Controlled flooding on the Colorado River : using GIS methods to assess sandbar development

Transcription

1 University of Iowa Iowa Research Online Theses and Dissertations Spring 2014 Controlled flooding on the Colorado River : using GIS methods to assess sandbar development Clara Thieme University of Iowa Copyright 2014 Clara Thieme This thesis is available at Iowa Research Online: Recommended Citation Thieme, Clara. "Controlled flooding on the Colorado River : using GIS methods to assess sandbar development." MS (Master of Science) thesis, University of Iowa, Follow this and additional works at: Part of the Geology Commons

2 CONTROLLED FLOODING ON THE COLORADO RIVER: USING GIS METHODS TO ASSESS SANDBAR DEVELOPMENT by Clara Thieme A thesis submitted in partial fulfillment of the requirements for the Master of Science degree in Geoscience in the Graduate College of The University of Iowa May 2014 Thesis Supervisor: Associate Professor Frank H. Weirich

3 Graduate College The University of Iowa Iowa City, Iowa CERTIFICATE OF APPROVAL MASTER'S THESIS This is to certify that the Master's thesis of Clara Thieme has been approved by the Examining Committee for the thesis requirement for the Master of Science degree in Geoscience at the May 2014 graduation. Thesis Committee: Frank Weirich, Thesis Supervisor Elmer A. Bettis III Jeffrey A. Dorale

4 ABSTRACT To replenish and restore sandbars and thus preserve aquatic and riparian habitats along the Colorado River, four high-flow controlled floods were conducted as part of the Glen Canyon Dam Adaptive Management Program. While studies of the most recent flood event in November 2012 are not yet available, scientific research has been completed on the March-April 1996, November 2004, and March 2008 artificial floods. Ground based research on pre- and post-dam sediment-transport has yielded insights into the process of sedimentation, the types of sediments conducive to sandbar formation, the significance of antecedent sand supply, and the effect of high-flow discharges from Glen Canyon Dam along seven reaches of the Colorado River where substantial data were collected over the years. On the basis of GIS data collected before and after the 2004 flood and assembled into an overall image in ArcGIS, this study tests the hypothesis that the post-flood sandbar area and volume show substantial increases over the pre-flood measurements. Analyzing extensive reaches along the main stem of the river allows a comprehensive overview of sandbar movement and development that may serve as a predictive tool for future high-flow experiments. ii

5 TABLE OF CONTENTS LIST OF TABLES......iv LIST OF FIGURES.v INTRODUCTION.. 1 BACKGROUND... 5 Rationale for High-flow Experimental Floods Processes of Sandbar Formation Antecedent Sand Supply Effects of 1996, 2004, and 2008 High-Flow Experiments TESTABLE HYPOTHESIS AND EXPECTED RESULTS...27 METHODS Study Sites Aerial Photography Analysis with ArcGIS Software RESULTS Images of Pre- and Post-Flow Sandbars Sandbars Unsuitable for Analysis Sandbar Measurements in Upper Marble Canyon, Lower Marble Canyon, and Eastern Central Canyon Reaches..41 DISCUSSION OF RESULTS...54 CONCLUSION.. 58 REFERENCES iii

6 LIST OF TABLES Table 1. Comparison of discharge and sand supply before 1996, 2004, and 2008 HFEs Summary of changes in sandbar size between February 1996 and October Measurement of sandbar in Figure Measurement of sandbar in Figure Measurement of sandbar in Figure Upper Marble Canyon pre- and post-flow sandbar measurements Lower Marble Canyon pre- and post-flow sandbar measurement Eastern Central Canyon pre- and post-flow sandbar measurements.50 iv

7 LIST OF FIGURES Figure 1. The Glen Canyon Dam Adaptive Management Program study area showing the Colorado River corridor Comparison of natural and controlled average flood discharge levels Continuous-discharge record for the Colorado River at Lees Ferry (River Mile 0) for 1921 through Beach Building Hydrograph Discharge of the Colorado River at Lees Ferry for the 1937, 1996, 2004,and 2008 HFEs Matched views at a sandbar near river mile 22 before and after 2004 HFE Matched views at a sandbar near river mile 30 after 2012 HFE Matched views at a sandbar near river mile 91 after 2012 HFE Perspective and cross section views of a typical eddy sandbar with sand above and below the reference stage of 8,000 cfs Sand on river bank before, during, and after a flood Colorado River from Lake Powell to Lake Mead Average monthly sand loads for Paria River and Little Colorado River Suspended-sand concentration collected during 2004 HFE Suspended-sand concentration collected during 2008 HFE Downstream changes in suspended-sand concentrations measured during 2004 and 2008 HFEs Normalized sandbar volume above reference stage in Marble Canyon and Grand Canyon from 1996 to Six study sites along the Colorado River in Lower Glen, Marble, and Grand Canyons Area chosen for aerial photography before and after 2004 HFE v

