British Columbia and Image Parks Canada.)

49

Figure 4.4: Saskatchewan River Study Reach with Mapped Splays. Flow is from southwest to northeast (left to right). (A) Smith et al. (1998, their Figure 4E) identified several splays (pink outline) along the New Channel of the Saskatchewan River, which is this study’s upstream reach. This main thread continues as the Centre Angling Channel in the northern portion of A. Note the relative dearth of splays upstream of the Torch River confluence (far left), which is due to this area being on the periphery of the avulsion zone, and these splays are excluded from the reach analysis, although they are included in quantification of splay attributes (e.g. aspect ratio). (B) Perez-Arlucea and Smith (1994, their Figure 2B) identified several splays and associated abandoned channels along the Centre Angling Channel. Note the large whitish green areas between splays, which are not mapped with any specific splay, but may be associated with distal splay development (e.g. “fen” in Figure 2.3). (Images are from Google Earth and copyrights are held by Cnes/Spot Image (A and B) and DigitalGlobe (B).) 50

Figure 4.5: Sandover River and Ovens River Study Reaches with Mapped Splays. (A) Flow is from west to east in the Sandover system (left to right). The Sandover River’s upstream floodout zone features numerous splays, most identified by Tooth (1999a, his Figures 9A and 13) and some also shown by Tooth (2005, his Figure 4A). Notable within this system is that major channels end in floodouts (purple outlines) at the downstream end of the study area, although unconfined flow continues downstream, as shown by the whitish tan coloration of the ground surface. (B) Flow is from southeast to northwest in the Ovens system (right to left). The three studied channel reaches—Pioneer (yellow), Tarrawingee (orange), and Deep Creek (white)—are outlined and numerous abandoned channel segments and reaches are evident on the floodplain (both within and outside the study area). Notably, no splay deposits are evident within this system. (Images are from Google Earth and copyrights are held by Cnes/Spot Image (A and B), DigitalGlobe (A), and GeoEye (B).)

51

Figure 4.6: Splay Measurements Made within Google Earth. Numerous measurements were made within Google Earth. These included measuring the splay length, which was typically done by measuring along the primary splay channel’s path, where evident. Other measurements included the coordinates (WGS 84) for the splay’s origin and terminus, splay width, parent channel width at the splay site, and the orthogonal extent. The orthogonal extent measures the width of the splay’s extent into the flood-basin perpendicular to the parent channel. (Image is from Google Earth and copyright is held by Province of British Columbia.)

52

Figure 4.7. Splay Examples from the Upper Columbia River.

(A) Small lobate splays are approximately 100 x 100 meters or less and feature a handful of small crevasse channels. (B) Elongate splays feature a fan-like origin and finger-like channel extension. Length scales are generally several hundred meters for these splays. Within both types, a portion of the mapped splays are likely below water level during floods. Machusick (2000) linked this dimension and planform variability to whether flood-basins contained downstream outlets that acted to guide flow, leading to elongate planforms. (Images are from Google Earth and copyrights are held by Province of British Columbia.)

53

Figure 4.8: Splay Examples from the Saskatchewan River. Splays within the Saskatchewan system generally feature a predominate channel emerging from a crevasse with multiple bifurcations downstream leading to splay anastomosis. Downstream extensions frequently feature a merging of distributary channels from initially distinct splays (e.g. right-center of photo). Splay development in this system resembles what might be expected in the Sandover with more cohesive floodplain sediments. Note how large splays develop long and narrow planforms, which contributes to the aspect ratio difference of the Saskatchewan with respect to the other systems. (Image is from Google Earth and copyrights are held by Cnes/Spot Image.)

54

Figure 4.9: Splay Examples from the Sandover River. (A) Splays within the Sandover system result from numerous, closely-spaced channel bank breaches. There is a high degree of inter-connectedness between these channels. Splay development in this system resembles what might be expected in the Saskatchewan with less cohesive floodplain sediments. Note the presence of the abandoned channel (pre 1974 avulsion) in the lower right (southeast) portion of the photo. (B) Proximal floodouts are similar to splays in deposition and growth, and feature main channels with many bifurcations leading to small distributary channels. Note that the floodout defined here is the depositional (splay-like) portion and that small channel pathways are evident downstream of the defined boundary. These small channels continue for approximately 15 kilometers before coalescing into another primary channel. (Images are from Google Earth and copyrights are held by DigitalGlobe (A and B).) 55

Figure 4.10: Channel Development and Floodplain Incisions along the Ovens River. (A) The Ovens River does not feature splay development, although the channel pathway prior to a 1974 localized avulsion is apparent (yellow arrows). Additionally, the headward (upstream) channel development of the Deep Creek reach is evident, as the downstream portion is much wider (aqua arrow) than the upstream portion (blue arrow). (B) Thin floodplain incisions or gullies are evident on this portion of the Ovens River floodplain. These features may serve as the development locations for new anabranches or portions thereof. The red arrows mark the same location on A and B; thus, the continued upstream development of Deep Creek can be observed. (Images are from Google Earth and copyrights are held by GeoEye (A and B).)

56

Table 4.1: Measured Splay Quantities in Modern Systems. The following are the results of the Google Earth analysis. Splay frequency represents the number of splays per kilometer of reach length. Splay proportion is the amount of the basin filled by splays when excluding channel areas. System Studied Identified Splay Study Channel Area Splay Area Channel Splay Reach Splays Frequency Area Size (excludes Islands) (km2) Proportion Proportion Length (km-1) (km2) (km2) (excluding (km) channels) Columbia River* 20.4 43* 2.10 7.60 1.56 1.06 0.205 0.176 Sandover River 23.9 32** 1.34 105 5.08 35.1 0.049 0.352

Saskatchewan 44.4 39*** 0.878 73.2 7.85 33.4 0.107 0.512 River

Ovens River 26.1 0 0 22.2 0.859 0 0.039 0

* -Values are for Main Channel Reach only. 33 additional splays were measured near Castledale (Figure 4.2). **-Includes two floodouts. ***-Values are for described reach only. 4 additional splays were measured along the New Channel before the confluence with the Torch River.

57

Figure 4.11: Splay Areas within the Three Systems. Splays areas for identified splays in the upper Columbia (n = 76), Sandover (n = 32), and Saskatchewan (n = 43) River systems.

58

Figure 4.12: Channel Widths within the Three Systems. Parent channel widths for identified splays in the upper Columbia (n = 76), Sandover (n = 32), and Saskatchewan (n = 43) River systems. 59

Figure 4.13: Orthogonal Extent into Flood-basin within the Three Systems. The orthogonal (with respect to the parent channel) extent splays extend into the flood-basin for identified splays in the upper Columbia (n = 76), Sandover (n = 32), and Saskatchewan (n = 43) River systems. 60

Figure 4.14: Measured Splay Widths within the Three Systems. Maximum splay widths for identified splays in the upper Columbia (n = 76), Sandover (n = 32), and Saskatchewan (n = 43) River systems.

61

Figure 4.15: Splay Channel Path Lengths within the Three Systems. Lengths of the primary splay channel for identified splays in the upper Columbia (n = 76), Sandover (n = 32), and Saskatchewan (n = 43) River systems.

62

Figure 4.16: Normalized Orthogonal Extent (by Channel Width). Orthogonal extent of identified splays adjusted for channel scale (width) in the upper Columbia (n = 76), Sandover (n = 32), and Saskatchewan (n = 43) River systems.

63

Figure 4.17: Aspect Ratio (Maximum Splay Width/Splay Channel Path Length). Aspect ratio (length/width) of identified splays in the upper Columbia (n = 76), Sandover (n = 32), and Saskatchewan (n = 43) River systems.

64

Table 4.2: Summary Statistics and T-test Comparison Results.

Means, Medians, and Standard Deviations are presented for each physical parameter measured in identified and mapped splays. Also presented are the results of two-tailed, heteroscedastic (i.e. assumes unequal variances for the different systems) Student t-tests that compare each of the systems. For the t-test, the null hypothesis is that the means of the parameters are equivalent in the corresponding systems. Aspect Splay Normalized Ratio Orthog- Maximum Log of Splay Log of Channel Log of Channel Orthogonal (Splay Log of System or Summary onal Splay Normalized Area Splay Width Channel Path Extent (by Length Aspect Comparison Statistic Extent Width Orthogonal (m2) Area (meters) Width Length Channel divided by Ratio (meters) (meters) Extent (meters) Width) Maximum Width) Mean 19500 4.02 46.2 1.54 139 111 268 5.19 0.535 2.46 0.326 Columbia Median 10200 4.01 37.7 1.58 112 97.1 196 3.47 0.539 2.16 0.335 (n = 76) Standard 24700 0.503 35.1 0.348 82.3 65.7 247 5.61 0.391 1.43 0.239 Deviation Mean 1100000 5.49 190 2.23 932 701 1620 6.07 0.538 2.27 0.305 Sandover Median 206000 5.31 156 2.19 485 412 813 3.25 0.511 2.02 0.305 (n = 32) Standard 1830000 0.721 94.9 0.213 1020 601 2010 6.28 0.492 1.09 0.221 Deviation Mean 917000 5.58 181 2.20 1040 667 2040 5.84 0.652 3.03 0.418 Saskatchewan Median 389000 5.59 161 2.21 743 560 1360 3.83 0.584 2.47 0.392 (n = 43) Standard 1180000 0.639 91.1 0.233 1050 458 2050 4.55 0.323 1.87 0.226 Deviation Student t- Columbia vs. test 0.000 0.000 0.972 0.663 Sandover Probability Student t- Columbia vs. Non- Non- Non- Non- Non- Non- test 0.000 0.000 Non-Normal 0.080 0.039 Saskatchewan Normal Normal Normal Normal Normal Normal Probability Student t- Sandover vs. test 0.578 0.584 0.259 0.033 Saskatchewan Probability

65

Figure 4.18: Splay Area vs. Channel Width in the Upper Columbia System. Scatter plot of the raw splay area and channel width values in the upper Columbia.

66

Figure 4.19: Splay Area vs.Channel Width in the Sanodver and Saskatchewan Systems. Scatter plot of the raw splay area and channel width values in the Sandover and Saskatchewan. 67

Figure 4.20: Orthogonal Extent vs. Channel Width in the Three Systems. Scatter plot of the raw splay orthogonal extent and channel width values.

68

Figure 4.21: Splay Channel Path Length vs. Maximum Splay Width in the Three Systems. Scatter plot of the raw splay lengths and widths.

69

Table 4.3. Estimated Splay Volumes and Normalized Splay Discharge. Using average splay areas determined in this study (Section 4.3.1) and estimated splay thicknesses from literature (Section 4.3.3), the sediment volume deposited in the studied reaches and an average sized splay is calculated. Further, a normalized sediment discharge is estimated and presented to account for splay development time (~30 years), and estimated sediment discharges in these systems (0.041 m3/s and 0.181 m3/s, respectively). Since the Sandover is not gauged and discharge is unknown, this system is excluded from the normalized calculations. "Normalized Sediment Discharge" is an estimate of the fraction of total sediment discharged diverted into crevasse-splay deposits in the Columbia and Saskatchewan rivers.

Estimated Estimated Normalized Average Average Total Total Average Sediment Splay Splay System Splay Area Splay Splay Discharge Thickness Area (m2) Volume Volume (m) (m2) (m3) (m3) Columbia 1,060,000 0.8 85,000 19,500 16,000 4.1 x 10-4 Sandover 35,100,000 0.1 3,500,000 1,110,000 110,000 ------Saskatchewan 33,400,000 1.8 60,000,000 917,000 1,700,000 9.6 x 10-3

70

5.0 Observations in Ancient Systems

Ancient deposits provide an important opportunity to study the relationships between grain-size, floodplain drainage, and splay deposits. Thus, during the summer of 2011, five weeks of field study focused on determining how grain-size distributions and floodplain drainage is related to crevasse-splay size and abundance in several ancient fluvial systems. In particular, this study examined the Late Cretaceous-Paleogene Ferris Formation (Hanna Basin, south-central Wyoming), Paleocene Fort Union Formation, and Paleocene-Eocene Willwood Formation (Bighorn Basin, north-central Wyoming) (Figure 3.2). Jones and Hajek (2007) described the Ferris Formation as stratigraphically abrupt with limited splays and the Fort Union and Willwood formations as stratigraphically transitional with abundant splays.

5.1 Formations

5.1.1 Ferris Formation

The Late Cretaceous-Paleocene Ferris Formation was deposited during the Laramide orogeny within the intermontane Hanna Basin (e.g. Jones, 2007; Jones and Hajek, 2007; Hajek et al., 2010b). Fossil evidence suggests the climate was tropical-temperate and the setting was deltaic lacustrine to estuarine (Wroblewski, 2004). Eberle and Lillegraven (1998) examined approximately 660 meters of Ferris section fully within the Puercan North American land mammal “age” (circa 65 Ma - 63.3 Ma), suggesting an accumulation rate of ~0.4 mm/year or more (Hajek et al., 2012). Within the formation, channel-belt sand bodies are generally tens to hundreds of meters wide, 1.5-4 meters thick, and contain 1-2 stories (Jones, 2007). Sand body grain sizes are coarse to very coarse sand and the average interpreted paleoflow depth is 0.6 meters (Hajek et al., 2012). Crevasse-splays are rare throughout the Ferris Formation and are never observed in the interval below a channel. Instead, channel deposits cut into and directly overlie dark carbonaceous floodplain mudstones (Jones, 2007; Jones and Hajek, 2007).

71

5.1.2 Fort Union Formation

The Paleocene Fort Union Formation also dates to the Laramide orogeny and was deposited within the intermontane Bighorn Basin (e.g. Bown, 1980; Kraus, 1998; Kraus and Wells, 1999). The depositional climate was humid continental with a mean annual temperature of approximately 10°C and a temperature range of 25ºC (Hickey, 1980). The formation is predominately fluvial, but areas of paludal and lacustrine conditions are known (Hickey, 1980). Kraus and Wells (1999) estimated basin-averaged sedimentation rates to be ~0.1 mm/yr during early Fort Union deposition and increasing to ~0.3 mm/yr by later stages of deposition. Larger “trunk” channel deposits are normally around 10 meters thick with 2 to 3 stories that are each 3 to 4.5 meters thick, suggesting paleoflow depths of 3-5 meters (Kraus, 1998; Kraus and Wells, 1999). These trunk channels are typically around 1 km wide with predominate grain sizes from very fine to (rarely) coarse sand, and are interpreted as meandering channels (Bown, 1980; Kraus, 1998; Kraus and Wells, 1999). Smaller distributary and crevasse-channel deposits are relatively narrow, generally single or weakly multi-story “ribbon” sandstones on the scale of tens of meters wide, typically 1 to 3 (up to 9) meters thick with grain-sizes from very fine to medium sand (generally finer) (Kraus and Wells, 1999). Trunk channels are commonly underlain by “heterolithic avulsion deposits” (Kraus and Wells, 1999) comprising abundant thin very-fine grained sandstones less than 1 meter thick encased in weakly pedogenically modified silty deposits found both lateral to and beneath paleochannels are collectively interpreted to represent crevasse-splays and related overbank deposition (Kraus, 1998; Kraus and Wells, 1999). Clay and silt distal floodplain deposits not associated with avulsion sequences feature stronger pedogenic development with yellow, brown, and black soils, which suggests slower aggradation rates (Kraus, 1998).

5.1.3 Willwood Formation

The Paleocene-Eocene Willwood Formation conformably overlies the Fort Union in the Bighorn Basin (although some peripheral basin areas feature an angular unconformity; Bown, 1980). During deposition climate shifted towards warmer and dryer conditions with mean annual temperatures around 13.5°C to 17°C and winter temperatures above freezing during the Eocene (Hickey, 1980; Greenwood and Wing, 1995; Kraus and Wells, 1999). Basin-averaged 72

sedimentation rates within the Willwood Formation are estimated at ~0.5 mm/yr (Kraus and Wells, 1999). Sedimentology within the Willwood Formation is similar to that of the Fort Union Formation (Kraus and Wells, 1999), with similar facies characteristics for channel bodies, avulsion-related deposits, splays, and other proximal overbank deposits (e.g. Kraus and Aslan, 1993, Kraus, 1996, Kraus, 1998; Kraus and Wells, 1999). The primary sedimentological difference between the two Bighorn Basin formations is the degree of paleosol development in the Willwood Formation (Kraus, 1998; Kraus and Wells, 1999). The Willwood Formation is characterized by its distinctive, well-developed paleosol horizons and successions (e.g., Kraus, 1996). These several meter-scale successions feature a basal light grey or greenish-gray siltstone overlain by a finer-grained (i.e. higher proportion of clay) yellow-brown paleosol horizon. The succession is capped by a clay-rich heavily mottled bright red and grey clayey paleosol horizon (Kraus, 1996). This succession reflects hiatal intervals during which thick, well-developed soils formed on the floodplain and likely captures an overall decline in sedimentation rates throughout the sequence (e.g. Bown and Kraus, 1987).

5.2 Field Observations

In order to test my hypotheses, field observations focused on obtaining information about paleo- sediment loads and paleo-floodplain drainage conditions, as well as crevasse-splay deposit characteristics in each formation. Visual grain-size estimates and hand samples were collected from channels, “proximal” overbank deposits (including splays and avulsion deposits), and “distal” overbank (i.e. floodplain) deposits. The thickness and lateral extent of these deposits was also measured and/or estimated, and paleosol development was described within overbank deposits. Collectively, these measurements test the stated hypotheses in three primary ways: (1) determination of the range of grain-sizes present, (2) the relative proportions of specific grain- sizes in the deposits, and (3) relative floodplain sedimentation rates and saturation conditions. Sampling at each outcrop included collection of three or more hand samples from paleo- channels to document grain sizes at the base, middle, and top of the deposit and any notable structures within the deposit (e.g. mud-plugs, sand bars, etc.). Multiple samples collected from individual interbeds within avulsion deposits preceding channels. Sampling of individual horizons—including splay sand-sheets and paleosol successions—is in the form of vertical

73

sections with increasing distance from the channel. A representation of this sampling pattern in presented as Figure 5.1. In the late summer and autumn of 2011, analytical grain-size distributions were obtained for characteristic channel, splay, and floodplain samples. Sample preparation consisted of weighing, disaggregation using a mortar and pestle, and dry sieving. Care was taken to ensure that as few grains as possible would be crushed or cracked during disaggregation and sieving, although this is a potential source of error. Dry sieving separated the disaggregated samples into fractions finer and coarser than 850 microns (0.85 mm), equivalent to upper coarse sand. Particle-size measurements of fine disaggregated samples were conducted at Penn State's Particle Characterization Laboratory (PCL) using the Malvern Mastersizer S. The Mastersizer S uses dynamic light scattering with a Helium-Neon laser (633 nm wavelength) to determine particle sizes ranging from 0.05 to 900 microns (0.9 mm). Prior to testing, each sample vile was filled with approximately 50 mL of water, placed in an ultrasonic bath for approximately 2 minutes to further disaggregate fine sediment, and hand shaken to suspend particles. With the finer sediment suspended, a portion of the sample was introduced to the Mastersizer via the wet sampler. A Malvern program then calculates and reports the percent volume of particles in each of 64 grain size intervals between 0.05 to 900 microns. These values were saved in a spreadsheet for further analysis.

5.3 Results

5.3.1 Ferris Formation

Channel Facies: Two sampled single story channel deposits were ~ 0.6 and 1.8 meters thick vertically and extend on the order of several tens of meters laterally. Visually estimated Grain sizes within channel bodies range from upper medium sand to upper very coarse sand (~354 – 2000 microns) with most in the upper coarse to lower very coarse range (~707 – 1410 microns) (Figures 5.2 and 5.3). Proximal Overbank Facies: Channel bases either directly overlie dark carbonaceous floodplain shales or thin (10 – 20 centimeters thick) interbedded clay-silt-sand horizons with most sediment in the silt to lower very fine range (~4 – 125 microns) (Figures 5.2 and 5.3). Notably, when channels overlie these thin vertical successions the contact is mostly non-erosive and 74

conformable, and the underlying vertical succession takes the same shape as that of the channel base, suggesting these deposits may represent the nascent development of the channel. At the lateral margins of channels, interbedded sand and clay – silt intervals (~1 – 30 centimeters thick) rapidly thin and fine away from the channel, pinching out within 10 - 20 meters of the channel margin. A lower fine to upper coarse (~125 – 1000 microns) splay sand up to 30 cm thick pinched out within only 7 meters. These deposits may reflect restricted channel- margin and crevasse-splay deposition. Distal Overbank (Floodplain) Facies: The formation’s most common facies is dark carbonaceous, peaty and coaly shale with little paleosol development. This is characteristic of swampy and waterlogged conditions (e.g. Mack et al., 1993). Distal overbank grain sizes are mostly in the clay to fine silt range (~1 – 16 microns), with small amounts (<25%) of medium silt to very fine sand (16 – 125 microns) (Figure 5.3).