8 19. Grid pattern laid over topographic area between upper Marble Canyon and Eastern Grand Canyon Pre-flow sandbar in Colorado River with red lines showing the x axis (perpendicular to the flow of the river) and y axis (parallel to the flow of the river) Post-flow sandbar in Colorado River with red lines showing the x axis (perpendicular to the flow of the river) and y axis (parallel to the flow of the river) Pre- and post-flow sandbar in Upper Marble Canyon Pre- and post-flow sandbar in Lower Marble Canyon Pre- and post-flow sandbar in Eastern Central Canyon Sandbar obscured by shadow Image of sandbar showing shift at random points Image with shift in middle of photo Changes in Upper Marble Canyon sandbar areas Area changes in Upper Marble Canyon in percentages Changes in Lower Marble Canyon sandbar areas Area changes in Lower Marble Canyon in percentages Area changes in Lower Marble Canyon in percentages, excluding LMC Changes in Eastern Grand Canyon sandbar areas Area changes in Eastern Grand Canyon in percentages 53 vi

9 1 INTRODUCTION Since the completion of the Glen Canyon Dam in 1963, the purpose of which was to create a reservoir that would allow for continuous water supply for the increasing population in Upper and Lower Basin states as well as generate hydroelectric power, balancing human needs with maintaining the ecological integrity of the Colorado River has been a significant challenge. There has been a growing concern about strategies to mimic naturally occurring floods to ensure the preservation and restoration of sandbars essential for the stability of the lifecycles of fish species, riparian vegetation, and the campsites for recreational boaters. Prior to the dam s construction, the Colorado River floods would deposit sediments and build up sandbars. However, the scarcity of sand below the dam and water fluctuations resulting from the operation of the hydroelectric plant led to the erosion of sandbars, and thus impacted the rest of the river ecology. Therefore, to facilitate decisions for restoring the natural river ecosystem while maintaining water availability for a variety of human uses, river management techniques were studied (Allan & Castillo, 2007). To obtain the benefits to the river system that natural flooding provides, while still maintaining control over the water level of the river and the distribution and usage of the water itself, the United States Geological Survey (USGS), in conjunction with the United States Bureau of Reclamation, conducted multiple experiments along the Colorado River to identify whether experimental controlled flooding the intentional release of floodlevel discharge down a river channel could feasibly be used as a method for the restoration and management of the riverine environment. A total of four controlled flooding experiments were conducted along the Colorado River. Detailed studies exist about the 1996, 2004, and 2008 controlled floods, which released flood-level discharge down the Colorado River within Grand Canyon National Park (Howard & Dolan, 1981; Topping et al., 2000; Topping et al., 2010 ). The most recent controlled flood took place

10 2 in November 2012, but field measurements pertaining to the effects of this event have not yet been published. During the 1996 event, sediment samples were collected at four sites downstream of Glen Canyon Dam, the farthest of which is located 166 river-miles downstream. Suspended sediment data were collected at each of the four sites, while bed sediment data were collected at one of the four. The suspended sediment data were accumulated only throughout the duration of the flooding event itself, whereas the bed sediment measurements were taken at intervals beginning one day prior to the start of the rising limb through the first day of the receding limb according to the hydrograph (Topping et. al., 2010). During the 2004 event, data-gathering methods increased, adding a fifth site to the previous four and expanding the methods of data gathering for both suspended and bed sediment. Suspended sediment data were gathered at all five of the sites, while bed sediment data were gathered at all but the site at the base of the dam, as the sediment at that location consisted primarily of larger gravels and boulders. In addition, both types of sediment were sampled from one day prior to the start of the rising limb through the first day of the receding limb, rather than only the bed sediments (Topping et. al., 2010). The sixth sample site was added during the 2008 event, and all six sites were sampled for suspended sediment, while all but the site at the base of the dam were sampled for bed sediment, for the same reason as during For the 2008 flooding event, sampling durations matched those of the previous event (Topping et al., 2010). Despite differences in sediment concentration and composition along the length of the river, an increase in suspended silt, clay, and sand content was observed at nearly all of the study sites, the degree of increase depending on the level of discharge at each location; however, the suspended sediment levels began to decrease before each flooding event had completed. This was unexpected and suggested a finite quantity of sediment available to feed the level of suspended sediment within the river during the period of

11 3 artificial flooding. Additionally, the USGS made several observations that held true at each of the sample sites during the steady discharge portion of the controlled floods: the concentration of suspended sand, silt, and clay decreased over time; the median grain size of suspended sand increased; the median grain size of the river bed increased; and the amount of silt and clay present in the river bed remained very low. Higher concentrations of suspended sand were also observed at all of the study sites in 2008, relative to 1996 and 2004 levels. The USGS suggested that more sand was likely present on the river bed within the Grand Canyon region in 1996 than in previous years, which may reflect longterm scouring conditions of the bed sediments caused by dam operations (Topping et al., 2010). Follow-up studies revealed that, approximately six months after each flood, sandbars that had been newly formed or that had been restored by the controlled floods were subject to severe erosion. Sediment deposits were washed away by dam releases as a result of hydroelectric power production, and there was an insufficient post-flood sediment supply volume from the Colorado River tributaries below the Glen Canyon dam to sustain the sandbars. However, compared to the 1996 sandbar volume, the October 2008 sandbar volume was typically larger. The 2008 controlled flood also increased the volume of sandbars in backwater habitats (Grams et al., 2010). In addition to ground-based field measurements taken to analyze the area, volume, and physical characteristics of sediments (both suspended and deposited), aerial photography has been used to assess the outcome of controlled floods. Of particular interest are Geographic Information System (GIS) methods that allow the creation of a visualized image of sandbars pre- and post-flood. Unfortunately, comparable data do not exist for all three controlled flood events. The 2008 floods had no aerial photography taken, and the 1996 data are inaccessible (the USGS has them in a warehouse, but does not wish to handle the data at this time for fear of damaging them; no digital copies exist).