5.3.2 Fort Union Formation

Channel Facies: Observations included two trunk channels with grain-sizes ranging from very fine sand to pebbles (Figure 5.2). Pebbles (~2 – 64 mm) were only found in numerous sand bars (roughly meter thick) in a single trunk channel with other non-bar portions in the fine to medium sand (~125 – 500 microns) range. The other observed trunk channel contained very fine to medium sand (~63 – 354 microns) (Figure 5.3). The trunk channels were several hundred meters to a kilometer or more in width and 6 to at least 10 meters thick. Trough cross-bed sets within the channel bodies were 10 - 40 centimeters thick (truncated). Proximal Overbank Facies: Ubiquitous coarsening-upward successions of interbedded silt and sand layers are found beneath paleochannels, which are the “heterolithic avulsion deposits” of Kraus and Wells (1999). Grain sizes range from clay to fine sand (~1 – 250 microns) in these intervals (Figure 5.4), and they feature weak paleosol development (rare orange mottling), which suggests rapid deposition (e.g. Kraus, 1998). Widths and thicknesses of proximal overbank facies sequences are on scale with the overlying channel deposits. A splay channel within a heterolithic complex was also measured and characterized. It was between 2 and 4 meters thick and around 30 meters wide. Grain-sizes within the channel

75

ranged from fine to coarse sand (~125 – 707 microns). Notably, this ribbon channel featured a full succession of lateral paleosols from weakly to heavy modified. Lateral to channel facies, proximal overbank deposits include splay sandstone sheets and fine-grained intervals. Sandstone sheets are very fine to medium sand (~63 – 500 microns) in decimeter scale beds (up to near a meter) with ripple laminations. These sheets extend for tens to hundreds of meters, although the full extent cannot be measured due to exposure. Finer-grained intervals generally consist of clay and silt and feature weak or moderate paleosol development (greenish grays to orange), suggesting fairly rapid deposition, although slower than for heterolithic deposits. Lateral extent is on the order of hundreds to thousands of meters. Distal Overbank (Floodplain) Facies: Distal floodplain sediments had varying paleosol development with one outcrop consisting of muted red, mottled paleosols and another outcrop consisting of blackish, brownish, and yellowish paleosols. These collectively suggest relatively low deposition rates (when compared with proximal overbank horizons) and some seasonality in exposure, but the latter type is indicative of more waterlogged conditions and is the more common type observed in the formation by others (e.g. Bown, 1980; Kraus, 1998). Grain sizes within distal floodplain deposits are clay to fine silt. Individual horizons are meter-scale in thickness and extend laterally for hundreds to thousands of meters.

5.3.3 Willwood Formation

Channel Facies: The five Willwood Formation channel deposits feature grain sizes from very fine to very coarse sand (~63 – 1410 microns) (Figure 5.5), although most samples were in the fine to medium sand range (~125 – 500 microns). A 0.9 meter sand bar was observed, and contained the coarsest grains observed in the Willwood Formation. Trough cross-beds up to approximately 50 centimeters in thickness were measured. Overall, channels were between 2.8 and 6 meters thick, and lateral extents were up to a kilometer and possibly more, although exposures were not complete. All identified channels were trunk channels (e.g. Kraus and Wells, 1999). Proximal Overbank Facies: The Willwood Formation features at least 2.5 meter thick sequences of coarsening-upward “heterolithic avulsion deposits” consisting of interbedded silt and sand (Kraus and Wells, 1999). Grain sizes range from clay to fine sand (Figure 5.5), and very

76

weak paleosol development (rare orange mottling), suggesting rapid deposition (e.g. Kraus and Aslan, 1993; Kraus, 1996). These features frequently extended below the ground-surface. Proximal overbank deposits lateral to channel deposits include splay sandstone sheets and fine-grained intervals. Observed sand sheets are slightly finer in the Willwood Formation compared to the Fort Union Formation, and feature very fine to fine sand in ripple-laminated, 10 – 50 centimeter thick beds. These sandstone sheets extend tens to hundreds of meters away from channels. Finer-grained intervals had clay and silt grain sizes, weak to moderate paleosol development (greenish grays to orange), and lateral extent on the order of hundreds to thousands of meters. Distal Overbank (Floodplain) Facies: These facies feature the most significant differences between the Fort Union and Willwood formations. Grain sizes in Willwood floodplain deposits are entirely clay to fine silt (~1 – 16 microns) (Figure 5.5), and distal overbank paleosols are always reddish and purplish with extensive mottling featuring various colors (greens, oranges, purples, and reds) and slickenslides. This development suggests extensive seasonal wetting and drying (e.g. Mack et al., 1993). Lateral extent of these deposits is from hundreds to thousands of meters, and individual horizons are between approximately 0.5 and 2 meters thick.

5.4 Discussion

Crevasse-splay development differs between the Ferris Formation and the Fort Union and Willwood formations as evidenced by the frequency of observed splays and their lateral extent. The Ferris Formation features few and laterally limited splays with none observed beneath channels, while splays are common, extensive laterally, and ubiquitous beneath channels in the Fort Union and Willwood Formations. These differences are potentially due to discharge, available grain sizes, and floodplain drainage characteristics of the formations. As presented in the preceding section, the channel scale within the Ferris Formation is smaller than that of the Fort Union and Willwood Formations, which may contribute to differences in the scale of splays. Individual Fort Union and Willwood channels are approximately 2 – 5 times deeper than Ferris channels and several times wider, which may account for some of the discrepancy in splay sizes. However, deposits lateral to Ferris channels pinch out in a distance equivalent to or less than the system’s channel width, while horizons

77

within the Fort Union and Willwood Formations are traceable for distances well in excess of the channel width (e.g. Kraus and Wells, 1999). This suggests that other factors play a role in the lateral extent of deposits. Overall, the Ferris Formation comprises an approximately bimodal grain-size distribution with peaks in coarse sand and clay – finer silt, and very little intermediate sized sediments. Laterally narrow (several tens of meters) and thin paleochannel bodies consist predominately of coarse and very coarse sand, while most overbank deposits clay and fine silt floodplain intervals. Presumably, coarse in-channel deposits reflect characteristic bedload in the Ferris Formation, while the clays and fine silts on the floodplain reflect the paleo-washload. Rare crevasse-splay or channel-margin deposits range from coarse silt to fine sand, and may reflect a relatively small fraction of coarse suspended load in Ferris rivers. In contrast, the Fort Union and Willwood Formations generally consist of a narrower, more unimodal distribution with a range of grain sizes predominately from clay to medium sand (with minor coarse sand to pebbles). Approximately kilometer-wide channels consist primarily of fine and medium sand, which, for rivers with few meter scale flow depths, would likely have been easily suspended (e.g. Shah-Fairbank et al., 2011). Crevasse-splay deposition in these formations is abundant, both beneath channels in the form of the ubiquitous "heterolithic avulsion deposits" of Kraus and Wells (1999) and lateral to channel deposits as overbank deposits. Floodplain-drainage conditions in the Ferris Formation also differ markedly from those in the Fort Union and Willwood Formations. Floodplains in the Ferris Formation show little paleosol development and feature highly carbonaceous and peaty, fissile shale, which suggests waterlogged conditions and rapid flood-basin sedimentation (Mack et al., 1993). On the other hand, distinct paleosol development in both the Fort Union and Willwood Formations consists of deep red paleosols with heavily mottled gray, green, and orange coloration, suggesting seasonal wetting and drying, and periods of low sedimentation (e.g. Simonson and Boersma, 1972; Mack et al., 1993). The Fort Union Formation also features some outcrops displaying black and yellowish paleosols, indicating wetter and more acidic conditions (Kraus, 1998). Differing floodplain inundation, sedimentation rates, and climate may account for these paleosol differences.

78

In general, observations in these formations validate the hypotheses that large and abundant splays will form most readily in systems that (1) have dry (at least seasonally) floodplains (facilitating a lateral water-surface slope across the floodplain) and (2) abundant coarse “suspended” sediment (e.g. coarse silt – fine sand) that is delivered to the floodplain via levee breaches. These conditions prevail in the Fort Union and Willwood Formations, which feature large splays, periodically dry floodplains, and an abundance coarse silt to fine sand, while the Ferris Formation, with small, rare splays, had persistently wet floodplains and very little coarse silt – fine sand. These results are consistent with observations from modern systems (Section 4.0).

79

Figure 5.1: Representation of Field Sampling. In the field, samples were collected from paleo-channel bodies, avulsion deposits beneath channels, and in multiple vertical sections with distance away from the channel. Within these sections, samples were collected from the various paleosol horizons and lateral avulsion deposits.

80

Figure 5.2: Range of Visually Estimated Grain-Sizes for Facies Types. The marker point represents the approximate visual mean grain-size estimate for an outcrop and the vertical bars represent the visual grain-size range for a specific outcrop. 81

Figure 5.3: Example Ferris Grain-Size Distribution. Due to limitations in the Mastersizer, the channel sample was tested using the CAMSIZER at Tulane University. 82

Figure 5.4: Example Fort Union Grain-Size Distribution.

83

Figure 5.5: Example Willwood Grain-Size Distribution.

84

6.0 Modeling Splay Development in Delft3D

Observations in modern and ancient systems presented in the preceding sections are consistent with the hypotheses that greater splay development occurs in systems featuring intermediate grain-size distributions and “dry”, well-drained (i.e. steep water-surface gradient) floodplains. However, these cannot help determine the importance of either grain-size or floodplain drainage on splay production in isolation, or alternatively, what specific combinations of grain-size and floodplain drainage will produce the most abundant and largest splays. In order to further evaluate the importance of these variables, I used Delft3D to: (1) document differences in splay morphology on “dry” versus “wet” floodplains (i.e. those dominated by advective or diffusive sediment transport) and (2) document differences resulting from three different grain-size distributions including a fine distribution (primarily clay and very fine to fine silt), an intermediate distribution (primarily coarse silt and very fine to fine sand), and a coarse distribution (primarily medium and coarse sand). Modeling follows the abbreviated workflow provided in Figure 6.1. There are two modeling domains to account for advective and diffusive flow routing from which each grain size distribution is modeled for a total of six final models.

6.1 Methods

This modeling was conducted used the software package Delft3D-FLOW from Deltares Systems. Delft3D is fully validated for modeling the 3-dimensional depth-averaged, non-linear, shallow water Navier-Stokes equations in accordance with the computational standards of the International Association of Hydraulic Research (IAHR, 1994). Six model runs with varying sediment distributions and floodplain drainage (Figure 6.1) were conducted on the same basic grid (Figure 6.2). The model setup assumes the channel is in near-equilibrium and graded conditions in the absence of the crevasse, as shown by model results discussed below in Section 6.1.3. Equations 4 – 9 below are solely utilized in determining parameter inputs in the Delft3D- FLOW interface, as fluid and sediment transport conditions are calculated within the Delft3D- FLOW program itself.

85

6.1.1 Model Grid and Initial Topography

The grid and bathymetry were constructed using Delft3D-RFGRID and Delft3D-QUICKIN, respectively. Important parameters for the grid and bathymetry are provided in Table 6.1 and the modeled grid and bathymetry are provided in Figure 6.2. To reduce the model computational time—but still allow for accurate morphodynamic and hydrodynamic estimates—the model incorporates a variable grid size and features a wider grid in the center of the domain to simulate the floodplain. Upstream and downstream of this wider grid, the grid is only 75 meters wide (15 grid cells in y-direction) to simulate the channel and levees. Within the simulated floodplain area, cell dimensions are 5 x 5 meters in a 1,000 x 1,000 meters area in vicinity of the crevasse and increase outward away from this area. In total, the wide central grid measures 3,067 (downstream) x 2,634 meters (orthogonal to downstream direction) and is 267 x 251 grid cells. Within the channel, the grid cells are always 5 meters wide orthogonal to the flow direction, but are variable longitudinally with a total channel length of 23,550 meters (321 grid cells, including the 267 cells discussed above in the middle reach). The modeled bankfull channel is 60 meters wide, has a bankfull depth of 3.0 meters, a floodplain elevation that is halfway between the channel bottom and levee height, side slopes in a 1:1 ratio, and transports a flow in the positive x-direction. The downstream channel bedslope is initially 0.0001 m/m. The left levee is two grid cells wide (10 meters) and the right levee is one grid cell wide (5 meters) and lies on the edge of the grid domain. Each levee cell is assigned a height 3.0 meters above the base of the channel. The crevasse is cut through the left levee and is located approximately 11,270 meters downstream of the upstream boundary, is ~10 meters wide (1 full/2 partial grid cells in the downstream x-direction), ~15 meters long (2 full/2 partial grid cells in the cross-channel y- direction), and the base is at floodplain height (i.e. ~1.5 meters above the channel bed). For ease of calculations, this floodplain does not feature a bedslope. Additionally, a second bathymetry featuring no crevasse was prepared for use in preliminary hydrodynamic and morphological testing.

6.1.2 Initial Bed Sediments

In Delft3D, sediment classes are user-defined and are either cohesive or non-cohesive, with cohesive sediments being silt-sized and finer (< 64 microns). Cohesive sediment erosion and 86

deposition is calculated using the Partheniades-Krone formulations based on user-defined critical shear stress thresholds, while cohesive sediment transport is determined via the advection- diffusion equation. Cohesive sediment classes are determined based on settling velocity, which for this study used Stokes’ Law

2 (휌 − 휌 ) 휔 = 푝 푓 푔푟2 [4] 9 휇 where ω is particle settling velocity, ρp is the particle density, ρf is the fluid density, μ is the fluid 3 dynamic viscosity, g is gravity, and r is the particle radius. For quartz particles (ρp = 2650 kg/m ) in 20°C water (μ = 10-3 N s/m2), the settling velocity is approximately 휔 = 900000퐷2 [m/s] [5] or 휔 = 900퐷2 [mm/s] [6] where D is the particle diameter. Within this model, there are three classifications for cohesive sediments, including categories for clay, “finer” silt (very fine and fine silt), and “coarser” silt (medium and coarse silt) (Table 6.2), for which the critical shear stress for entrainment is set. The listed settling velocities correspond to grain diameters of 1.95, 7.81, and 31.25 microns for these classifications, respectively. Hydrodynamic testing in a solid-walled channel (i.e. no crevasse) revealed that thalweg bed-shear stress was approximately 2.86 N/m2 for the middle reach. Consequently, given the assumption of a graded, near-equilibrium channel, I set the critical shear stress for erosion to 2.8 N/m2 for these three classes. This value is similar to critical shear stress values for other cohesive bed rivers (Mier and Garcia, 2011). For non-cohesive sediments, Delft3D calculates transport automatically using the formulations of Van Rijn (1993) and grain diameter explicitly defines sediment classifications. Within this study, there is one classification for each sand size from very fine sand to coarse sand, and the modeled grain sizes for all classifications are shown in Table 6.2. In Delft3D, sediment-transport conditions are determined based on the proportion of sediments available in the channel bed; thus, designation of appropriate relative sediment thicknesses is necessary. Delft3D-QUICKEN was used to develop the three sets of depth files utilizing these 7 grain-size classifications for the fine, intermediate, and coarse channel distributions (Table 6.3). Model runs with fine grain-size distributions are 84.1% clay and finer

87

silt, middle grain-size distribution runs contain 80.1% coarse silt, very fine sand, and fine sand, and coarse grain-size distribution runs contain 84.1% medium and coarse sand (Figure 6.3). Within each distribution, the relationship between adjoining grain types is 2.5:1, such that there is 2.5 times as much clay as finer silt in the fine-grained channel distribution or vice versa for the intermediate-grained and coarse-grained channel distributions. All six runs contained the same floodplain bed-sediment, which was set to the fine channel grain-size distribution.

6.1.3 Initial and Boundary Conditions, Including Water and Sediment Discharge

A series of set-up models helped determine parameters, initial conditions, and boundary conditions utilized in the final models. Three models—one for each channel sediment distribution profile—featured no crevasse or morphological changes for the duration of the simulation period, but included derived sediment thickness profiles from Table 6.3, which allowed equilibrium sediment concentrations to be characterized. This model was necessary because Delft3D-FLOW does not determine the equilibrium concentration profiles for cohesive sediments at model boundaries, so these boundary conditions must be user-defined. As Delft3D determines the equilibrium concentrations for non-cohesive sediments at model boundaries, values for the four sand classifications are not listed in Table 6.4 and were not user-defined in the final model runs. The set-up models were also used to determine the appropriate near-bankfull upstream water discharge (130 m3/s) and floodplain water-surface levels for the “dry” and “wet” model runs, 2.78 meters and 3.98 meters, respectively (Table 6.4). The normal depth (2.939 meters) corresponding to this discharge was determined via solving the following equation

3 퐶 × 푄3 퐻 = √ 푓 [7] 푔 × 푆 × 푤2 with 푓 퐶 = [8] 푓 8 and

1/6 8푔 푅 퐶 ≡ √ ≈ ℎ [9] 푓 푛 where H is normal water depth, Cf is a friction factor equal to the Darcy-Weisbach friction factor (f) divided by 8, Q is discharge, g is gravity, S is the channel friction slope, which is equal to 88

bedslope under uniform flow conditions (flow does not change from point to point at a given time interval), w is channel bankfull width, C is the Chezy friction factor, Rh is the hydraulic radius, and n is the Manning friction factor. A typical Manning’s “n” for straight sandy-bed rivers is 0.03 and this roughly equates to a Chezy friction factor of 43 m1/2/s (e.g. Julien, 2002; Chaundry, 2007). This Chezy value was used in all model simulations. To determine whether the designed channel was near equilibrium and graded conditions, three preliminary models utilized the sediment thickness profiles of Table 6.3, the derived sediment concentrations from Table 6.4, featured a solid-walled channel (i.e. no crevasse), 80x scaling factor, and morphological changes. This model was simulated for 45 days and the cumulative erosion and sedimentation profiles for the thalweg of the middle reach are presented in Figure 6.4. Over the 45 days, none of the three channels featured erosion or sedimentation in excess of 4 centimeters and the maximum difference is approximately 1.2% of the channel depth. Given the small magnitude of this variance, the model design is considered to be near equilibrium and graded conditions. Six additional set-up models—one for each of the final models— included the crevasse, sediment thickness files, and floodplain boundary conditions, but excluded morphological changes for the duration of the model. These models were used to determine the start-up conditions for each of the final run models, as these final runs featured a “hot start.” Additional user-defined input parameters are presented in Table 6.4. Of importance, the downstream channel boundary is a QH-relation to account for varying discharge (maximum value corresponds full discharge normal depth), which accounts for discharge loss through the levee crevasse. Also, each of the floodplain boundaries is either assigned a water-surface level (WSL) approximately 5 cm above the floodplain bed surface (2.78 meters for dry “advective” floodplain models) or ~20 centimeters below the water-surface of the solid-walled channel (3.98 meters for wet “diffusive” floodplain models), such that water entering the floodplain is effectively flowing into a lake with set water-levels. These conditions simulate high (dry) and low (wet) cross-floodplain water-surface gradients (e.g. Adams et al., 2004). Models were run at an 80x morphological scaling factor, which means each simulated days is equivalent to 80 days of morphological change.

89

6.2 Results

Initial flow magnitudes (i.e. Day 0) in proximity to the crevasse are presented for the dry and wet model set-ups in Figures 6.5 and 6.6, respectively. For these two set-up conditions, the initial crevasse discharges are 24.4 m3/s and 10.1 m3/s, respectively, and the associated downstream discharges are 105.6 m3/s and 119.9 m3/s, respectively. The reduction in crevasse discharge exemplifies the importance of water-surface gradient in the delivery of water and sediment to the floodplain, as the bathymetry for these two model domains is otherwise identical. Through- crevasse flow velocities are approximately 3.3 m/s and 0.7 m/s for the dry and wet models, respectively. Additionally, the turbulent jet propagating from the crevasse is directed in a more downstream direction for the diffusive transport (wet floodplain) run, and the expansion angle of this turbulent jet is much narrower, while the expansion angle of the advective transport (dry floodplain) run is nearly 45°, which is the theoretical maximum (Machusick, 2000).

6.2.1 Run 1: Fine Channel Sediment with Dry Floodplain

Crevasse and downstream discharges (Figure 6.7) show a highly erosive initial phase as crevasse discharge increases to over 60 m3/s within a few simulation days and stays around this value for the duration of the model. Floodplain water flow is represented as a small lobate flow expansion (Figure 6.8) emerging from the crevasse mouth and features through crevasse velocities of ~ 0.7– 0.9 m/s. Following the 162 simulation days, significant erosion (~ 6–8 meters) occurred within the crevasse (Figure 6.9), but only limited deposition occurred elsewhere. Maximum deposition of 0.5–0.6 meters occurred within two “wings” on either side of the crevasse mouth, and these wings may be attributable to deposition from diffusive eddies. Elsewhere, deposition greater than 0.4 meters presents as a bird’s foot shaped pattern, and cross-sections (Figures 6.10 – 6.11) reveal a deeply incised channel near the crevasse and generally gentle slopes towards splay margins. Deposition greater than 0.3 meters covers an area approximately 190,000 m2, and deposition greater than 0.1 meters covers approximately 1,290,000 m2 (Table 6.5). Sediment transport through the crevasse (Figures 6.12 – 6.14) is extensive in the first 5–10 simulation days and eventually levels off to a rate of approximately 0.0002 m3/s, which is much less than the equilibrium rate observed in Run 1’s set-up model (Figure 6.14). 90

6.2.2 Run 2: Fine Channel Sediment with Wet Floodplain

Over the 153.5 day simulation, relatively little flow is diverted from the main channel (Figure 6.15), as there is a minor increase in through crevasse discharge to 10–14 m3/s, which is roughly 10% of the upstream discharge. Final flow magnitudes show a slight downstream oriented zone of flow expansion with a narrow zone of higher velocities extending away from the crevasse opening (Figure 6.16). Through crevasse flow magnitudes are mostly 0.5–0.7 m/s. Maximum erosion (0.2–0.3 meters) occurs within and adjacent to the crevasse (Figure 6.17) and sediment deposition is limited (Figures 6.17 – 6.19), with only 1 (non-channel) grid cell having deposition in excess of 0.3 meters and deposition greater than 0.1 meters limited to 24,250 m2 (Table 6.5). Of the six model runs, sediment transport through the crevasse (Figures 6.12) is lowest for this scenario and the instantaneous sediment discharge (Figure 6.13) is less than 0.00005 m3/s. Normalized sediment discharge (Figure 6.14) is also lowest in this model run and is around 20% of the equivalent set-up model.

6.2.3 Run 3: Intermediate Channel Sediment with Dry Floodplain

Run 3 crevasse discharges show an erosive initial phase, a brief quasi-equilibrium around 50 m3/s lasting until approximately until day 25, and finally an overall increase that persists until the end of the 198 day model run (Figure 6.20). At the end of the simulated period, the flow magnitudes (Figure 6.21) show the development of a distinct splay-channel bifurcation approximately 50 meters from the crevasse and two channels extending around 350 meters into the flood-basin. Flow velocities in these channels range from 0.7 m/s to 1.4 m/s. The splay channels eroded approximately 2–5 meters into the floodplain before the bifurcation and generally between 0.5 and 2 meters in the branches (Figure 6.22). Maximum deposition was between 0.8 and 1.0 meters in narrow bands surrounding the bifurcation, with a broad 300 x 250 meters area consisting of deposition between 0.6 and 0.8 meters. This broad area presents as a plateau on cross-sections (Figures 6.23 – 6.24) and is surrounded by gentle slopes. The 0.3 meter contour encompasses an area of ~321,000 m2 and the 0.1 meter contour an area of ~1,200,000 m2 (Table 6.5). This model run features the highest cumulative sediment transport (Figure 6.12) and is the only run with a distinctly increasing instantaneous sediment

91

discharge (Figure 6.13). Normalized sediment discharge indicates that sediment discharge is greater in this simulation than in the hydrodynamic set-up model (Figure 6.14).

6.2.4 Run 4: Intermediate Channel Sediment with Wet Floodplain

Crevasse and downstream discharges show a sudden drop in discharge through the crevasse at the onset of modeling and then crevasse discharge around 6 m3/s for the duration of the model (162 days) (Figure 6.25). Final flow magnitudes in the crevasse’s proximity are around 0.7–0.8 m/s, and there is a rapid decline away from this plume with magnitudes dropping to less than 0.3 m/s within 50 meters (Figure 6.26). Net deposition of 1.0-1.2 meters is found in a band extending from near the crevasse to approximately 290–300 meters outward, which is further encircled by a thin 10–15 meter band of sedimentation between 0.1 and 1.0 meters (Figure 6.27). Cross-sections (Figures 6.28 – 6.29) show a plateau with steep, peripheral slip-faces. Thus, the transition from splay-associated deposits to flood-basin associated deposits is rapid, and the area encompassed by the 0.3 meter contour (~142,000 m2) is only slightly smaller than the area encompassed by the 0.1 meter contour (~151,000 m2) (Table 6.5). Cumulative total transport (Figure 6.12) is an order of magnitude less for Run 4 when compared to Run 3, and instantaneous sediment discharge is approximately 0.0001 m3/s, which is the second smallest value of any model run (Figure 6.13). Normalized sediment discharge is approximately 0.4, which indicates that sediment discharge is only around 40% of the value observed in the hydrodynamic set-up model (Figure 6.14).

6.2.5 Run 5: Coarse Channel Sediment with Dry Floodplain

Crevasse and downstream discharges (Figure 6.30) show the highly erosive initial phase with crevasse discharge increasing to greater than 50 m3/s, a mostly chaotic phase from Day 10 to Day 80, and a slow decline phase from Day 80 until the end of the simulation (Day 196) with final through crevasse discharge of approximately 41 m3/s. Flow magnitudes (Figure 6.31) show a similar pattern to Run 3, with a single channel extending approximately 80 meters out from the crevasse before a bifurcation, and these two secondary channels extending over 300 meters farther into the basin. Crevasse-channel flow velocities are approximately 0.7–1.0 m/s in all of these channels. 92

Splay channel erosion is approximately 1–3 meters extending from the crevasse to the bifurcation and generally between 0 and 1 meter in the secondary channels (Figure 6.32). Deposition greater than 0.8 meters extends approximately 40–60 meters out from the primary splay channel, and a broad (440 x 340 meters) area features deposition greater than 0.5 meters. Cross-sections C, D, and F (Figures 6.33 – 6.34) capture this broad area of deposition and all cross-sections collectively show the gently sloping margins of the crevasse-splay. The 0.3 meter contour surrounds an area of ~315,000 m2 and the 0.1 meter contour surrounds an area of ~1,160,000 m2 (Table 6.5). Both areas are slightly smaller than the equivalent measurements for Run 3. The second highest values for cumulative sediment transport (Figure 6.12) and sediment discharge (Figure 6.13) are observed in Run 5. However, this run features the highest normalized sediment discharge indicating the greatest sediment discharge in excess of the baseline hydrodynamic set-up model.

6.2.6 Run 6: Coarse Channel Sediment with Wet Floodplain

Crevasse and downstream discharges (Figure 6.35) indicate a brief depositional phase at the onset of the model followed by quasi-equilibrium with crevasse discharges of 4–7 m3/s for the duration of the model (141.5 Days). Flow magnitudes (Figure 6.36) are approximately 0.7–0.9 m/s in a plume extending 40 meters outward from the crevasse and flow magnitudes greater than 0.3 meters are confined to an area extending 65–80 meters outward from the crevasse. Similar patterns were also seen in Run 4 (Figure 6.26). Deposition patterns are also similar to Run 4 (Figure 6.27), with deposition between 1.0 and 1.2 meters extending approximately 290–310 meters from the crevasse and this area surrounded by a thin 5–10 meters band with sediment thicknesses between 0.1 and 1.0 meters (Figure 6.37). Cross-sections (Figures 6.38 and 6.39) show this crevasse-splay as a plateau with steep slip-faces. Deposition greater than 0.3 meters covers an area of ~149,000 m2, and deposition greater than 0.1 meters covers ~153,000 m2 (Table 6.5), and both areas are slightly greater than the equivalent measurements in Run 4. Cumulative sediment transport (Figure 6.12) and sediment discharge (Figure 6.13) are also slightly greater than those of Run 4, with the latter value ~0.0001 m3/s. Interestingly, the normalized sediment discharge differs from Run 4 and is actually above 1, which indicates that more sediment is passing through the crevasse in this

93

model run than in its hydrodynamic set-up model (Figure 6.14). Along with Runs 3 and 5, these are the only models to feature a normalized value greater than 1.

6.3 Discussion

These models give many insights into the influence of grain size and floodplain drainage on crevasse-splay development, and the parameter space in which small and large scale splay growth occurs. Systems dominated by clay and finer silt (i.e. Runs 1 and 2) show very little splay deposition overall both in surface area and sediment volume relative to equivalent intermediate and coarse-grained model runs (Table 6.5), and, consequently, also feature a lower through- crevasse sediment discharge than runs with coarser grain sizes (Figures 6.12 and 6.13). There are differences between Runs 1 and 2, however. In Run 1, erosion dominates proximal crevasse settings, which is due to rapid crevasse flows and associated elevated shear stresses preventing fine sediments from falling out of suspension. The most extensive deposition presents as a bird’s foot shaped wedge a few hundred meters from the crevasse. While crevasse flow velocity is lower in Run 2, the corresponding decrease in crevasse discharge limits sediment supply—both from the main channel and eroded floodplain sediments—leading to very limited deposition right at the crevasse (~0.2–0.3 meters). All other model runs feature an equivalent amount of deposition hundreds of meters away from the crevasse throat. A modern system featuring fine- grained small “splay-like” features is Red Creek in Wyoming (Schumann, 1989). Like model runs 1 and 2, splay deposition on Red Creek floodplains may be limited because the system lacks a sufficient supply of sediment that will settle from suspension on the proximal floodplain near levee crevasses. Outside of the fine-grained models, the primary control on crevasse-splay development appears to be floodplain inundation style. The “wet” intermediate and coarse-grained models (i.e. Runs 4 and 6) produce small splays extending approximately 300 meters outward from the crevasse with steep slip-faces on their splay margins. Splay area and volume calculations (Table 6.5) indicate these two models produce crevasse-splays that are similar in extent and basin-filling characteristics. These models also feature low crevasse discharges and immediate deposition within the crevasse at the start of the models. This suggests that highly inundated floodplains

94

suppress not only the basin-ward sediment advection, but through crevasse water transport as well. In contrast, splays produced in Runs 3 and 5, the “dry” floodplain runs, featured more expansive splay deposits and gently sloping profiles. Deposition greater than 0.3 meters and 0.1 meters extends approximately 400 meters and 750 meters, respectively, from the crevasse. In proximal settings, these splays are generally 0.7–0.9 meters thick, which is thinner than in equivalent portions of the “wet” model runs. Unlike these wet model runs, there is distinct development of erosive splay channels in the crevasse and extending ~ 350 meters into the flood- basin with a bifurcation approximately 50 (Run 3) – 80 (Run 5) meters from the crevasse. Maximum erosion extends well below the base of the primary channel and erosion declines along the length of the ~15–25 meter-wide splay channels. This erosive development phase is also captured in the sudden increase of through-crevasse discharge at the onset of modeling. Splay area and volume calculations (Table 6.5) show that these splays are much larger in area, and a greater volume of sediment is deposited in the region contained by the 0.1 meter contour. Within the region contained by the 0.3 meter contour, the sediment volume is nearly equivalent for Runs 3 – 6. The overall distributary pattern, gently sloping profiles, and more expansive scale are most similar to described splays in the Saskatchewan system (e.g. Smith and Perez-Arlucea, 1994; Perez-Arlucea and Smith, 1999; Farrell, 2001) and possibly the ancient Fort Union and Willwood Formations (e.g. Kraus and Wells, 1999). Ultimately, these six model runs suggest that overbank sedimentation via levee crevasses may be common—as splay-like deposits were produced in all of the models, except Run 2—but that extensive basin-filling crevasse-splays may only occur in a rather limited parameter space. Expansive splays only occurred in well-drained systems featuring relatively abundant sand and large differences between channel and floodplain water-surface elevations. It’s possible that in other settings, the appropriate grain sizes may shift. For example, fluvial systems in the upland Northern Pennines, United Kingdom produced splays featuring cobbles and boulders in a sand matrix (Table 1.1) (Macklin et al., 1992), but the river is substantially steeper.

95

Figure 6.1: Abbreviated modeling flow plan. Modeling of the three grain-size distributions will consist of two sets of computations comprising the same grid space and parameters except for exterior floodplain boundary conditions that will be varied to account for “wet” and “dry” floodplain conditions.

96

Figure 6.2: Modeled grid and bathymetry. (A) Upper figure shows the full grid. The long blue “line” at the bottom is the channel and the blue “rectangle” in the middle is the wider grid that includes the floodplain. Grid spacing is too fine to resolve much detail in channel or most of the floodplain. (B) The grey rectangle from A is the region in the crevasse’s vicinity and is shown in the lower figure (B). In this area, grid spacing is 5 x 5 meters. 97

Table 6.1: Grid and Bathymetry Conditions. Domain Parameter Value(s) Channel and Levee Grid Cells in X Direction 321 Grid Cells Channel and Levee Grid Cells in Y Direction 15 Grid Cells Floodplain Grid Cells in X Direction 267 Floodplain Grid Cells in Y Direction 236 Range in X Grid Cell sizes 5 – 1000 meters Range in Y Grid Cell sizes 5 – 100 meters Length of Grid in X Direction 23550 meters Length of Grid in Y Direction 2644 meters Area of Grid near Crevasse (“Wider Grid”) 3067 x 2634 meters Width of Channel-Levee Banks 5 meters (1 grid cell) Height of Channel-Levee (initial lip) Banks 3 meters Width of Levee-Floodplain Bank 5 meters (1 grid cell) Width of Levee Top 10 meters (2 grid cells) Width of Crevasse Opening at Levee Base/Top 5 meters (1 grid cell)/15 meters (3 grid cells) Approximate Distance to Crevasse Opening 11270 meters

98

Table 6.2: Modeled Sediment Fractions. Sediment Grain Type Corresponding Assigned Settling Assigned Type Grain Size Velocity Critical (Cohesive) Shear Stress (n/m2) Cohesive Clay 1.95 Microns 0.0034 mm/s 2.8 Cohesive Very Fine-Fine Silt 7.81 Microns 0.055 mm/s 2.8 Cohesive Medium-Coarse Silt 31.25 Microns 0.88 mm/s 2.8 Non-Cohesive Very Fine Sand 93.75 Microns -- -- Non-Cohesive Fine Sand 187.5 Microns -- -- Non-Cohesive Medium Sand 375 Microns -- -- Non-Cohesive Coarse Sand 750 Microns -- --

99

Table 6.3: Vertical Sediment Thicknesses of Modeled Grain-size Distributions. Grain Type Fine-Grained Intermediate- Coarse-Grained Distribution Models Grained Distribution Models (meters) Distribution Models (meters) (Channel/Floodplain) (meters) (Channel/Floodplain) (Channel/Floodplain) Clay 12 / 12 0.56832 / 12 0.049152 / 12 Very Fine-Fine Silt 4.8 / 4.8 1.4208 / 4.8 0.12288 / 4.8 Medium-Coarse Silt 1.92 / 1.92 3.552 / 1.92 0.3072 / 1.92 Very Fine Sand 0.768 / 0.768 8.88 / 0.768 0.768 / 0.768 Fine Sand 0.3072 / 0.3072 3.552 / 0.3072 1.92 / 0.3072 Medium Sand 0.12288 / 0.12288 1.4208 / 0.12288 4.8 / 0.12288 Coarse Sand 0.049152 / 0.049152 0.56832 / 0.049152 12 / 0.049152 Cumulative Thickness 19.967232 / 19.967232 19.96224 / 19.967232 19.967232 / 19.967232

100

Figure 6.3: Proportions of Each Sediment Fraction in Model Runs. This depicts the proportions of sediments in the channel beds for each model run. The fine channel fraction is the same as the floodplain bed fractions used in all of the models. Note that specific cohesive sediment fractions assigned for the upstream boundary condition are presented in Table 6.4 and that non-cohesive sediment flow fractions are determined within the model from these bed fractions. 101

Table 6.4: Final Model Parameters, Initial Conditions, and Boundary Conditions. These are the exact values changed from defaults or utilized in the final model runs. Parameter/Condition Advective Inundation Diffusive Inundation Model Length (days) Variable Variable Model Time Step (minutes) 0.2 0.2 Surface Roughness (m1/2/s) 43 43 Initial Hydrodynamic Conditions From Set-up Model Map File From Set-up Model Map File Sediment Effect on Fluid Density Yes Yes Morphological Scaling Factor 80x 80x Morphological Spin-up Time (hours) 0 0 Boundary Conditions Upstream Channel: Discharge (m3/s) 130 130 Upstream Channel: Fine Model Clay: 5.6 x 10-3 Clay: 5.6 x 10-3 Sediment Concentration (kg/m3) Finer Silt: 1.9 x 10-3 Finer Silt: 1.9 x 10-3 Coarser Silt: 1.9 x 10-4 Coarser Silt: 1.9 x 10-4 Upstream Channel: Intermediate Clay: 1.2 x 10-4 Clay: 1.2 x 10-4 Model Sediment Concentration Finer Silt: 2.5 x 10-4 Finer Silt: 2.5 x 10-4 (kg/m3) Coarser Silt: 1.6 x 10-4 Coarser Silt: 1.6 x 10-4 Upstream Channel: Coarse Model Clay: 8.1 x 10-6 Clay: 8.1 x 10-6 Sediment Concentration (kg/m3) Finer Silt: 1.8 x 10-5 Finer Silt: 1.8 x 10-5 Coarser Silt: 1.1 x 10-5 Coarser Silt: 1.1 x 10-5 Downstream Channel: QH Relation Minimum—0 m3/s : 0 meters Minimum—0 m3/s : 0 meters (Discharge: Water-surface Level) Maximum—130 m3/s : 2.939 meters Maximum—130 m3/s : 2.939 meters 3 Floodplain Boundaries: WSL 2.78 meters (~5 cm above floodplain 3.98 meters (~20 cm below water- surface) surface in channel without crevasse)

102

Figure 6.4: Cumulative Sedimentation in Channel with No Crevasse. This graph documents the cumulative erosion and sedimentation in the middle reach of the channel over the course of 45 simulated days. Given the small variance of channel bed elevation, this graph shows that the modeled channel is near equilibrium and graded.

103

Figure 6.5: Initial Flow Magnitudes for Dry Floodplain Models. The expansion angle for each side of the turbulent jet is nearly 45° from the primary flow direction, and flow is primarily directed away from the crevasse. 104

Figure 6.6: Initial Flow Magnitudes for Wet Floodplain Models. The expansion angle for each side of the turbulent jet is less than in the dry floodplain runs, and flow is directed in a more downstream direction than for the dry floodplain runs (Figure 6.5). 105

Figure 6.7: Crevasse and Downstream Discharges for Run 1.

106

Figure 6.8: Final Flow Magnitudes (Day 162) for Run 1: Fine Channel Sediment with Dry Floodplain.

107

Figure 6.9: Cumulative Erosion/Sedimentation (Day 162) for Run 1: Fine Channel Sediment with Dry Floodplain.

108

Figure 6.10: Run 1: Cross-sections A, B, and C.

109

Figure 6.11: Run 1: Cross-sections D, E, and F.

110

Figure 6.12: Cumulative Total Sediment Transport Through Crevasse. These are the sediment volumes passed through a cross-section placed at the exit of the crevasse. Note that Delft3D calculates transport values ignoring the morphological scaling factor and assigned bulk sediment densities, thus this number is much lower than values based on simulated deposition (as shown in Table 6.5, for example).

111

Figure 6.13: Instantaneous Total Sediment Discharge Through Crevasse. These are the instantaneous sediment discharges passed through a cross-section placed at the exit of the crevasse. Note that Delft3D calculates transport values ignoring the morphological scaling factor and assigned bulk sediment densities, thus this number is much lower than values assumed simulated deposition (as shown in Table 6.5, for example).

112

Figure 6.14: Normalized Instantaneous Sediment Discharge Through Crevasse. These are the instantaneous sediment discharges normalized by the equilibrium instantaneous sediment discharge observed in preliminary set-up models used to determine the initial hydrodynamics for the final runs.

113

Table 6.5: Surface Areas and Volumes for Splays Produced within Models. Note that the following contour intervals include everything within (or beyond) that contour on the floodplain. Therefore, even areas within or near the crevasse that underwent erosion or saw deposition less than the given value are included in the > 0.3 and > 0.1 meter classifications. Also, Run 2 featured limited deposition and is listed with a maximum contour interval of 0.2 meters.

Model Contour Interval # Grid Cells Surface Area (m2) Volume (m3) > 0.3 meter 7591 189,775 61,643 Run 1 > 0.1 meter 40429 1,291,693 261,730 < 0.1 meter 22322 6,556,782 105,400 > 0.2 meter 17 425 61.4 Run 2 > 0.1 meter 970 24,250 2,968 < 0.1 meter 61781 7,824,225 72,688 > 0.3 meter 12833 320,825 138,637 Run 3 > 0.1 meter 39132 1,200,450 287,179 < 0.1 meter 23619 6,648,025 101,297 > 0.3 meter 5694 142,350 160,962 Run 4 > 0.1 meter 6027 150,675 162,446 < 0.1 meter 56724 7,697,800 9050 > 0.3 meter 12595 314,875 146,582 Run 5 > 0.1 meter 38245 1,156,855 289,042 < 0.1 meter 24506 6,691,620 97,416 > 0.3 meter 5959 148,975 170,634 Run 6 > 0.1 meter 6134 153,350 171,399 < 0.1 meter 56617 7,695,125 2,369

114

Figure 6.15: Crevasse and Downstream Discharges for Run 2.

115

Figure 6.16: Final Flow Magnitudes (Day 153.5) for Run 2: Fine Channel Sediment with Wet Floodplain.

116

Figure 6.17: Cumulative Erosion/Sedimentation (Day 153.5) for Run 2: Fine Channel Sediment with Wet Floodplain.

117

Figure 6.18: Run 2: Cross-sections A, B, and C.

118

Figure 6.19: Run 2: Cross-sections D, E, and F.

119

Figure 6.20: Crevasse and Downstream Discharges for Run 3.

120

Figure 6.21: Final Flow Magnitudes (Day 198) for Run 3: Intermediate Channel Sediment with Dry Floodplain.

121

Figure 6.22: Cumulative Erosion/Sedimentation (Day 198) for Run 3: Intermediate Channel Sediment with Dry Floodplain.

122

Figure 6.23: Run 3: Cross-sections A,B, and C.

123

Figure 6.24: Run 3: Cross-sections D, E, and F.

124

Figure 6.25: Crevasse and Downstream Discharges for Run 4. 125

Figure 6.26: Final Flow Magnitudes (Day 148) for Run 4: Intermediate Channel Sediment with Wet Floodplain.

126

Figure 6.27: Cumulative Erosion/Sedimentation (Day 148) for Run 4: Intermediate Channel Sediment with Wet Floodplain.

127

Figure 6.28: Run 4: Cross-sections A, B, and C.

128

Figure 6.29: Run 4: Cross-sections D, E, and F.

129

Figure 6.30: Crevasse and Downstream Discharges for Run 5.

130

. Figure 6.31: Final Flow Magnitudes (Day 196) for Run 5: Coarse Channel Sediment with Dry Floodplain.

131

Figure 6.32: Cumulative Erosion/Sedimentation (Day 196) for Run 5: Coarse Channel Sediment with Dry Floodplain.

132

Figure 6.33: Run 5: Cross-sections A, B, and C.

133

Figure 6.34: Run 5: Cross-sections D, E, and F.

134

Figure 6.35: Crevasse and Downstream Discharges for Run 6. 135

Figure 6.36: Final Flow Magnitudes (Day 141.5) for Run 6: Coarse Channel Sediment with Wet Floodplain.

136

. Figure 6.37: Cumulative Erosion/Sedimentation (Day 75) for Run 6: Coarse Channel Sediment with Wet Floodplain.

137

Figure 6.38: Run 6: Cross-sections A, B, and C.

138

Figure 6.39: Run 6: Cross-sections D, E, and F.

139

7.0 Synthesis and Discussion

The importance of the two proposed primary controls—grain-size distribution and floodplain inundation—on crevasse-splay development and growth is evident in the study of actual deposits (i.e. the modern and ancient study phases). In the modern systems, I observed the intermediate- grained (predominately medium silt through fine sand) and well-drained (“dry” floodplain with abundant advective-transport of sediment) Saskatchewan River produced large-scale distributary splays responsible for 50% or more of flood-basin filling. In contrast, the coarse-grained (primarily medium to very coarse sand) and poorly drained (“wet” floodplain with limited advective-transport of sediment) upper Columbia River produced numerous splays, but these relatively small, and fill only a small proportion of the floodbasin. The coarse-grained, dryland Sandover River produced abundant basin-covering splays, but these splays are highly erosive and do not contribute significant sediment towards basin-filling (e.g. Tooth, 2005), whereas, the Ovens River produced no observed splays. Similar trends were observed in the ancient systems, where the Fort Union and Willwood formations feature abundant medium silt to fine sand and well-drained floodplains, and produced large-scale splays and avulsion deposits, while the Ferris Formation had a waterlogged floodplain and lacked substantial intermediate-sized sediment and produced very few splays and no avulsion-related deposits. Delft3D modeling helped compare the relative influence of sediment size and floodplain drainage on crevasse-splay development. Simplified models show that large-scale, distributary- type splay deposits only form in the intermediate and coarse-grained dry floodplain (advective- transport-dominated) runs, which suggests that this type of splay formation requires well-drained floodplains and a sufficient supply of "intermediate" sediment sizes that can escape the channel as suspended sediment but are readily deposited on the floodplain. In the other runs, limited or no splay development occurred. Another important modeling observation is the amount of floodplain incision within the wet floodplain (diffusive-transport) runs. All three of these runs featured limited crevasse and proximal floodplain incision, which would limit the crevasse’s potential discharge and possibly encourage crevasse healing. Additionally, this has implications for the analogous upper Columbia River. Since there is little floodplain incision, it implies that produced splays/channels

140

are already perched and nearly superelevated. Thus, the very method of channel production in this system may encourage frequent avulsion. While large-scale, distributary style splay development is closely linked to progradational style avulsion processes in the modern Saskatchewan River, Rhine-Meuse Delta, and Mississippi River, this type of deposition is not inherent in incisionally avulsive systems. The study of modern systems found no splay development in the Ovens River despite the production of an alluvial ridge, crevasses, and floodplain siltation (Binnie and Partners, 1984; Schumm et al., 1984; Judd et al., 2007). While it could be argued that net aggradation in this system is insufficient for splay production, splays are not observed in other incisionally avulsive systems with long-term net aggradation such as the Okavango Fan (Botswana) (McCarthy et al., 1992; 1993; Smith et al., 1997) and Australia’s “Channel Country” (Gibling et al., 1998), nor are splays (or levees and alluvial ridges) observed in the rapidly aggrading (~1–2 mm/y) Narew River (Poland) (Gradzinski et al., 2003). Notably, the “Channel Country” and Narew River systems feature limited quantities of intermediate-sized sediments, while the Okavango Fan features such a low bed slope that even finer sands are transported as bedload (e.g. McCarthy et al., 1992; Smith et al., 1997). Elsewhere, the Baghmati River (India) and Magela Creek (Australia) avulse via incisional processes, are rapidly aggrading, and deposit splays (Jain and Sinha, 2004; Sinha et al., 2005; Tooth et al., 2008). These systems all transport significant amounts of fine sand (Sinha et al., 2005; Tooth et al., 2008). Additionally, Sinha et al. (2005) note that overbank deposition in the Baghmati system appears to result from repeated flooding, rather than extensive avulsion-related deposition, which contrasts with the dominant style in modern progradational systems and ancient stratigraphically transitional systems. Overall, these results also have important implications for the inter-connectedness of sub- surface reservoirs. As observed in all three study components, systems with relatively wide, well- drained floodplains, and abundant intermediate-sized sediments produce large-scale splay deposits, which should promote increased reservoir size and inter-connectedness. On the other hand, systems with poorly drained floodplains (with no cross-floodplain water-surface slope) and fine-grained sediments should produce small, isolated sand bodies.

141

8.0 Conclusions

This research aimed to test the hypotheses that large splays develop in systems (1) where floodplain drainage conditions allow for a distinct water-surface gradient to develop between the channel and floodplain, which drives sediment onto the floodplain via advective currents, and (2) with an intermediate grain-size distribution (i.e. coarse silt to fine sand in sand-bed rivers) that is fine enough to be easily suspended within channels, but also coarse enough to fall out of suspension during overbank flow expansion. This study’s results suggest that splay deposition is greater in systems with well-drained floodplains dominated by advective transport, as seen in the Saskatchewan River and the ancient Fort Union and Willwood formations. Additionally, Delft3D models show that for a given grain- size distribution, splays deposited on “dry” (well-drained) floodplain runs with steep cross- floodplain water-surface slopes should be larger and more voluminous than splays produced in “wet” (poorly-drained) systems. Similarly, the through-crevasse sediment discharge is always greater within the “dry” floodplain run. Beyond its implications for advective and diffusive sediment-transport processes, floodplain water-surface slope is also an important control on water discharge through crevasses. The wet model runs featured significantly lower discharge levels than the dry model runs. Thus, the effect of water-surface slope is two-fold: (1) it controls the distance sediment is advected onto the floodplain, and (2) it controls the amount of sediment that can be deposited on the floodplain. Overall, these findings support Hypothesis I. This study’s results also provide evidence that systems featuring an intermediate grain- size distribution will produce the most extensive splays. Again, the Saskatchewan is a modern system featuring abundant intermediate-sized sediments, as do the ancient Fort Union and Willwood Formations. The splays produced within these systems are also larger than those within the upper Columbia and Ferris Formation, which feature coarse-grained distributions. These results are supported by Delft3D modeling, where the model runs featuring predominately fine-grained sediments produced the smallest and least voluminous splays, and had the lowest through-crevasse sediment discharges relative to equivalent intermediate and coarse-grained runs. However, Delft3D models featuring a predominately coarse-grained distribution were effectively equivalent in size, volume, and sediment discharge to the corresponding intermediate- grained distribution with the same floodplain drainage. Furthermore, the fine-grained, “dry” 142

floodplain run produced a splay larger, more voluminous, and with a higher sediment discharge than any “wet” floodplain run. Thus, based on these physical and modeling results, Hypothesis II is supported for differences between intermediate and fine-grained systems, but is inconclusive for differences between intermediate and coarse-grained systems. Ultimately, these results suggest that while floodplain conditions promoting a steep, lateral water-surface gradient are necessary for producing large splays, it is not sufficient on its own. Abundant sandy sediment must also be available to produce large, basin-filling crevasse- splay deposits.

143

References

ABBADO, D., SLINGERLAND, R., SMITH, N.D., 2005. Origin of anastomosis in the upper Columbia River, British Columbia, Canada. In: Blum, M.D., Marriot, S.B., Leclair, S.M. (Eds.), Fluvial Sedimentology VII. Special Publication of the International Association of Sedimentologists, vol. 35. Blackwell, Oxford, UK, pp. 3–15.

ADAMS, P.N., SLINGERLAND, R.L., and SMITH, N.D., 2004, Variations in natural levee morphology in anastomosed channel flood plain complexes: Geomorphology, v. 61, p. 127-142.

ARCO, L.J., ADELSBERGER, K.A., HUNG, L.Y., and KIDDER, T.R., 2006, Alluvial geoarchaeology of a Middle Archaic mound complex in the lower Mississippi valley, USA: Geoarchaeology- an International Journal, v. 21, p. 591-614.

ASHMORE, P.E., and DAY, T.E., 1988, Spatial and temporal patterns of suspended-sediment yield in the Saskatchewan River basin: Canadian Journal of Earth Science, v. 25, p. 1450-1463.

ASLAN, A., and AUTIN, W.J., 1999, Evolution of the Holocene Mississippi River floodplain, Ferriday, Louisiana: Insights on the origin of fine-grained floodplains: Journal of Sedimentary Research, v. 69.

ASLAN, A., and BLUM, M.D., 1999, Contrasting styles of Holocene avulsion, Texas Gulf Coastal Plain, USA: Fluvial Sedimentology Vi, v. 28.

BERNAL, C., CHRISTOPHOUL, F., DARROZES, J., LARAQUE, A., BOURREL, L., SOULA, J-C., GUYOT,

J-L., and BABY, P., in press, Crevassing and capture by floodplain drains as a cause of partial avulsion and anastomosis (lower Rio Pastaza, Peru), Journal of South American Earth Sciences.

BINNIE and PARTNERS, 1984, Ovens River floodplains and river management study: Report to Rural Water Commission of Victoria, 233 p.

BLUM, M.D., GUCCIONE, M.J., WYSOCKI, D.A., ROBNETT, P.C., and RUTLEDGE, E.M., 2000, Late Pleistocene evolution of the lower Mississippi River valley, southern Missouri to Arkansas: Geological Society of America Bulletin, v. 112, p. 221-235.

BOWN, T.M., 1980, Summary of latest Cretaceous and Cenozoic sedimentary, tectonic, and erosional events, Bighorn Basin, Wyoming. In: Gingerich, P.D. (Ed.), Early Cenozoic Paleontology and Stratigraphy of the Bighorn Basin, Wyoming, p. 25-32.

BOWN, T.M., and KRAUS, M.J., 1987, Integration of channel and floodplain suites i. developmental 144

sequence and lateral relations of alluvial paleosols: Journal of Sedimentary Petrology, v. 57.

BRIDGE, J.S., and DEMICCO, R.V., 2008, Earth Surface Processes, Landforms, and Sediment Deposits, Cambridge University Press, pp. 815.

BRIDGE, J.S., and LEEDER, M.R., 1979, Simulation-model of alluvial stratigraphy: Sedimentology, v. 26.

BRISTOW, C.S., SKELLY, R.L., and ETHRIDGE, F.G., 1999, Crevasse splays from the rapidly aggrading, sand-bed, braided Niobrara River, Nebraska: effect of base-level rise: Sedimentology, v. 46.

CAZANACLI, D., and SMITH, N.D., 1998, A study of morphology and texture of natural levees - Cumberland Marshes, Saskatchewan, Canada: Geomorphology, v. 25. nd CHAUNDRY,M.H., 2007, Open-Channel Flow, 2 edition, Springer, p. 505.

COTTINGHAM, P., HANNAN, G., HILLMAN, T., KOEHN, J., METZLING, L., ROBERTS, J., RUTHERFURD, I., 2001, Report of the Ovens Scientific Panel on the Environmental Condition and Flows of the Ovens River, Technical Report 9/2001, 90 p.

EBERLE, J.J., and LILLEGRAVEN, J.A., 1998, A new important record of earliest Cenozoic mammalian history: geologic setting, Multituberculata, and Peradectia: Rocky Mountain Geology, v. 33, p. 3-47.

EBERTH, D.A., and MIALL, A.D., 1991, Stratigraphy, sedimentology and evolution of a vertebrate- bearing, braided to anastomosed fluvial system, Cutler Formation (Permian Pennsylvanian), north-central New-Mexico: Sedimentary Geology, v. 72.

FARRELL, K.M., 2001, Geomorphology, facies architecture, and high-resolution, non-marine sequence stratigraphy in avulsion deposits, Cumberland Marshes, Saskatchewan: Sedimentary Geology, v. 139, p. 93-150.

FEDELE, J.J., and PAOLA, C., 2007, Similarity solutions for fluvial sediment fining by selective deposition: Journal of Geophysical Research-Earth Surface, v. 112.

FISHER, J.A., KRAPF, C.B.E., LANG, S.C., NICHOLS, G.J., and PAYENBERG, T.H.D., 2008, Sedimentology and architecture of the Douglas Creek terminal splay, Lake Eyre, central Australia: Sedimentology, v. 55, p. 1915-1930.

FLORSHEIM, J.L., and MOUNT, J.F., 2002, Restoration of floodplain topography by sand-splay complex formation in response to intentional levee breaches, Lower Cosumnes River, 145

California: Geomorphology, v. 44, p. 67-94.

GIBLING, M.R., NANSON, G.C., and MAROULIS, J.C., 1998, Anastomosing river sedimentation in the Channel Country of central Australia: Sedimentology, v. 45.

GRADZINSKI, R., BARYLA, J., DOKTOR, M., GMUR, D., GRADZINSKI, M., KEDZIOR, A., PASZKOWSKI,

M., SOJA, R., ZIELINSKI, T., and ZUREK, S., 2003, Vegetation-controlled modern anastomosing system of the upper Narew River (NE Poland) and its sediments: Sedimentary Geology, v. 157.

GREENWOOD, D.R., and WING, S.L., 1995, Eocene continental climates and latitudinal temperature-gradients: Geology, v. 23.

HAJEK, E.A., HELLER, P.L., and SHEETS, B.A., 2010, Significance of channel-belt clustering in alluvial basins: Geology, v. 38.

HAJEK, E.A., HELLER, P.L., SCHUR, E.L., 2012, Field test of autogenic control on alluvial stratigraphy (Ferris Formation, Upper Cretaceous-Paleogene, Wyoming), Geological Society of America Bulletin.

HAJEK, E.A., and WOLINSKY, M.A., 2012, Simplified process modeling of river avulsion and alluvial architecture: Connecting models and field data: Sedimentary Geology, v. 257.

HESSELINK, A.W., WEERTS, H.J.T., and BERENDSEN, H.J.A., 2003, Alluvial architecture of the human-influenced river Rhine, The Netherlands: Sedimentary Geology, v. 161, p. 229- 248.

HICKEY, L.J., 1980, Paleocene stratigraphy and flora of the Clark's Fork Basin. In: Gingerich, P.D. (Ed.), Early Cenozoic Paleontology and Stratigraphy of the Bighorn Basin, Wyoming, p. 33-49.

INTERNATIONAL ASSOCIATION for HYDRO-ENVIRONMENT ENGINEERING and RESEARCH (IAHR), 1994, Guidelines for documenting the validity of computational modeling software, 24 pp.

JAIN, V., and SINHA, R., 2004, Fluvial dynamics of an anabranching river system in Himalayan foreland basin, Baghmati river, north Bihar plains, India: Geomorphology, v. 60.

JAIN, V., and SINHA, R., 2005, Response of active tectonics on the alluvial Baghmati River, Himalayan foreland basin, eastern India: Geomorphology, v. 70.

JONES, H.L., 2007, Characterizing Ancient Avulsion Stratigraphy and its Significance. Unpublished Ph.D. dissertation, Department of Geology and Geophysics, University of 146

Wyoming, pp. 223.

JONES, H.L., and HAJEK, E.A., 2007, Characterizing avulsion stratigraphy in ancient alluvial deposits: Sedimentary Geology, v. 202.

JORGENSEN, P.J., and FIELDING, C.R., 1996, Facies architecture of alluvial floodbasin deposits: Three-dimensional data from the Upper Triassic Callide Coal Measures of east-central Queensland, Australia: Sedimentology, v. 43, p. 479-495.

JUDD, D.A., RUTHERFURD, I.D., TILLEARD, J.W., and KELLER, R.J., 2007, A case study of the processes displacing flow from the anabranching Ovens River, Victoria, Australia: Earth Surface Processes and Landforms, v. 32.

JULIEN, P.Y., 2002, River Mechanics, University Press, Cambridge, p. 621.

KLEINHANS, M.G., WEERTS, H.J.T., and COHEN, K.M., 2010, Avulsion in action: Reconstruction and modeling sedimentation pace and upstream flood water levels following a Medieval tidal-river diversion catastrophe (Biesbosch, The Netherlands, 1421-1750 AD): Geomorphology, v. 118, p. 65-79.

KRAUS, M.J., 1996, Avulsion deposits in lower Eocene alluvial rocks, Bighorn Basin, Wyoming: Journal of Sedimentary Research, v. 66.

KRAUS, M.J., 1998, Development of potential acid sulfate paleosols in Paleocene floodplains, Bighorn Basin, Wyoming, USA: Palaeogeography Palaeoclimatology Palaeoecology, v. 144.

KRAUS, M.J., and ASLAN, A., 1993, Eocene hydromorphic paleosols - significance for interpreting ancient floodplain processes: Journal of Sedimentary Petrology, v. 63.

KRAUS, M.J., and GWINN, B., 1997, Facies and facies architecture of Paleogene floodplain deposits, Willwood Formation, Bighorn Basin, Wyoming, USA: Sedimentary Geology, v. 114.

KRAUS, M.J., and WELLS, T.M., 1999, Recognizing avulsion deposits in the ancient stratigraphical record: Fluvial Sedimentology Vi, v. 28.

LANG, S.C., PAYENBERG, T.H.D., REILLY, M.R.W, HICKS, T., BENSON, J., and KASSAN, J., 2004, Modern analogues for dryland sandy fluvial-lacustrine deltas and terminal splay reservoirs, The APPEA Journal, v. 44, p. 329-356.

LAVOOI, E., 2010, Origin of anastomosis, upper Columbia River, British Columbia, Canada. Unpublished MSc. Thesis, Faculty of Geosciences, Utrecht University, 70 pp. 147

LOCKING, T., 1983, Hydrology and sediment transport in an anastomosing reach of the Upper Columbia River, B.C. Unpublished MSc. Thesis, Department of Geography, University of Calgary, Canada, 107 pp.

MACHUSICK, M.D., 2000, The effect of floodbasin type on crevasse splay form. Unpublished BSc. Thesis, Department of Geosciences, The Pennsylvania State University, 55 pp.

MACK, G.H., JAMES, W.C., and MONGER, H.C., 1993, Classification of Paleosols: Geological Society of America Bulletin, v. 105.

MACKLIN, M.G., RUMSBY, B.T., and HEAP, T., 1992, Flood alluviation and entrenchment - Holocene valley-floor development and transformation in the British uplands: Geological Society of America Bulletin, v. 104.

MAKASKE, B., 1998. Anastomosing Rivers; Forms, Processes and Sediments. Nederlandse Geografische Studies 249. Koninklijk Nederlands Aardrijkskundig Genootschap/ Faculteit RuimtelijkeWetenschappen, Universiteit Utrecht, Utrecht, The Netherlands.

MAKASKE, B., SMITH, D.G., and BERENDSEN, H.J.A., 2002, Avulsions, channel evolution and floodplain sedimentation rates of the anastomosing upper Columbia River, British Columbia, Canada: Sedimentology, v. 49.

MAKASKE, B., SMITH, D.G., BERENDSEN, H.J.A., BOER, A.G.D., VAN NIELEN-KIEZEBRINK, M.F., and

LOCKING, T., 2009, Hydraulic and sedimentary processes causing anastomosing morphology of the upper Columbia River, British Columbia, Canada: Geomorphology, v. 111.

MCCARTHY, T.S., ELLERY, W.N., and ELLERY, K., 1993, Vegetation-induced, subsurface precipitation of carbonate as an aggradational process in the permanent swamps of the Okavango (Delta) Fan, Botswana: Chemical Geology, v. 107.

MCCARTHY, T.S., ELLERY, W.N., and STANISTREET, I.G., 1992, Avulsion mechanisms on the Okavango Fan, Botswana - the control of a fluvial system by vegetation: Sedimentology, v. 39.

MEADE, R.H., 1985, Suspended sediment in the Amazon River and its tributaries in Brazil during 1982-84: United States Geological Survey, Open-File Report 85-492, 39 pp.

MERTES, L.A.K., SMITH, M.O., and ADAMS, J.B., 1993, Estimating suspended sediment concentrations in surface waters of the Amazon River wetlands from Landsat images: Remote Sensing of Environment, v. 43, p. 281-301. 148

MICHAELSEN, P., HENDERSON, R.A., CROSDALE, P.J., and MIKKELSEN, S.O., 2000, Facies architecture and depositional dynamics of the Upper Permian Rangal Coal Measures, Bowen Basin, Australia: Journal of Sedimentary Research, v. 70, p. 879-895.

MIER, J.M., and GARCIA, M.H., 2011, Erosion of glacial till from the St. Clair River (Great Lakes basin): Journal of Great Lakes Research, v. 37.

MILLER, J.R., 1991, Development of anastomosing channels in south-central Indiana: Geomorphology, v. 4.

MJOS, R., WALDERHAUG, O., PRESTHOLM, E., 1993, Crevasse splay sandstone geometries in the Middle Ravenscar Group of Yorkshire, UK, Special Publications of the International Association of Sedimentologists, v. 17, p. 167-184.

MOHRIG, D., HELLER, P.L., PAOLA, C., and LYONS, W.J., 2000, Interpreting avulsion process from ancient alluvial sequences: Guadalope-Matarranya system (northern Spain) and Wasatch Formation (western Colorado): Geological Society of America Bulletin, v. 112.

OBRIEN, P.E., and WELLS, A.T., 1986, A small, alluvial crevasse splay: Journal of Sedimentary Petrology, v. 56, p. 876-879.

PARKER, G., 1991, Selective sorting and abrasion of river gravel .1. Theory: Journal of Hydraulic Engineering-Asce, v. 117, p. 131-149.

PEREZ-ARLUCEA, M., and SMITH, N.D., 1999, Depositional patterns following the 1870s avulsion of the Saskatchewan River (Cumberland Marshes, Saskatchewan, Canada): Journal of Sedimentary Research, v. 69.

PIZZUTO, J.E., 1987, Sediment diffusion during overbank flows, Sedimentology, v. 34, p. 301 – 317.

PONTEN, A., and PLINK-BJORKLUND, P., 2007, Depositional environments in an extensive tide- influenced delta plain, Middle Devonian Gauja Formation, Devonian Baltic Basin: Sedimentology, v. 54, p. 969-1006.

PRANTER, M.J., COLE, R.D., PANJAITAN, H., and SOMMER, N.K., 2009, Sandstone-body dimensions in a lower coastal-plain depositional setting: Lower Williams Fork Formation, Coal Canyon, Piceance Basin, Colorado: Aapg Bulletin, v. 93, p. 1379-1401.

RAJARATNAM, N., and AHMADI, R.M., 1979, Interaction between main channel and flood-plain flows: Journal of the Hydraulics Division, v. 105, p. 573-588.

REYNOLDS, A.D., 1999, Dimensions of paralic sandstone bodies: AAPG Bulletin-American 149

Association of Petroleum Geologists, v. 83, p. 211-229.

RIPPIN, K. G., and SHEEHAN, D. B., 1986, Ovens River floodplain management, Hydrology and Water Resources Symposium, 1986, Griffith University, Brisbane, 25–27 November 1986: Sydney, Institution of Engineers, Australia, p. 181–185.

ROBERTS, H.H., 1997, Dynamic changes of the Holocene Mississippi River delta plain: The delta cycle: Journal of Coastal Research, v. 13, p. 605-627.

ROUSE, H., 1937, Modern conceptions of mechanics of fluid turbulence. Trans. Am. Soc. Civ. Eng., 102, p. 436-505.

SARTI, G., ZANCHETTA, G., MAZZA, P., and GRASSI, R., 2001, Sedimentological and palaeontological features of an ancient alluvial plain in the Lucca Basin (Central Italy): Eclogae Geologicae Helvetiae, v. 94, p. 107-117.

SCHUMANN, R.R., 1989, Morphology of Red Creek, Wyoming, an arid-region anastomosing channel system: Earth Surface Processes and Landforms, v. 14.

SCHUMM, S.A., ERSKINE, W.D., and TILLEARD, J.W., 1996, Morphology, hydrology, and evolution of the anastomosing ovens and King Rivers, Victoria, Australia: Geological Society of America Bulletin, v. 108.

SHAH-FAIRBANK, S.C., JULIEN, P.Y., BAIRD, D.C., 2011, Total Sediment Load from SEMEP Using Depth-Integrated Concentration Measurements, Journal of Hydraulic Engineering, v. 137, p. 1606-1614.

SIMONSON, G.H., and BOERSMA, L., 1972, Soil morphology and water table relations .2. Correlation between annual water table fluctuations and profile features: Soil Science Society of America Proceedings, v. 36, p. 649-&.

SINHA, R., FRIEND, P.F., and SWITSUR, V.R., 1996, Radiocarbon dating and sedimentation rates in the Holocene alluvial sediments of the northern Bihar plains, India: Geological Magazine, v. 133, p. 85-90.

SINHA, R., GIBLING, M.R., JAIN, V., and TANDON, S.K., 2005, Sedimentology and avulsion patterns of the anabranching Baghmati River in the Himalayan foreland basin, India. In: Blum, M.D., Marriot, S.B., Leclair, S.M. (Eds.), Fluvial Sedimentology VII. Special Publication of the International Association of Sedimentologists, v. 35. Blackwell, Oxford, UK, p. 3–15.

SLINGERLAND, R., and SMITH, N.D., 1998, Necessary conditions for a meandering-river avulsion: 150

Geology, v. 26.

SLINGERLAND, R., and SMITH, N.D., 2004, River avulsions and their deposits: Annual Review of Earth and Planetary Sciences, v. 32, p. 257-285.

SMITH, D.G., 1986, Anastomosing river deposits, sedimentation-rates and basin subsidence, Magdalena River, northwestern Colombia, South America: Sedimentary Geology, v. 46.

SMITH, N.D., CROSS, T.A., DUFFICY, J.P., and CLOUGH, S.R., 1989, Anatomy of an avulsion: Sedimentology, v. 36.

SMITH, N.D., MCCARTHY, T.S., ELLERY, W.N., MERRY, C.L., and RUTHER, H., 1997, Avulsion and anastomosis in the panhandle region of the Okavango Fan, Botswana: Geomorphology, v. 20.

SMITH, N.D., and PEREZ-ARLUCEA, M., 1994, Fine-grained splay deposition in the avulsion belt of the lower Saskatchewan River, Canada: Journal of Sedimentary Research Section B- Stratigraphy and Global Studies, v. 64.

SMITH, N.D., and PEREZ-ARLUCEA, M., 2008, Natural levee deposition during the 2005 flood of the Saskatchewan River, Geomorphology, v. 101, p. 583-594.

SMITH, N.D., SLINGERLAND, R.L., PEREZ-ARLUCEA, M., and MOROZOVA, G.S., 1998, The 1870s avulsion of the Saskatchewan River: Canadian Journal of Earth Sciences, v. 35.

SMITH, R.M.H., 1993, Sedimentology and ichnology of floodplain paleosurfaces in the Beaufort Group (Late Permian), Karoo Sequence, South Africa: Palaios, v. 8, p. 339-357.

STOUTHAMER, E., 2001, Sedimentary products of avulsions in the Holocene Rhine-Meuse Delta, The Netherlands: Sedimentary Geology, v. 145.

TABATA, K.K., and HICKIN, E.J., 2003, Interchannel hydraulic geometry and hydraulic efficiency of the anastomosing Columbia River, southeastern British Columbia, Canada: Earth Surface Processes and Landforms, v. 28.

TAYLOR, C.F.H., 1999, The role of overbank flow in governing the form of an anabranching river: the Fitzroy River, northwestern Australia: Fluvial Sedimentology Vi, v. 28, p. 77-91.

TOOTH, S., 1999a, Downstream changes in floodplain character on the Northern Plains of arid central Australia: Fluvial Sedimentology Vi, v. 28.

TOOTH, S.,1999b, Floodouts in Central Australia. In: MILLER, A.J. and GUPTA, A. (Eds.), Varieties of Fluvial Form, Publication No. 7, International Association of Geomorphologists, p. 219-248. 151

TOOTH, S., 2000a, Downstream changes in dryland river channels: the Northern Plains of arid central Australia: Geomorphology, v. 34.

TOOTH, S., 2000b, Process, form and change in dryland rivers: a review of recent research: Earth- Science Reviews, v. 51.

TOOTH, S., and NANSON, G.C., 1999, Anabranching rivers on the Northern Plains of arid central Australia: Geomorphology, v. 29.

TOOTH, S., and NANSON, G.C., 2000, Equilibrium and nonequilibrium conditions in dryland rivers: Physical Geography, v. 21.

TOOTH, S., 2005, Splay formation along the lower reaches of ephemeral rivers on the Northern Plains of arid central Australia, Journal of Sedimentary Research, v. 75, p. 636-649.

TOOTH, S., RODNIGHT, H., DULLER, G.A.T., MCCARTHY, T.S., MARREN, P.M., BRANDT, D., 2007, Chronology and controls of avulsion along a mixed bedrock-alluvial river, Geological Society of America Bulletin, v. 119, p. 452-461.

TOOTH, S., JANSEN, J., NANSON, G.C., COULTHARD, T.J., PIETSCH, T., 2008, Riparian vegetation and the late Holocene development of an anabranching river: Magela Creek, northern Australia, Geological Society of America Bulletin, v. 120, p. 1021-1035.

TYE, R.S., 2004, Geomorphology: An approach to determining subsurface reservoir dimensions: AAPG Bulletin, v. 88, p. 1123-1147.

VAN RIJN|, 1993. Principles of Sediment Transport in Rivers, Estuaries and Coastal Seas. Aqua Publications, The Netherlands.

VICTORIAN WATER RESOURCES DATA WAREHOUSE, online data, < http://www.vicwaterdata.net/vicwaterdata/data_warehouse_content.aspx?option=9 > Checked September, 10, 2012.

WATER SURVEY of CANADA, 1991. Historical Streamflow Summary, British Columbia, 1990. Water Survey of Canada, Ottawa.

WEERTS, H.J.T., and BIERKENS, M.F.P., 1993, Geostatistical analysis of overbank deposits of anastomosing and meandering fluvial systems - Rhine-Meuse Delta, the Netherlands: Sedimentary Geology, v. 85, p. 221-232.

WILLIS, B.J., and BEHRENSMEYER, A.K., 1994, Architecture of Miocene overbank deposits in northern Pakistan: Journal of Sedimentary Research Section B-Stratigraphy and Global Studies, v. 64. 152

WROBLEWSKI, A.F.J., 2004, New Selachian paleofaunas from "Fluvial" deposits of the Ferris and Lower Hanna Formations (Maastrichtian-Selandian : 66-58 Ma), southern Wyoming: Palaios, v. 19.

153

Appendix A: Splay Polygon Raw Values

Aspect Parent Maximum Splay Log Ratio Log Log Orthogona Normalized 10 Log Splay Area 10 Perimeter Channel 10 Splay Path Normalized (Path 10 Splay System Reach Source Origin Coordinates Terminus Coordinates Splay Channel l Extent Orthogonal Aspect (meters2) (meters) Width Width Length Orthogonal Length/ Area Width (meters) Extent Ratio (meters) (meters) (meters) Extent Maximum Width) Upper Makaske, Splay 1 Castledale 51.0238599°, -116.5476038° 51.0251180°, -116.5532656° 24,127.109 4.383 1,081.485 20.669 1.315 92.222 132.846 504.122 6.427 0.808 5.466 0.738 Columbia 1998 Upper Makaske, Splay 2 Castledale 51.0236113°, -116.5553082° 51.0238763°, -116.5523640° 11,940.916 4.077 684.503 20.677 1.315 102.064 176.271 253.083 8.525 0.931 2.480 0.394 Columbia 1998 Upper Castledale Makaske, Splay 3 51.0215804°, -116.5519178° 51.0213911°, -116.5488865° 18,995.693 4.279 684.226 45.592 1.659 141.548 120.818 318.138 2.650 0.423 2.248 0.352 Columbia /Main 1998 Upper Castledale Makaske, Splay 4 51.0202281°, -116.5543022° 51.0210174°, -116.5571491° 33,398.827 4.524 1,352.130 59.120 1.772 212.190 229.784 314.679 3.887 0.590 1.483 0.171 Columbia /Main 1998 Upper Makaske, Splay 5 Castledale 51.0194389°, -116.5630581° 51.0193119°, -116.5604544° 12,343.677 4.091 506.215 5.919 0.772 80.825 183.642 217.532 31.028 1.492 2.691 0.430 Columbia 1998 Upper Makaske, Splay 6 Castledale 51.0242801°, -116.5416118° 51.0252660°, -116.5415525° 9,349.983 3.971 380.873 12.867 1.109 107.937 111.683 111.985 8.680 0.939 1.037 0.016 Columbia 1998 Upper Makaske, Splay 7 Castledale 51.0241205°, -116.5411021° 51.0252870°, -116.5404769° 8,848.425 3.947 432.127 10.426 1.018 72.493 142.220 209.532 13.641 1.135 2.890 0.461 Columbia 1998 Upper Makaske, Splay 8 Castledale 51.0214393°, -116.5345344° 51.0222867°, -116.5351380° 4,493.990 3.653 350.908 13.206 1.121 89.198 51.436 106.455 3.895 0.590 1.193 0.077 Columbia 1998 Upper Makaske, Splay 9 Castledale 51.0205486°, -116.5318258° 51.0220872°, -116.5337195° 7,662.246 3.884 662.105 20.359 1.309 50.331 105.998 227.702 5.206 0.717 4.524 0.656 Columbia 1998 Upper Makaske, Splay 10 Castledale 51.0190409°, -116.5289995° 51.0221856°, -116.5308901° 21,562.256 4.334 925.687 17.457 1.242 123.795 212.828 415.191 12.192 1.086 3.354 0.526 Columbia 1998 Upper Makaske, Splay 11 Castledale 51.0188797°, -116.5290827° 51.0184294°, -116.5240653° 28,892.000 4.461 888.667 16.775 1.225 161.260 239.525 393.294 14.279 1.155 2.439 0.387 Columbia 1998 Upper Makaske, Splay 12 Castledale 51.0161845°, -116.5249252° 51.0176622°, -116.5215754° 24,407.015 4.388 1,010.291 21.900 1.340 135.345 303.544 367.347 13.860 1.142 2.714 0.434 Columbia 1998 Upper Makaske, Splay 13 Castledale 51.0149140°, -116.5238089° 51.0157590°, -116.5228747° 3,186.361 3.503 302.842 22.419 1.351 40.324 88.006 121.112 3.925 0.594 3.003 0.478 Columbia 1998 Upper Castledale Makaske, Splay 14 51.0101501°, -116.5177481° 51.0154155°, -116.5221041° 119,422.067 5.077 2,074.349 56.720 1.754 310.078 286.049 725.071 5.043 0.703 2.338 0.369 Columbia /Main 1998 Upper Castledale Makaske, Splay 15 51.0162456°, -116.5397341° 51.0169371°, -116.5395083° 2,785.882 3.445 234.348 55.082 1.741 45.808 46.907 82.427 0.852 -0.070 1.799 0.255 Columbia /Main 1998 Upper Castledale Makaske, Splay 16 51.0147089°, -116.5388545° 51.0155524°, -116.5379854° 7,732.166 3.888 375.853 50.460 1.703 85.897 102.140 119.228 2.024 0.306 1.388 0.142 Columbia /Main 1998 Upper Castledale Splay 17 51.0102222°, -116.5271459° 51.0074554°, -116.5274544° 33,472.229 4.525 1,489.793 70.224 1.846 193.355 260.122 355.421 3.704 0.569 1.838 0.264 Columbia /Main Upper Splay 18 Castledale 51.0130800°, -116.5432521° 51.0140778°, -116.5419678° 12,306.219 4.090 568.873 15.290 1.184 157.583 150.695 171.304 9.856 0.994 1.087 0.036 Columbia Upper Makaske, Splay 19 Castledale 51.0117527°, -116.5236852° 51.0172472°, -116.5348355° 102,875.922 5.012 3,087.511 9.682 0.986 251.810 266.125 1,329.039 27.486 1.439 5.278 0.722 Columbia 1998 Upper Splay 20 Castledale 51.0154076°, -116.5281956° 51.0167248°, -116.5313095° 13,878.880 4.142 650.853 10.014 1.001 113.226 151.006 273.907 15.080 1.178 2.419 0.384 Columbia Upper Castledale Splay 21 51.0103074°, -116.5332823° 51.0094685°, -116.5314852° 9,251.645 3.966 652.396 62.267 1.794 82.135 139.457 260.288 2.240 0.350 3.169 0.501 Columbia /Main Upper Splay 22 Castledale 51.0213298°, -116.5404673° 51.0208368°, -116.5398464° 3,699.620 3.568 287.144 11.745 1.070 77.904 71.029 84.723 6.047 0.782 1.088 0.036 Columbia Upper Splay 23 Castledale 51.0216849°, -116.5407439° 51.0224305°, -116.5423989° 6,838.156 3.835 429.046 12.655 1.102 85.425 145.789 154.289 11.520 1.061 1.806 0.257 Columbia Upper Splay 24 Castledale 51.0120156°, -116.5415913° 51.0116306°, -116.5423043° 1,884.745 3.275 201.090 13.042 1.115 43.142 58.923 71.471 4.518 0.655 1.657 0.219 Columbia Upper Splay 25 Castledale 51.0134014°, -116.5459842° 51.0129695°, -116.5479880° 9,986.477 3.999 517.464 17.037 1.231 166.683 104.801 164.706 6.151 0.789 0.988 -0.005 Columbia 154

Aspect Parent Maximum Splay Log Ratio Log Log Orthogona Normalized 10 Log Splay Area 10 Perimeter Channel 10 Splay Path Normalized (Path 10 Splay System Reach Source Origin Coordinates Terminus Coordinates Splay Channel l Extent Orthogonal Aspect (meters2) (meters) Width Width Length Orthogonal Length/ Area Width (meters) Extent Ratio (meters) (meters) (meters) Extent Maximum Width) Upper Splay 26 Castledale 51.0141838°, -116.5482154° 51.0149560°, -116.5492446° 7,033.135 3.847 415.945 12.240 1.088 70.957 74.087 142.520 6.053 0.782 2.009 0.303 Columbia Upper Splay 27 Castledale 51.0155425°, -116.5499147° 51.0150584°, -116.5512554° 10,677.574 4.028 492.520 19.567 1.292 137.344 108.677 139.113 5.554 0.745 1.013 0.006 Columbia Upper Splay 28 Castledale 51.0156188°, -116.5490924° 51.0163926°, -116.5502264° 8,657.255 3.937 513.233 21.064 1.324 57.409 94.359 193.722 4.480 0.651 3.374 0.528 Columbia Upper Castledale Splay 29 51.0182741°, -116.5483077° 51.0192075°, -116.5503587° 10,854.812 4.036 701.666 37.619 1.575 62.848 81.012 235.456 2.153 0.333 3.746 0.574 Columbia /Main Upper Splay 30 Castledale 51.0182737°, -116.5533572° 51.0184231°, -116.5563826° 31,810.972 4.503 883.215 10.078 1.003 146.095 151.923 349.360 15.075 1.178 2.391 0.379 Columbia Upper Splay 31 Castledale 51.0042283°, -116.5221833° 51.0048943°, -116.5225252° 12,279.978 4.089 666.315 32.631 1.514 110.990 89.372 132.225 2.739 0.438 1.191 0.076 Columbia Upper Splay 32 Castledale 51.0067072°, -116.5231229° 51.0068637°, -116.5210321° 6,770.348 3.831 491.257 37.330 1.572 47.341 104.364 191.489 2.796 0.446 4.045 0.607 Columbia Upper Splay 33 Castledale 51.0061091°, -116.5210558° 51.0068194°, -116.5225375° 8,471.542 3.928 438.858 35.641 1.552 67.261 61.996 157.844 1.739 0.240 2.347 0.370 Columbia Upper Splay 34 Castledale 51.0056062°, -116.5245423° 51.0060952°, -116.5246408° 1,099.172 3.041 146.711 33.071 1.519 28.492 45.656 58.983 1.381 0.140 2.070 0.316 Columbia Upper Splay 35 Castledale 51.0065522°, -116.5245635° 51.0063283°, -116.5247129° 344.708 2.537 82.685 33.195 1.521 14.061 28.158 28.817 0.848 -0.071 2.049 0.312 Columbia Upper Splay 36 Castledale 51.0071764°, -116.5262844° 51.0079219°, -116.5257558° 4,683.055 3.671 305.901 38.713 1.588 80.883 73.299 91.514 1.893 0.277 1.131 0.054 Columbia Upper Makaske, Splay 37 Castledale 51.0149429°, -116.5426861° 51.0165465°, -116.5428746° 8,949.577 3.952 482.891 10.393 1.017 74.396 180.454 198.015 17.364 1.240 2.662 0.425 Columbia 1998 Upper Splay 38 Castledale 51.0167120°, -116.5447960° 51.0163337°, -116.5447014° 961.284 2.983 124.232 4.595 0.662 27.273 42.137 60.741 9.170 0.962 2.227 0.348 Columbia Upper Castledale Splay 39 51.0225306°, -116.5578486° 51.0230656°, -116.5600197° 10,468.005 4.020 625.773 86.705 1.938 136.704 98.340 206.002 1.134 0.055 1.507 0.178 Columbia /Main Upper Splay 40 Castledale 51.0176201°, -116.5309909° 51.0170657°, -116.5306498° 2,392.452 3.379 241.626 12.819 1.108 50.817 51.960 78.209 4.053 0.608 1.539 0.187 Columbia Upper Castledale Splay 41 51.0103009°, -116.5245651° 51.0111367°, -116.5239144° 2,479.386 3.394 254.265 75.580 1.878 31.201 102.664 105.808 1.358 0.133 3.391 0.530 Columbia /Main Upper Splay 42 Castledale 51.0161875°, -116.5378117° 51.0155533°, -116.5384106° 4,390.870 3.643 276.614 31.262 1.495 67.167 82.940 86.005 2.653 0.424 1.280 0.107 Columbia Upper Splay 43 Castledale 51.0237548°, -116.5590639° 51.0233913°, -116.5592941° 2,634.224 3.421 218.591 17.669 1.247 76.983 41.961 49.230 2.375 0.376 0.639 -0.194 Columbia Upper Splay 44 Main 51.0238239°, -116.5633038° 51.0244455°, -116.5633262° 5,136.127 3.711 459.670 40.201 1.604 106.585 82.533 122.193 2.053 0.312 1.146 0.059 Columbia Upper Splay 45 Main 51.0249220°, -116.5770891° 51.0244688°, -116.5771713° 1,664.382 3.221 180.326 108.768 2.037 46.652 53.104 51.257 0.488 -0.311 1.099 0.041 Columbia Upper Splay 46 Main 51.0382694°, -116.5963944° 51.0429861°, -116.5958333° 60,842.423 4.784 1,726.777 153.244 2.185 219.486 460.993 582.547 3.008 0.478 2.654 0.424 Columbia Upper Splay 47 Main 51.0387699°, -116.5975437° 51.0400839°, -116.5976720° 10,028.169 4.001 484.027 125.003 2.097 133.096 110.364 154.974 0.883 -0.054 1.164 0.066 Columbia Upper Splay 48 Main 51.0365135°, -116.5969603° 51.0374339°, -116.6009339° 33,588.675 4.526 1,078.396 151.199 2.180 161.286 137.959 362.923 0.912 -0.040 2.250 0.352 Columbia Upper Splay 49 Main 51.0407157°, -116.6016100° 51.0431328°, -116.6038534° 29,731.527 4.473 1,517.636 111.258 2.046 186.655 186.603 391.037 1.677 0.225 2.095 0.321 Columbia Upper Splay 50 Main 51.0087069°, -116.5051787° 51.0088347°, -116.5020635° 12,774.194 4.106 641.787 56.073 1.749 90.738 85.375 236.440 1.523 0.183 2.606 0.416 Columbia Upper Splay 51 Main 51.0083001°, -116.5074804° 51.0078877°, -116.5061249° 6,893.473 3.838 420.680 53.168 1.726 149.762 74.922 108.236 1.409 0.149 0.723 -0.141 Columbia Upper Splay 52 Main 51.0094037°, -116.4997247° 51.0097474°, -116.4969445° 6,106.691 3.786 524.103 61.480 1.789 39.655 146.501 232.600 2.383 0.377 5.866 0.768 Columbia 155

Aspect Parent Maximum Splay Log Ratio Log Log Orthogona Normalized 10 Log Splay Area 10 Perimeter Channel 10 Splay Path Normalized (Path 10 Splay System Reach Source Origin Coordinates Terminus Coordinates Splay Channel l Extent Orthogonal Aspect (meters2) (meters) Width Width Length Orthogonal Length/ Area Width (meters) Extent Ratio (meters) (meters) (meters) Extent Maximum Width) Upper Splay 53 Main 51.0084871°, -116.4978397° 51.0086620°, -116.4944048° 13,451.855 4.129 722.832 56.985 1.756 74.171 184.505 287.288 3.238 0.510 3.873 0.588 Columbia Upper Splay 54 Main 51.0047790°, -116.4908022° 51.0081814°, -116.4944368° 49,118.773 4.691 1,442.502 59.168 1.772 131.753 154.291 571.820 2.608 0.416 4.340 0.637 Columbia Upper Splay 55 Main 51.0062073°, -116.4957135° 51.0072318°, -116.5023722° 41,612.881 4.619 1,380.194 56.829 1.755 155.458 214.953 631.940 3.782 0.578 4.065 0.609 Columbia Upper Splay 56 Main 51.0006493°, -116.4820888° 51.0005064°, -116.4804069° 5,540.359 3.744 346.960 54.388 1.736 46.312 103.343 139.171 1.900 0.279 3.005 0.478 Columbia Upper Splay 57 Main 50.9979983°, -116.4794648° 50.9976210°, -116.4749158° 48,153.489 4.683 1,130.022 56.550 1.752 182.170 232.604 357.859 4.113 0.614 1.964 0.293 Columbia Upper Splay 58 Main 50.9988647°, -116.4803868° 50.9989680°, -116.4793310° 4,529.771 3.656 355.571 53.250 1.726 62.339 83.294 126.862 1.564 0.194 2.035 0.309 Columbia Upper Splay 59 Main 50.9935942°, -116.4738209° 50.9958910°, -116.4719728° 85,494.161 4.932 1,481.691 53.839 1.731 371.709 297.776 485.705 5.531 0.743 1.307 0.116 Columbia Upper Splay 60 Main 50.9767965°, -116.4534564° 50.9777484°, -116.4534890° 16,351.135 4.214 560.047 37.838 1.578 109.652 113.142 230.568 2.990 0.476 2.103 0.323 Columbia Upper Splay 61 Main 50.9759280°, -116.4512292° 50.9774285°, -116.4515292° 19,471.807 4.289 774.766 33.246 1.522 167.613 180.819 321.219 5.439 0.735 1.916 0.282 Columbia Upper Splay 62 Main 50.9754201°, -116.4490637° 50.9761500°, -116.4475354° 8,752.899 3.942 601.366 30.881 1.490 83.263 144.741 237.404 4.687 0.671 2.851 0.455 Columbia Upper Splay 63 Main 50.9748478°, -116.4533294° 50.9745987°, -116.4516221° 3,158.891 3.500 316.837 38.030 1.580 35.409 95.741 133.499 2.517 0.401 3.770 0.576 Columbia Upper Splay 64 Main 50.9747324°, -116.4491550° 50.9750024°, -116.4499532° 2,113.626 3.325 196.500 29.749 1.473 34.829 55.993 71.066 1.882 0.275 2.040 0.310 Columbia Upper Splay 65 Main 50.9733009°, -116.4450869° 50.9754309°, -116.4452065° 16,453.583 4.216 861.551 26.859 1.429 101.885 134.307 328.929 5.000 0.699 3.228 0.509 Columbia Upper Splay 66 Main 50.9715370°, -116.4440829° 50.9745892°, -116.4438705° 44,508.385 4.648 1,171.320 31.862 1.503 194.663 237.265 436.189 7.447 0.872 2.241 0.350 Columbia Upper Splay 67 Main 50.9693457°, -116.4436095° 50.9716290°, -116.4460437° 42,256.908 4.626 2,003.017 39.828 1.600 153.092 178.298 603.270 4.477 0.651 3.941 0.596 Columbia Upper Splay 68 Main 50.9706230°, -116.4420856° 50.9723233°, -116.4422661° 17,805.298 4.251 761.724 49.330 1.693 130.825 188.512 251.566 3.821 0.582 1.923 0.284 Columbia Upper Machusic Splay 69 Main 51.0523705°, -116.6261431° 51.0530786°, -116.6247237° 12,132.567 4.084 411.489 104.898 2.021 135.380 114.821 127.832 1.095 0.039 0.944 -0.025 Columbia k, 2000 Upper Splay 70 Main 51.0642422°, -116.6366064° 51.0577287°, -116.6362327° 99,844.267 4.999 2,549.209 85.063 1.930 160.192 386.974 1,109.186 4.549 0.658 6.924 0.840 Columbia Upper Splay 71 Main 51.0607157°, -116.6310897° 51.0614398°, -116.6307842° 3,654.514 3.563 286.478 75.699 1.879 57.059 57.021 85.498 0.753 -0.123 1.498 0.176 Columbia Upper Splay 72 Main 51.0594081°, -116.6300494° 51.0598540°, -116.6290531° 4,014.878 3.604 316.834 73.130 1.864 70.263 87.292 88.083 1.194 0.077 1.254 0.098 Columbia Upper Splay 73 Main 51.0557422°, -116.6280695° 51.0498941°, -116.6222179° 75,707.808 4.879 3,237.972 89.720 1.953 148.400 250.055 1,210.718 2.787 0.445 8.158 0.912 Columbia Upper Splay 74 Main 51.0669962°, -116.6423672° 51.0666658°, -116.6435031° 10,278.229 4.012 478.357 61.753 1.791 146.586 88.828 101.640 1.438 0.158 0.693 -0.159 Columbia Upper Splay 75 Main 51.0661595°, -116.6410450° 51.0660035°, -116.6426713° 4,765.797 3.678 372.418 77.187 1.888 49.585 80.618 160.988 1.044 0.019 3.247 0.511 Columbia Upper Splay 76 Main 51.0357038°, -116.5953288° 51.0343958°, -116.5964837° 18,272.534 4.262 672.207 155.261 2.191 186.564 167.446 183.731 1.078 0.033 0.985 -0.007 Columbia Floodout Tooth, Sandover -21.7942555°, 135.2998021° -21.7911245°, 135.3249305° 4,198,856.631 6.623 9,167.155 134.604 2.129 1,636.439 2,824.883 2,934.927 20.987 1.322 1.793 0.254 Zone 1 1999a Floodout Tooth, Sandover -21.7785228°, 135.3328441° -21.7785566°, 135.3549149° 3,041,026.556 6.483 7,442.292 140.012 2.146 1,635.238 2,455.664 2,467.979 17.539 1.244 1.509 0.179 Zone 2 1999a Splay Tooth, Sandover -21.7878143°, 135.2859388° -21.7781338°, 135.2979953° 1,185,182.158 6.074 4,771.860 118.931 2.075 1,121.863 1,089.742 1,866.399 9.163 0.962 1.664 0.221 Complex 1 1999a 156

Aspect Parent Maximum Splay Log Ratio Log Log Orthogona Normalized 10 Log Splay Area 10 Perimeter Channel 10 Splay Path Normalized (Path 10 Splay System Reach Source Origin Coordinates Terminus Coordinates Splay Channel l Extent Orthogonal Aspect (meters2) (meters) Width Width Length Orthogonal Length/ Area Width (meters) Extent Ratio (meters) (meters) (meters) Extent Maximum Width) Splay Tooth, Sandover -21.7885277°, 135.2553267° -21.7898156°, 135.2754747° 1,411,695.270 6.150 7,614.799 248.239 2.395 1,052.643 1,320.834 2,627.026 5.321 0.726 2.496 0.397 Complex 2 1999a Splay Tooth, Sandover -21.7909396°, 135.2783351° -21.7936861°, 135.2814692° 88,854.119 4.949 1,326.133 109.116 2.038 281.779 317.294 457.967 2.908 0.464 1.625 0.211 Complex 3 1999a Splay Tooth, Sandover -21.7896730°, 135.3010795° -21.7909832°, 135.3043400° 119,490.125 5.077 1,330.571 288.278 2.460 371.050 368.697 416.869 1.279 0.107 1.123 0.051 Complex 4 1999a Splay Tooth, Sandover -21.7905484°, 135.2988274° -21.7923627°, 135.3017012° 43,888.371 4.642 1,016.094 247.217 2.393 181.316 321.006 388.019 1.298 0.113 2.140 0.330 Complex 5 1999a Splay Tooth, Sandover -21.7848268°, 135.3031153° -21.7790173°, 135.3052617° 134,865.537 5.130 1,616.572 247.383 2.393 270.583 495.948 740.889 2.005 0.302 2.738 0.437 Complex 6 1999a Splay Tooth, Sandover -21.7813239°, 135.3221756° -21.7861109°, 135.3338095° 666,097.243 5.824 3,537.424 138.786 2.142 668.912 1,190.085 1,462.793 8.575 0.933 2.187 0.340 Complex 7 1999a Splay Sandover -21.7784291°, 135.3293938° -21.7710056°, 135.3384879° 792,185.285 5.899 4,023.026 72.092 1.858 1,204.075 1,021.412 1,353.172 14.168 1.151 1.124 0.051 Complex 8 Splay Sandover -21.7885731°, 135.2666074° -21.7878743°, 135.2724237° 193,670.709 5.287 1,847.416 265.082 2.423 434.446 354.757 631.249 1.338 0.127 1.453 0.162 Complex 9 Splay Tooth, Sandover -21.7907962°, 135.2809375° -21.7976025°, 135.2985816° 1,012,809.492 6.006 4,546.432 117.381 2.070 766.597 795.816 2,034.351 6.780 0.831 2.654 0.424 Complex 10 1999a Splay Tooth, Sandover -21.8225184°, 135.1345564° -21.8229054°, 135.1533870° 1,542,789.231 6.188 5,465.469 99.744 1.999 1,075.549 1,229.465 2,038.770 12.326 1.091 1.896 0.278 Complex 11 1999a Splay Tooth, Sandover -21.8079449°, 135.1850078° -21.7988266°, 135.1975848° 1,231,459.835 6.090 4,502.561 258.156 2.412 1,131.853 1,033.898 1,762.714 4.005 0.603 1.557 0.192 Complex 12 1999a Splay Sandover -21.7793768°, 135.3260978° -21.7825055°, 135.3328835° 118,245.113 5.073 1,755.787 96.492 1.984 237.479 291.036 809.838 3.016 0.479 3.410 0.533 Complex 13 Splay Tooth, Sandover -21.7896850°, 135.2796962° -21.7809965°, 135.2833680° 218,527.334 5.340 2,441.399 109.936 2.041 305.881 915.221 1,078.935 8.325 0.920 3.527 0.547 Complex 14 1999a Splay Tooth, Sandover -21.7970993°, 135.2311464° -21.7897048°, 135.2457984° 629,969.348 5.799 4,747.792 193.506 2.287 519.881 712.512 2,140.722 3.682 0.566 4.118 0.615 Complex 15 1999a Splay Tooth, Sandover -21.7966190°, 135.2387411° -21.8049122°, 135.2806773° 4,832,618.442 6.684 15,003.438 125.546 2.099 1,919.194 1,980.933 4,556.010 15.779 1.198 2.374 0.375 Complex 16 1999a Splay Tooth, Sandover -21.7898016°, 135.2759420° -21.7854508°, 135.2796753° 119,559.817 5.078 1,621.660 87.279 1.941 243.802 473.833 622.864 5.429 0.735 2.555 0.407 Complex 17 1999a Splay Tooth, Sandover -21.7806253°, 135.3147821° -21.7783188°, 135.3172033° 85,211.591 4.930 1,216.242 168.148 2.226 388.590 273.721 362.102 1.628 0.212 0.932 -0.031 Complex 18 1999a Splay Tooth, 10,428.48 Sandover -21.8224198°, 135.1287490° -21.7893057°, 135.1762397° 7,884,106.130 6.897 26,966.533 160.808 2.206 2,341.178 3,565.447 22.172 1.346 4.454 0.649 Complex 19 1999a 3 Splay Sandover -21.8037468°, 135.2116471° -21.7642467°, 135.2283170° 4,506,129.094 6.654 13,359.090 355.978 2.551 1,656.753 4,156.512 5,157.354 11.676 1.067 3.113 0.493 Complex 20 Splay Tooth, Sandover -21.8017739°, 135.2194471° -21.7965036°, 135.2298210° 319,178.891 5.504 2,716.682 152.172 2.182 381.767 530.521 1,263.469 3.486 0.542 3.310 0.520 Complex 21 1999a Splay Tooth, Sandover -21.7802204°, 135.3212194° -21.7769186°, 135.3275300° 121,261.518 5.084 1,862.645 138.467 2.141 226.047 247.588 815.621 1.788 0.252 3.608 0.557 Complex 22 1999a Splay Sandover -21.8074941°, 135.2004227° -21.8053115°, 135.2074932° 68,894.336 4.838 1,842.049 469.158 2.671 185.705 313.968 786.140 0.669 -0.174 4.233 0.627 Complex 23 Splay Sandover -21.8052609°, 135.2153552° -21.8044942°, 135.2208447° 66,496.943 4.823 1,316.808 294.400 2.469 149.109 220.700 573.503 0.750 -0.125 3.846 0.585 Complex 24 Splay Sandover -21.7869560°, 135.2998517° -21.7851796°, 135.2994673° 37,969.193 4.579 763.569 312.230 2.494 264.323 185.903 217.168 0.595 -0.225 0.822 -0.085 Complex 25 Splay Tooth, Sandover -21.7909838°, 135.2586778° -21.7924803°, 135.2626402° 109,569.419 5.040 1,763.209 303.160 2.482 648.333 203.680 441.512 0.672 -0.173 0.681 -0.167 Complex 26 1999a Splay Tooth, Sandover -21.7912127°, 135.2678469° -21.7923134°, 135.2696175° 28,537.612 4.455 725.117 251.741 2.401 185.181 141.188 229.944 0.561 -0.251 1.242 0.094 Complex 27 1999a Splay Tooth, Sandover -21.7911094°, 135.2701097° -21.7921257°, 135.2723463° 34,729.872 4.541 822.022 127.492 2.105 137.510 148.146 262.103 1.162 0.065 1.906 0.280 Complex 28 1999a 157

Aspect Parent Maximum Splay Log Ratio Log Log Orthogona Normalized 10 Log Splay Area 10 Perimeter Channel 10 Splay Path Normalized (Path 10 Splay System Reach Source Origin Coordinates Terminus Coordinates Splay Channel l Extent Orthogonal Aspect (meters2) (meters) Width Width Length Orthogonal Length/ Area Width (meters) Extent Ratio (meters) (meters) (meters) Extent Maximum Width) Splay Tooth, Sandover -21.7909439°, 135.2742658° -21.7922006°, 135.2764847° 43,745.197 4.641 937.765 68.183 1.834 181.933 188.328 272.766 2.762 0.441 1.499 0.176 Complex 29 1999a Splay Tooth, Sandover -21.7844968°, 135.3046685° -21.7791988°, 135.3074539° 228,596.922 5.359 2,224.455 195.227 2.291 626.308 453.427 718.535 2.323 0.366 1.147 0.060 Complex 30 1999a Splay 1 Saskatchewan 53.9437115°, -102.9934599° 53.9565335°, -102.9900242° 892,749.430 5.951 4,892.786 218.133 2.339 1,021.853 1,024.397 1,732.380 4.696 0.672 1.695 0.229

Splay 2 Saskatchewan 53.9454404°, -102.9817185° 53.9501411°, -102.9650452° 456,806.440 5.660 3,711.382 278.226 2.444 452.360 764.693 1,804.204 2.748 0.439 3.988 0.601

Splay 3 Saskatchewan 53.9526921°, -102.9649948° 53.9564839°, -102.9536336° 388,677.802 5.590 3,227.318 186.059 2.270 675.663 452.307 966.904 2.431 0.386 1.431 0.156

Splay 4 Saskatchewan 53.9511023°, -102.9627844° 53.9522694°, -102.9299984° 1,247,798.359 6.096 7,252.677 183.250 2.263 1,030.060 1,063.553 2,567.365 5.804 0.764 2.492 0.397

Splay 5 Saskatchewan 53.9643677°, -102.9081338° 53.9686614°, -102.8967406° 230,923.059 5.363 2,441.375 150.293 2.177 305.431 427.945 951.089 2.847 0.454 3.114 0.493

Splay 6 Saskatchewan 53.9700644°, -102.8968524° 53.9840452°, -102.8603549° 1,036,802.309 6.016 10,179.509 63.274 1.801 746.479 1,274.867 3,228.466 20.148 1.304 4.325 0.636

Splay 7 Saskatchewan 53.9735368°, -102.8882873° 53.9788236°, -102.8749256° 258,543.096 5.413 2,594.106 65.137 1.814 305.291 448.725 1,137.760 6.889 0.838 3.727 0.571

Splay 8 Saskatchewan 53.9899828°, -102.8687017° 53.9952082°, -102.8659113° 176,571.995 5.247 2,101.230 117.207 2.069 445.525 468.886 645.239 4.000 0.602 1.448 0.161

Splay 9 Saskatchewan 53.9916629°, -102.8632737° 53.9944894°, -102.8564500° 108,503.075 5.035 1,866.607 116.103 2.065 385.700 350.209 610.194 3.016 0.479 1.582 0.199

Perez- Arlucea Splay 10 Saskatchewan and 53.9950817°, -102.7779820° 53.9975004°, -102.7735995° 44,001.953 4.643 1,208.798 83.951 1.924 200.759 381.572 412.189 4.545 0.658 2.053 0.312

Smith, 1999 Perez- Arlucea Splay 11 Saskatchewan and 53.9941920°, -102.7752255° 53.9960553°, -102.7697776° 64,804.829 4.812 1,169.413 88.648 1.948 257.386 255.820 436.393 2.886 0.460 1.695 0.229

Smith, 1999 Perez- Arlucea Splay 12 Saskatchewan and 53.9938673°, -102.7702434° 53.9959025°, -102.7675715° 48,366.413 4.685 930.638 127.681 2.106 193.388 282.104 304.380 2.209 0.344 1.574 0.197

Smith, 1999 Perez- Arlucea Splay 13 Saskatchewan and 53.9944331°, -102.7603840° 53.9984328°, -102.7542741° 206,096.808 5.314 2,551.044 204.286 2.310 535.814 480.563 767.576 2.352 0.372 1.433 0.156

Smith, 1999 Perez- Arlucea Splay 14 Saskatchewan and 53.9997205°, -102.7608581° 54.0000761°, -102.7494522° 88,011.793 4.945 2,030.830 136.519 2.135 175.977 273.272 875.912 2.002 0.301 4.977 0.697

Smith, 1999 Perez- Arlucea Splay 15 Saskatchewan and 54.0015585°, -102.7607346° 54.0002568°, -102.7663029° 74,863.519 4.874 1,721.458 138.094 2.140 213.597 241.588 521.382 1.749 0.243 2.441 0.388

Smith, 1999 Perez- Arlucea Splay 16 Saskatchewan and 54.0035783°, -102.7535002° 54.0104595°, -102.7414002° 343,536.855 5.536 3,486.404 209.843 2.322 557.241 510.839 1,442.595 2.434 0.386 2.589 0.413

Smith, 1999

158

Aspect Parent Maximum Splay Log Ratio Log Log Orthogona Normalized 10 Log Splay Area 10 Perimeter Channel 10 Splay Path Normalized (Path 10 Splay System Reach Source Origin Coordinates Terminus Coordinates Splay Channel l Extent Orthogonal Aspect (meters2) (meters) Width Width Length Orthogonal Length/ Area Width (meters) Extent Ratio (meters) (meters) (meters) Extent Maximum Width) Perez- Arlucea Splay 17 Saskatchewan and 53.9909383°, -102.8528963° 53.9944770°, -102.8558508° 76,422.499 4.883 1,515.626 129.712 2.113 307.388 427.410 498.805 3.295 0.518 1.623 0.210

Smith, 1999 Perez- Arlucea Splay 18 Saskatchewan and 53.9909876°, -102.8475701° 53.9956737°, -102.8217026° 617,441.105 5.791 6,524.681 160.932 2.207 609.846 591.743 1,956.760 3.677 0.565 3.209 0.506

Smith, 1999 Perez- Arlucea Splay 19 Saskatchewan and 53.9901321°, -102.7794640° 54.0030653°, -102.7200566° 2,699,293.478 6.431 13,004.828 249.580 2.397 918.720 1,821.626 5,052.219 7.299 0.863 5.499 0.740

Smith, 1999 Splay 20 Saskatchewan 54.0084638°, -102.7396228° 54.0043101°, -102.7238221° 559,542.653 5.748 4,076.363 213.103 2.329 730.633 754.556 1,425.803 3.541 0.549 1.951 0.290

Splay 21 Saskatchewan 53.8529733°, -103.0862659° 53.8378296°, -103.1001753° 498,048.804 5.697 6,456.944 269.132 2.430 791.394 1,909.497 2,350.196 7.095 0.851 2.970 0.473

Splay 22 Saskatchewan 53.8565947°, -103.0933656° 53.8517414°, -103.1012034° 589,555.021 5.771 3,592.071 325.289 2.512 854.786 883.995 898.561 2.718 0.434 1.051 0.022

Splay 23 Saskatchewan 53.8575643°, -103.0873584° 53.8732139°, -103.0196162° 4,438,324.778 6.647 14,040.963 319.200 2.504 1,289.111 4,901.045 9,461.671 15.354 1.186 7.340 0.866

Splay 24 Saskatchewan 53.8641352°, -103.0891632° 53.8777589°, -103.0181327° 2,533,817.639 6.404 13,952.228 394.160 2.596 812.244 4,571.381 8,129.545 11.598 1.064 10.009 1.000

Splay 25 Saskatchewan 53.9904741°, -102.8307442° 53.9899204°, -102.8322901° 10,831.180 4.035 474.327 119.650 2.078 103.657 122.113 145.915 1.021 0.009 1.408 0.149

Splay 26 Saskatchewan 53.8494706°, -103.0660530° 53.8563121°, -103.0529731° 446,286.512 5.650 3,520.479 426.519 2.630 618.652 1,119.651 1,377.625 2.625 0.419 2.227 0.348

Splay 27 Saskatchewan 54.0049580°, -102.6987415° 54.0002549°, -102.7159325° 336,892.322 5.527 3,773.996 127.704 2.106 559.958 489.657 1,358.380 3.834 0.584 2.426 0.385

Perez- Arlucea Splay 28 Saskatchewan and 53.9946815°, -102.8232918° 53.9934874°, -102.8113957° 125,750.084 5.100 2,382.181 48.096 1.682 275.538 290.599 812.001 6.042 0.781 2.947 0.469

Smith, 1999 Perez- Arlucea Splay 29 Saskatchewan and 53.9955624°, -102.8143838° 53.9960965°, -102.7866342° 192,949.417 5.285 4,200.632 62.063 1.793 244.297 217.114 1,884.070 3.498 0.544 7.712 0.887

Smith, 1999 Splay 30 Saskatchewan 53.9840924°, -102.8789320° 53.9863450°, -102.8739296° 26,628.261 4.425 963.597 173.998 2.241 91.064 101.705 477.285 0.585 -0.233 5.241 0.719

Smith et Splay 31 Saskatchewan 53.8730490°, -103.0873248° 53.8828256°, -103.0368573° 3,842,355.529 6.585 10,695.573 216.873 2.336 1,950.629 2,328.516 4,579.819 10.737 1.031 2.348 0.371 al., 1998 Smith et Splay 32 Saskatchewan 53.8990845°, -103.0455731° 53.8941668°, -103.0206139° 2,995,805.205 6.477 8,423.143 211.986 2.326 1,920.619 1,873.942 2,685.900 8.840 0.946 1.398 0.146 al., 1998 Smith et Splay 33 Saskatchewan 53.9303396°, -103.0041553° 53.9111974°, -103.0206832° 2,684,592.295 6.429 10,048.920 271.804 2.434 1,154.642 2,046.036 4,059.208 7.528 0.877 3.516 0.546 al., 1998 Smith et Splay 34 Saskatchewan 53.9061655°, -103.0415501° 53.9220490°, -103.0147456° 1,417,795.994 6.152 7,315.791 267.916 2.428 796.394 945.781 2,938.675 3.530 0.548 3.690 0.567 al., 1998 Smith et Splay 35 Saskatchewan 53.9353044°, -102.9954449° 53.9476925°, -102.9600420° 1,684,816.389 6.227 6,902.592 336.469 2.527 1,084.255 1,229.899 3,732.453 3.655 0.563 3.442 0.537 al., 1998 Smith et Splay 36 Saskatchewan 53.9130570°, -103.0432620° 53.9545752°, -103.0328979° 4,014,785.223 6.604 17,420.282 217.350 2.337 1,599.869 2,835.505 6,291.769 13.046 1.115 3.933 0.595 al., 1998 Splay 37 Saskatchewan 53.9860958°, -102.8623047° 53.9820731°, -102.8520913° 424,213.889 5.628 3,426.728 51.640 1.713 691.486 950.326 1,017.143 18.403 1.265 1.471 0.168

159

Aspect Parent Maximum Splay Log Ratio Log Log Orthogona Normalized 10 Log Splay Area 10 Perimeter Channel 10 Splay Path Normalized (Path 10 Splay System Reach Source Origin Coordinates Terminus Coordinates Splay Channel l Extent Orthogonal Aspect (meters2) (meters) Width Width Length Orthogonal Length/ Area Width (meters) Extent Ratio (meters) (meters) (meters) Extent Maximum Width) Perez- Arlucea Splay 38 Saskatchewan and 53.9892584°, -102.8450825° 53.9785604°, -102.8309368° 1,256,005.014 6.099 7,453.140 168.501 2.227 1,198.402 1,669.428 1,999.612 9.908 0.996 1.669 0.222

Smith, 1999 Perez- Arlucea Splay 39 Saskatchewan and 53.9891477°, -102.8242336° 53.9854139°, -102.7966984° 1,386,227.740 6.142 7,247.082 118.812 2.075 1,080.004 1,427.919 2,666.278 12.018 1.080 2.469 0.392

Smith, 1999 Perez- Arlucea Splay 40 Saskatchewan and 53.9902169°, -102.8182195° 53.9866139°, -102.8060844° 305,541.616 5.485 3,077.108 112.997 2.053 444.975 734.786 1,140.216 6.503 0.813 2.562 0.409

Smith, 1999 Perez- Arlucea Splay 41 Saskatchewan and 53.9914517°, -102.8126929° 53.9875533°, -102.8012886° 305,158.564 5.485 2,691.664 141.052 2.149 455.697 743.330 971.871 5.270 0.722 2.133 0.329

Smith, 1999 Perez- Arlucea Splay 42 Saskatchewan and 53.9934141°, -102.8047136° 53.9895845°, -102.8021800° 59,850.691 4.777 1,264.192 123.167 2.090 154.298 449.958 560.467 3.653 0.563 3.632 0.560

Smith, 1999 Perez- Arlucea Splay 43 Saskatchewan and 53.9931931°, -102.8012661° 53.9862387°, -102.8007219° 221,162.875 5.345 3,293.431 157.693 2.198 443.416 769.106 789.724 4.877 0.688 1.781 0.251

Smith, 1999

160

Appendix B: Flood-basin, Channel, and Island Measurements

Mapped Area Length Centroid Origin Terminus System Reach Shape Type Notes Source (meters2) (meters) (WGS 84) (WGS 84) (WGS 84) Castledale Flood-basin 51.0167323°, - Columbia Castledale 5078010.725 Reach Polygon 116.5369031° Main Channel Main Flood-basin 51.0206110°, - Columbia 7597637.376 Reach Area Channel Polygon 116.5489309° Channel 1 Name/Number Channel 51.0028912°, - 51.0213886°, - (Makaskie, Columbia Castledale 69430.239 3760.454 follows Makaske Polygon/Linestring 116.5206542° 116.5584656° 1998) (1998) Channel 2a Name/Number Channel 51.0127769°, - 51.0167891°, - (Makaskie, Columbia Castledale 7301.561 706.353 follows Makaske Polygon/Linestring 116.5390090° 116.5447995° 1998) (1998) Channel 2b Name/Number Channel 51.0171893°, - 51.0183312°, - (Makaskie, Columbia Castledale 1788.382 304.984 follows Makaske Polygon/Linestring 116.5435453° 116.5471345° 1998) (1998) Channel 3 Castledale/ Name/Number Channel 51.0086115°, - 51.0217096°, - (Makaskie, Columbia Main 343670.751 4151.720 follows Makaske Polygon/Linestring 116.5149451° 116.5620877° 1998) Channel (1998) Channel 4 Name/Number Channel 51.0114972°, - 51.0184366°, - (Makaskie, Columbia Castledale 23009.927 1001.974 follows Makaske Polygon/Linestring 116.5354923° 116.5412755° 1998) (1998) Channel 5 Name/Number Channel 51.0099908°, - 51.0255460°, - (Makaskie, Columbia Castledale 71033.500 3547.895 follows Makaske Polygon/Linestring 116.5184959° 116.5582643° 1998) (1998) Channel 6 Name/Number Channel 51.0185562°, - 51.0285979°, - (Makaskie, Columbia Castledale 29776.010 2195.299 follows Makaske Polygon/Linestring 116.5291840° 116.5552258° 1998) (1998) Unnamed Channel shown in Channel 1 Channel 51.0176426°, - 51.0201187°, - Makaskie (1998), Columbia Castledale 5314.143 568.003 (Makaskie, Polygon/Linestring 116.5578224° 116.5636757° but no name 1998) given. Unnamed Channel shown in Channel 2 Channel 51.0224425°, - 51.0240183°, - Makaskie (1998), Columbia Castledale 7337.489 451.439 (Makaskie, Polygon/Linestring 116.5560134° 116.5597894° but no name 1998) given. Unnamed Channel shown in Channel 3 Channel 51.0064129°, - 51.0080326°, - Makaskie (1998), Columbia Castledale 6960.168 328.953 (Makaskie, Polygon/Linestring 116.5171322° 116.5209386° but no name 1998) given. 161

Mapped Area Length Centroid Origin Terminus System Reach Shape Type Notes Source (meters2) (meters) (WGS 84) (WGS 84) (WGS 84) Unnamed Channel shown in Channel 4 Channel 51.0079912°, - 51.0090740°, - Makaskie (1998), Columbia Castledale 6120.473 293.764 (Makaskie, Polygon/Linestring 116.5155723° 116.5193451° but no name 1998) given. Channel A Name/Number Channel 51.0137408°, - 51.0176383°, - (Makaskie, Columbia Castledale 17458.650 1248.884 follows Makaske Polygon/Linestring 116.5401655° 116.5552579° 1998) (1998) Channel Baldy Name/Number Channel 51.0043189°, - 51.0105478°, - (Makaskie, Columbia Castledale 53771.283 1395.971 follows Makaske Polygon/Linestring 116.5192286° 116.5348146° 1998) (1998) Channel C Name/Number Channel 51.0091537°, - 51.0246234°, - (Makaskie, Columbia Castledale 36413.014 3662.912 follows Makaske Polygon/Linestring 116.5203369° 116.5577533° 1998) (1998) Downstream portion of Downstream Channel 3. Reach Main Channel 51.0217289°, - 51.0675798°, - Extended Columbia 890877.114 8211.032 beginning Channel Polygon/Linestring 116.5620803° 116.6418892° Reach corresponds to end of Channel 3 reach. Upstream portion of Channel 3. Upstream Main Channel 50.9660197°, - 51.0086086°, - Reach end Extended Columbia 383503.722 8068.285 Channel Polygon/Linestring 116.4398730° 116.5149250° corresponds to Reach beginning of Channel 3 reach. Castledale/ Columbia 51.0190387°, - Columbia Main Island Polygon 17698.467 Island 1 116.5469814° Channel Castledale/ Columbia 51.0207216°, - Columbia Main Island Polygon 7408.660 Island 2 116.5521521° Channel Columbia Main 51.0231462°, - Columbia Island Polygon 30426.705 Island 3 Channel 116.5663525° Columbia Main 51.0236949°, - Columbia Island Polygon 4152.501 Island 4 Channel 116.5705826° Columbia Main 51.0415263°, - Columbia Island Polygon 982.473 Island 5 Channel 116.6065056° Columbia Main 51.0627016°, - Columbia Island Polygon 251.982 Island 6 Channel 116.6343849°

162

Mapped Area Length Centroid Origin Terminus System Reach Shape Type Notes Source (meters2) (meters) (WGS 84) (WGS 84) (WGS 84) Columbia Main 51.0644350°, - Columbia Island Polygon 340.501 Island 7 Channel 116.6355509° Columbia Main 50.9835905°, - Columbia Island Polygon 292.996 Island 8 Channel 116.4608092° Ovens River Flood-basin -36.4305226°, Floodplain Ovens 22160610.576 Polygon 146.4955230° Limits Channel -36.4597445°, -36.4233074°, Deep Creek Ovens 189605.673 7629.428 Polygon/Linestring 146.5349821° 146.4934195° Channel -36.4597286°, -36.4234525°, Pioneer Reach Ovens 356337.364 13088.411 Polygon/Linestring 146.5348912° 146.4933478° Tarrawingee Channel -36.4234522°, -36.4105198°, Ovens 313466.077 5366.139 Reach Polygon/Linestring 146.4933250° 146.4487012° Sandover Flood-basin -21.7952534°, Sandover 104783679.089 Floodout Zone Polygon 135.2388396° Sandover Channel -21.8258608°, -21.7943372°, Sandover 4435929.270 19690.730 Main Channel Polygon/Linestring 135.1203977° 135.2997367° Sandover Old Channel -21.7908161°, -21.7785123°, Sandover 647423.980 4192.040 Channel Polygon/Linestring 135.2953856° 135.3328000° Saskatchewan Saskatche Flood-basin 53.9417091°, - 73177671.057 Flood-basin wan Polygon 102.9403012° Major Island 1 Major Island 1 is Saskatche 53.9555293°, - is not within Major Island 2 Island Polygon 8583.730 not within the wan 102.9266260° the studied studied reach. reach. Saskatche 53.9712766°, - Major Island 3 Island Polygon 116844.581 wan 102.8965705° Saskatche 53.9855409°, - Major Island 4 Island Polygon 74454.409 wan 102.8769376° Saskatche 53.9855524°, - Major Island 5 Island Polygon 558800.315 wan 102.8683809° Saskatche 53.9918151°, - Major Island 6 Island Polygon 158450.498 wan 102.8252530° Saskatche 53.9939233°, - Major Island 7 Island Polygon 217151.835 wan 102.8157919° Saskatche 53.9951752°, - Major Island 8 Island Polygon 36942.623 wan 102.7989885° Saskatche 53.9939047°, - Major Island 9 Island Polygon 216022.620 wan 102.7822994°

163

Mapped Area Length Centroid Origin Terminus System Reach Shape Type Notes Source (meters2) (meters) (WGS 84) (WGS 84) (WGS 84) In the Saskatchewan, the Centre- Channel Polygon Saskatche 53.8605760°, - 54.0069857°, - Angling Channel River Channel (Combined)/Linestri 9233854.206 35953.465 wan 103.0926103° 102.6931465° polygon ng incorporates side channels and islands. Only the reach Saskatche 53.9676895°, - 53.9749529°, - length was Side Channel 1 Channel LineString 1391.251 wan 102.9024050° 102.8878496° explicitly measured. Only the reach Saskatche 53.9828523°, - 53.9871320°, - length was Side Channel 2 Channel LineString 868.952 wan 102.8811507° 102.8730352° explicitly measured. Only the reach Saskatche 53.9825683°, - 53.9896197°, - length was Side Channel 3 Channel LineString 2039.931 wan 102.8779736° 102.8579476° explicitly measured. Only the reach Saskatche 53.9943245°, - 53.9951752°, - length was Side Channel 4 Channel LineString 1430.152 wan 102.8250037° 102.8037993° explicitly measured. Only the reach Saskatche 53.9949126°, - 53.9952450°, - length was Side Channel 5 Channel LineString 742.748 wan 102.8034609° 102.7928729° explicitly measured. Only the reach Saskatche 53.9945875°, - 53.9925179°, - length was Side Channel 6 Channel LineString 1998.777 wan 102.7973539° 102.7683359° explicitly measured.

164

Appendix C: Values from Representative Ancient Grain-Size Distributions

Ferris Formation: Sample 623-012 623-004 623-011 623-002 623-015 (Ferris) Distal Distal Proximal Proximal Channel Size Size Overbank Overbank Overbank Overbank (CAMSIZER) CAMSIZER (Microns) (Percent (Percent (Percent (Percent (Percent (Microns) Volume) Volume) Volume) Volume) Mass) 0.0582 0.0007 0.0003 0.0000 0.0007 26 0 0.0679 0.0018 0.0009 0.0001 0.0018 32 0 0.0791 0.0037 0.0019 0.0002 0.0036 38 0 0.0921 0.0074 0.0043 0.0004 0.0073 45 0 0.1073 0.0154 0.0099 0.0010 0.0150 53 0 0.1250 0.0330 0.0236 0.0027 0.0314 63 0.1 0.1456 0.0707 0.0564 0.0073 0.0650 75 0 0.1697 0.1463 0.1287 0.0189 0.1288 90 0.2 0.1977 0.2836 0.2712 0.0444 0.2376 106 0.3 0.2303 0.5041 0.5158 0.0925 0.4002 125 0.7 0.2683 0.8049 0.8704 0.1663 0.6057 150 1.2 0.3125 1.1254 1.2750 0.2500 0.8058 180 1.4 0.3641 1.3685 1.6138 0.3106 0.9369 212 1.7 0.4242 1.5028 1.8331 0.3339 0.9831 250 1.9 0.4941 1.5850 1.9860 0.3351 0.9812 300 2.5 0.5757 1.6374 2.0934 0.3204 0.9462 355 2.9 0.6707 1.6313 2.1141 0.2855 0.8694 425 3.8 0.7813 1.6346 2.1240 0.2511 0.7923 500 4.1 0.9103 1.7117 2.1742 0.2312 0.7500 600 5.9 1.0604 1.8011 2.2297 0.2154 0.7159 710 6.6 1.2354 1.9403 2.3358 0.2096 0.7031 850 8.6 1.4393 2.1561 2.5219 0.2145 0.7163 1000 10 1.6767 2.4000 2.7223 0.2189 0.7358 1180 10.3 1.9534 2.6292 2.8885 0.2185 0.7506 1400 10.3 2.2757 2.8864 3.0800 0.2245 0.7773 1700 11.3 2.6512 3.1375 3.2561 0.2340 0.8097 2000 7.4 3.0887 3.3634 3.3963 0.2470 0.8468 2360 5.7 3.5983 3.5592 3.4987 0.2643 0.8894 2800 2.5 4.1920 3.7280 3.5692 0.2864 0.9378 3350 0.6 4.8837 3.8504 3.5873 0.3109 0.9882 4000 0 5.6895 3.9428 3.5722 0.3388 1.0431 4750 0 6.6283 3.9918 3.5152 0.3678 1.0986 5600 0 7.7219 4.0073 3.4305 0.3982 1.1557 6700 0 8.9960 3.9973 3.3318 0.4311 1.2173 10.4804 3.9726 3.2318 0.4678 1.2871 165

Sample 623-012 623-004 623-011 623-002 623-015 (Ferris) Distal Distal Proximal Proximal Channel Size Size Overbank Overbank Overbank Overbank (CAMSIZER) CAMSIZER (Microns) (Percent (Percent (Percent (Percent (Percent (Microns) Volume) Volume) Volume) Volume) Mass) 12.2096 3.9389 3.1374 0.5108 1.3720 14.2242 3.8990 3.0474 0.5623 1.4794 16.5712 3.8622 3.0136 0.6253 1.6209 19.3055 3.7970 2.9927 0.7009 1.8082 22.4909 3.6673 2.9555 0.7886 2.0527 26.2019 3.4408 2.8744 0.8854 2.3637 30.5252 3.1038 2.7333 0.9875 2.7474 35.5618 2.6598 2.5247 1.0881 3.2003 41.4295 2.2159 2.2515 1.1793 3.7041 48.2654 1.7719 1.9289 1.2546 4.2268 56.2292 1.3280 1.5822 1.3111 4.7216 65.5070 0.8840 1.2354 1.3532 5.1558 76.3157 0.0000 0.9161 1.3943 5.5081 88.9077 0.0000 0.6421 1.4575 5.7918 103.5775 0.0000 0.4233 1.5751 5.6441 120.6678 0.0000 0.2631 1.7863 5.2862 140.5780 0.0000 0.1591 2.1335 4.7889 163.7733 0.0000 0.0551 2.6526 4.1994 190.7959 0.0000 0.0000 3.3652 3.5587 222.2773 0.0000 0.0000 4.2666 2.8945 258.9530 0.0000 0.0000 5.3215 2.2009 301.6802 0.0000 0.0000 6.4724 1.5072 351.4575 0.0000 0.0000 7.6816 0.8135 409.4479 0.0000 0.0000 8.9178 0.1199 477.0068 0.0000 0.0000 9.3033 0.0000 555.7130 0.0000 0.0000 8.9642 0.0000 647.4056 0.0000 0.0000 7.6847 0.0000 754.2275 0.0000 0.0000 5.5931 0.0000 878.6750 0.0000 0.0000 2.8842 0.0000

166

Fort Union Formation: Sample (Fort Union) 713-003 713-013 713-014 713-006 713-012 Distal Distal Proximal Proximal Channel Size Overbank Overbank Overbank Overbank (Percent (Microns) (Percent (Percent (Percent (Percent Volume) Volume) Volume) Volume) Volume) 0.0582 0.0031 0.0003 0.0002 0.0001 0.0001 0.0679 0.0071 0.0009 0.0005 0.0002 0.0003 0.0791 0.0129 0.0020 0.0011 0.0005 0.0008 0.0921 0.0218 0.0045 0.0025 0.0013 0.0021 0.1073 0.0362 0.0107 0.0061 0.0038 0.0059 0.1250 0.0592 0.0257 0.0157 0.0116 0.0174 0.1456 0.0943 0.0615 0.0401 0.0353 0.0507 0.1697 0.1440 0.1395 0.0975 0.1009 0.1380 0.1977 0.2081 0.2911 0.2164 0.2582 0.3366 0.2303 0.2822 0.5476 0.4291 0.5769 0.7190 0.2683 0.3559 0.9149 0.7459 1.1045 1.3235 0.3125 0.4129 1.3310 1.1095 1.7701 2.0573 0.3641 0.4390 1.6795 1.4048 2.3648 2.6926 0.4242 0.4356 1.9071 1.5764 2.7539 3.0938 0.4941 0.4155 2.0646 1.6727 3.0018 3.3337 0.5757 0.3844 2.1661 1.7099 3.1331 3.4398 0.6707 0.3428 2.1605 1.6492 3.0717 3.3383 0.7813 0.3027 2.1154 1.5605 2.9511 3.1691 0.9103 0.2786 2.0226 1.4632 2.8258 2.9951 1.0604 0.2576 1.9226 1.3701 2.7078 2.8313 1.2354 0.2423 1.8363 1.3052 2.6665 2.7459 1.4393 0.2331 1.7752 1.2780 2.7316 2.7683 1.6767 0.2267 1.7000 1.2491 2.7955 2.7991 1.9534 0.2212 1.6006 1.2049 2.8054 2.7856 2.2757 0.2188 1.5265 1.1876 2.8751 2.8223 2.6512 0.2186 1.4718 1.1864 2.9709 2.8742 3.0887 0.2197 1.4333 1.1976 3.0806 2.9224 3.5983 0.2209 1.4145 1.2214 3.2114 2.9666 4.1920 0.2203 1.4157 1.2547 3.3637 3.0058 4.8837 0.2159 1.4304 1.2856 3.5117 3.0193 5.6895 0.2069 1.4610 1.3124 3.6626 3.0183 6.6283 0.1933 1.5023 1.3241 3.7901 2.9892 7.7219 0.1773 1.5570 1.3210 3.8905 2.9362 8.9960 0.1627 1.6305 1.3119 3.9738 2.8678 10.4804 0.1535 1.7262 1.3137 3.9412 2.7925 12.2096 0.1525 1.8476 1.3541 3.8387 2.7047 14.2242 0.1598 1.9929 1.4661 3.6609 2.6152 16.5712 0.1721 2.1583 1.6879 3.4098 2.5170 167

Sample (Fort Union) 713-003 713-013 713-014 713-006 713-012 Distal Distal Proximal Proximal Channel Size Overbank Overbank Overbank Overbank (Percent (Microns) (Percent (Percent (Percent (Percent Volume) Volume) Volume) Volume) Volume) 19.3055 0.1827 2.3340 2.0504 3.0822 2.3928 22.4909 0.1833 2.5066 2.5710 2.6808 2.2252 26.2019 0.1680 2.6645 3.2448 2.2794 2.0094 30.5252 0.1377 2.8030 4.0453 1.8781 1.7599 35.5618 0.1051 2.9226 4.9274 1.4767 1.5016 41.4295 0.0983 3.0266 5.8385 1.0753 1.2602 48.2654 0.1619 3.1215 6.7407 0.6739 1.0585 56.2292 0.3551 3.2120 7.0671 0.0000 0.9089 65.5070 0.7437 3.3021 6.8576 0.0000 0.8123 76.3157 1.3898 3.3626 6.1016 0.0000 0.7586 88.9077 2.3357 3.3627 4.9213 0.0000 0.7301 103.5775 3.5803 3.2735 3.5442 0.0000 0.7080 120.6678 5.0547 3.0893 2.2315 0.0000 0.6787 140.5780 6.6122 2.8287 1.1858 0.0000 0.6350 163.7733 8.0357 2.5089 0.1400 0.0000 0.5679 190.7959 9.1374 2.1468 0.0000 0.0000 0.4705 222.2773 9.8759 1.7568 0.0000 0.0000 0.3731 258.9530 10.3427 1.3668 0.0000 0.0000 0.2757 301.6802 9.3539 0.9768 0.0000 0.0000 0.1783 351.4575 7.8384 0.5868 0.0000 0.0000 0.0000 409.4479 6.3229 0.0000 0.0000 0.0000 0.0000 477.0068 4.8073 0.0000 0.0000 0.0000 0.0000 555.7130 3.2918 0.0000 0.0000 0.0000 0.0000 647.4056 1.7763 0.0000 0.0000 0.0000 0.0000 754.2275 0.0000 0.0000 0.0000 0.0000 0.0000 878.6750 0.0000 0.0000 0.0000 0.0000 0.0000

168

Willwood Formation: Sample 708-004 708-002 708-001 708-009 708-018 (Willwood) Distal Distal Proximal Proximal Channel Size Overbank Overbank Overbank Overbank (Percent (Microns) (Percent (Percent (Percent (Percent Volume) Volume) Volume) Volume) Volume) 0.0582 0.0088 0.0000 0.0009 0.0000 0.0000 0.0679 0.0193 0.0001 0.0022 0.0001 0.0001 0.0791 0.0329 0.0002 0.0048 0.0004 0.0002 0.0921 0.0512 0.0005 0.0102 0.0010 0.0005 0.1073 0.0770 0.0016 0.0222 0.0031 0.0018 0.1250 0.1129 0.0051 0.0490 0.0100 0.0063 0.1456 0.1609 0.0161 0.1069 0.0322 0.0222 0.1697 0.2206 0.0474 0.2213 0.0963 0.0722 0.1977 0.2893 0.1238 0.4235 0.2551 0.2063 0.2303 0.3609 0.2807 0.7359 0.5825 0.5040 0.2683 0.4258 0.5441 1.1446 1.1274 1.0341 0.3125 0.4718 0.8834 1.5635 1.8076 1.7435 0.3641 0.4898 1.1987 1.8679 2.3922 2.4092 0.4242 0.4827 1.4201 2.0183 2.7336 2.8626 0.4941 0.4614 1.5738 2.0803 2.8968 3.1550 0.5757 0.4311 1.6684 2.0782 2.9113 3.3071 0.6707 0.3932 1.6608 1.9817 2.7216 3.2349 0.7813 0.3583 1.6135 1.8679 2.4737 3.0872 0.9103 0.3435 1.5388 1.7739 2.2485 2.9293 1.0604 0.3328 1.4635 1.6857 2.0460 2.7812 1.2354 0.3294 1.4210 1.6251 1.9249 2.7254 1.4393 0.3348 1.4235 1.6003 1.8979 2.7919 1.6767 0.3466 1.4165 1.5713 1.8727 2.8502 1.9534 0.3631 1.3782 1.5211 1.8108 2.8352 2.2757 0.3854 1.3638 1.4873 1.7946 2.8732 2.6512 0.4123 1.3528 1.4628 1.7930 2.9184 3.0887 0.4423 1.3380 1.4478 1.7991 2.9611 3.5983 0.4736 1.3201 1.4483 1.8196 3.0119 4.1920 0.5036 1.2999 1.4668 1.8596 3.0795 4.8837 0.5285 1.2715 1.4962 1.9083 3.1468 5.6895 0.5453 1.2442 1.5377 1.9791 3.2344 6.6283 0.5508 1.2168 1.5806 2.0684 3.3244 7.7219 0.5437 1.1943 1.6219 2.1865 3.4176 8.9960 0.5245 1.1811 1.6603 2.3473 3.5145 10.4804 0.4962 1.1754 1.6975 2.5597 3.6186 12.2096 0.4630 1.1720 1.7432 2.8298 3.7048 14.2242 0.4305 1.1607 1.8115 3.1516 3.7408 16.5712 0.4040 1.1353 1.9236 3.5178 3.6941 169

Sample 708-004 708-002 708-001 708-009 708-018 (Willwood) Distal Distal Proximal Proximal Channel Size Overbank Overbank Overbank Overbank (Percent (Microns) (Percent (Percent (Percent (Percent Volume) Volume) Volume) Volume) Volume) 19.3055 0.3867 1.0956 2.1008 3.9107 3.5241 22.4909 0.3786 0.7840 2.3580 4.3100 3.2036 26.2019 0.3752 0.7775 2.7002 4.5605 2.7402 30.5252 0.3679 0.8565 3.1213 4.6242 2.1831 35.5618 0.3464 1.0703 3.6032 4.4624 1.5999 41.4295 0.3034 1.4767 4.1235 4.0668 1.0575 48.2654 0.2403 2.1313 4.6619 3.4754 0.6089 56.2292 0.1713 3.0716 4.9372 2.7649 0.2825 65.5070 0.1254 4.3242 4.9212 2.0213 0.0000 76.3157 0.1470 5.8586 4.5883 1.3374 0.0000 88.9077 0.2941 7.5578 3.9854 0.7813 0.0000 103.5775 0.6315 8.3594 3.2071 0.2253 0.0000 120.6678 1.2169 8.0745 2.3650 0.0000 0.0000 140.5780 2.0859 6.8232 1.5774 0.0000 0.0000 163.7733 3.2329 5.0347 0.9511 0.0000 0.0000 190.7959 4.5937 3.3476 0.5389 0.0000 0.0000 222.2773 6.0305 1.7951 0.3312 0.0000 0.0000 258.9530 7.3404 0.4558 0.2810 0.0000 0.0000 301.6802 8.3503 0.0000 0.3084 0.0000 0.0000 351.4575 9.0129 0.0000 0.3368 0.0000 0.0000 409.4479 9.4369 0.0000 0.3147 0.0000 0.0000 477.0068 8.8712 0.0000 0.2201 0.0000 0.0000 555.7130 7.9444 0.0000 0.1254 0.0000 0.0000 647.4056 6.7140 0.0000 0.0000 0.0000 0.0000 754.2275 5.0032 0.0000 0.0000 0.0000 0.0000 878.6750 2.7977 0.0000 0.0000 0.0000 0.0000

170

Appendix D: Summary Statistics for Studied Outcrops

Values for outcrops studied in Section 5.0. Note that estimated average grain-sizes are for selected samples only (i.e. mean grain-sizes may be skewed) and the abundance of samples is not indicative of deposit extents (e.g. proximal overbank samples are heavily oversampled in the Ferris Formation). Micron values are based on the minimum/maximum values for visual ranges. For example, 125 microns corresponds to the boundary between Very Fine Upper and Fine Lower sand, so a sample with a minimum visual grain-size of Fine Lower is listed as 125 microns, and a sample with maximum visual grain-size of Very Fine Upper is also listed as 125 microns. Visual Grain-Size Estimated Visual Grain- Outcrop Sample Number of Visual Grain-Size Formation Maximum Size Mean for Collected Designation Classification Samples Minimum (Microns) (Microns) Samples (Microns) 4000 (Very Fine 1630 (Very Coarse Channel 1 707 (Coarse Upper) Pebble) Upper) Proximal 622 Ferris 10 1 (Clay) 1000 (Coarse Upper) 123.8 (Very Fine Upper) Overbank Distal 3 1 (Clay) 16 (Fine Silt) 2.67 (Clay) Overbank 2000 (Very Coarse Channel 3 354 (Medium Upper) 912.3 (Coarse Upper) Upper) Proximal 623 Ferris 9 1 (Clay) 1000 (Coarse Upper) 188.1 (Fine Upper) Overbank Distal 4 1 (Clay) 32 (Medium Silt) 3 (Clay) Overbank 627 Ribbon 3 125 (Fine Lower) 707 (Coarse Lower) 328 (Medium Lower) (Underlying Channel Ribbon Additional 125 (Very Fine Channel— Proximal 19 1 (Clay) 8 (Very Fine Silt) Upper) Not Fort Union Overbank reported in Figure 5.2 Distal 24 1 (Clay) 16 (Fine Silt) 2.13 (Clay) as a Overbank channel) 171

Visual Grain-Size Estimated Visual Grain- Outcrop Sample Number of Visual Grain-Size Formation Maximum Size Mean for Collected Designation Classification Samples Minimum (Microns) (Microns) Samples (Microns) 627 64000 (Very Coarse Channel 5 125 (Fine Lower) 1009 (Coarse Lower) (Overlying Pebble) Trunk Proximal Fort Union 8 4 (Very Fine Silt) 707 (Coarse Lower) 206 (Fine Upper) Channel— Overbank Reported in Distal 125 (Very Fine 5 1 (Clay) 13.2 (Fine Silt) Figure 5.2) Overbank Lower) 62.5 (Very Fine Channel 3 707 (Coarse Lower) 223 (Fine Upper) Lower) Proximal 707 Willwood 15 1 (Clay) 250 (Fine Upper) 52.5 (Coarse Silt) Overbank Distal 0 Not Exposed Not Exposed Not Exposed Overbank 1410 (Very Coarse 523.3 (Coarse Lower) Lower) Note: One sample was Channel 4 177 (Fine Upper) Note: One sample from a sandbar, skewing was from a sandbar, results 708 Willwood skewing results Proximal 125 (Very Fine 6 1 (Clay) 24.8 (Medium Silt) Overbank Upper) Distal 8 1 (Clay) 4 (Clay) 2 (Clay) Overbank 1 (Clay)—Found in a 354 (Medium Channel 5 small, in-channel mud 148.8 (Fine Lower) Lower) plug 710 Willwood Proximal 354 (Medium 8 1 (Clay) 36.8 (Coarse Silt) Overbank Lower) Distal 10 1 (Clay) 16 (Fine Silt) 2.3 (Clay) Overbank

172

Visual Grain-Size Estimated Visual Grain- Outcrop Sample Number of Visual Grain-Size Formation Maximum Size Mean for Collected Designation Classification Samples Minimum (Microns) (Microns) Samples (Microns) 354 (Medium Channel 4 88 (Very Fine Upper) 169.3 (Fine Lower) Lower) Proximal 125 (Very Fine 713 Fort Union 6 1 (Clay) 10.3 (Fine Silt) Overbank Upper) Distal 4 1 (Clay) 4 (Clay) 2 (Clay) Overbank 500 (Medium Channel 5 88 (Very Fine Upper) 226 (Fine Upper) Upper) Proximal 714 Willwood 8 1 (Clay) 250 (Fine Upper) 47.1 (Coarse Silt) Overbank Distal 4 1 (Clay) 8 (Very Fine Silt) 2.5 (Clay) Overbank 500 (Medium Channel 3 125 (Fine Lower) 228 (Fine Upper) Upper) Proximal 354 (Medium 716 Willwood 8 1 (Clay) 52.8 (Coarse Silt) Overbank Lower) Distal 15 1 (Clay) 16 (Fine Silt) 2.2 (Clay) Overbank

173

Appendix E: Location of Data Files

Data used in the production of this thesis are provided in the file folder “Craig Millard’s Thesis Data” on the “D” Drive partition of the Hajek Lab 1 computer. Of primary interest is the folder entitled “Files Pertaining to the Final Thesis Draft.” Final editions of the .kml file and thesis draft are presented here along with three additional sub-folders. The “Final Delft3D Model Runs” sub- folder contains the final model runs, associated files, and other relevant model runs from Section 6.0, the “Final Draft ArcMap and CSV Files” sub-folder contains the files utilized in analyzing these Delft model runs, while other final draft data are presented in the “Final Data, Figures, Tables, and Spreadsheets” sub-folder. The latter folder contains several sub-folders with one each for the three analytical chapters (Sections 4.0 – 6.0), an additional folder with every figure saved as a .png file, and a folder with the parent Adobe Illustrator file for many figures. Data for tables and figures are found in the corresponding chapter folder. Within the “Chapter 5 Ancient Systems Study” folder are three .doc(x) files that describe the data collected in the field, which are primarily presented in “Combined File Final.xls.” Returning to the parent folder (i.e. “Craig Millard’s Thesis Data”), the two additional sub- folders are entitled “Journal Articles” and “Other Thesis Related Files.” Relevant and cited journal articles are found in the “Journal Articles” folder. The “Other Thesis Related Files” folder contains mostly older, “raw” data used in earlier study phases, but also contains various thesis drafts, all data from the Delft3D Runs, and all Google Earth .kml files within several sub- folders.

174