12 4 Therefore, this study focuses on the GIS data available for the 2004 controlled flood. Using ArcGIS, a program that allows the visual analysis of a sequence of about 30 aerial images taken both pre- and post-flood, this study aims at demonstrating and characterizing the development of sandbars along six study sites along the stretch of the river. The goal of the study is to test the hypothesis that there is a significant increase or decrease in the volume and area of sandbars before and after the 2004 flood. The images created may also indicate that sandbars have shifted in shape or position. The GIS generated images are a snapshot in time that may make a substantial contribution to understanding the impact of controlled floods on sandbar development and serve as a predictive tool for subsequent artificial floods.

13 5 BACKGROUND Glen Canyon Dam, which is located just south of the Arizona-Utah border, forms a reservoir that allows the Upper Basin states Colorado, New Mexico, Utah, and Wyoming to store water in wet years and release it during drier periods, thus guaranteeing the availability of water for the three Lower Basin states California, Arizona, and Nevada at all times throughout the year. At the same time, the dam produces electricity that is sold to approximately 240 wholesale customers in these seven states (Melis, 2011). Figure 1 below shows the geographic area that depends on water from the Coloradoo and is subject to the Glen Canyon Dam Adaptive Management Program. Figure 1: The Glen Canyon Dam Adaptive Management Program study area showing the Colorado River corridor (modified from Melis, 2011) When the dam was constructed ( ), little consideration was given to the dam s potential impact on the ecology of the Colorado. No comprehensive survey or description

14 6 of the pre-dam environment existed (Andrews & Pizzi, 2000). Whitewater boaters were the first to realize the changes in the ecology of the canyon. Sandbars along the channel margin where the flow of the main stem separates from the river bank and creates a recirculating cell seemed to be affected by erosion or complete loss of sedimentation (Schmidt & Graf, 1990). Initially, the National Park Service decided to limit the number of river runners in the hope of preventing the erosion of sandbars. In 1983, responding to pressure from conservation and whitewater recreation groups, the Bureau of Reclamation initiated the Glen Canyon Environmental Studies (GCES) (Andrews & Pizzi, 2000). In 1995, the Operation of Glen Canyon Dam: Final Environmental Impact Statement (EIS) was filed, and in 1996, the Secretary of the Interior signed the Record of Decision. The goal of this initiative was to permit recovery and long-term sustainability of downstream resources while limiting hydropower capability and flexibility only to the extent necessary to achieve recovery and long-term sustainability (cited in Melis, 2011, p. 6). However, it became clear that adjusting the operation of the hydroelectric plant would have an insufficient effect on the restoration of sandbars. Fewer and smaller sandbars meant not only smaller camping beaches for visitors (Kaplinksi et al., 2005), but also continued erosion of cultural sites (Hereford et al. 1993) and a decreased habitat for native fish. As a result of the dam, the water quality, temperature, and turbidity changed and affected organisms consumed by fish as food. This change, for example, has led to advantageous conditions for the rainbow trout at the expense of the humpback chuck ( Glen Canyon beach/habitat building test flow, 1996). A detailed discussion of the importance of sandbars for maintaining an ecological balance for wildlife and vegetation as well as the conservation of Native American Indian archaeological sites would be beyond the scope of this study.

15 7 Rationale for High-Flow Experimental Floods After the closure of Glen Canyon Dam, peak discharges that sent sediment downstream decreased significantly. The average peak discharge during periods of flooding dropped from 93,400 cfs to less than 20,000 cfs ( Controlled Flooding, 1996). Figure 2 demonstrates that natural flood discharge levels were much higher than controlled flood discharge levels. Figure 2: Comparison of natural and controlled average flood discharge levels ( Controlled Flooding, 1996) To combat the continuing erosion of sandbars and preserve an ecological balance, the U.S. Bureau of Reclamation authorized High-Flow Experimental floods (HFEs). Studies indicated that restricted flows from the hydroelectric power plant would only be able to minimize sand transport and thus reduce erosion rates. If sandbars were to be replenished, high flows would need to take place with some frequency (Melis, 2011). These HFEs, which have also been called beach-habitat building flows, are defined as a planned

16 8 release from the dam that exceeds the peak capacity of the hydroelectric power plant (~ 940m 3 /s or ~33,000 cfs) by at least 30 percent (U.S. Bureau of the Interior, 1995). Figure 3: Continuous-discharge record for the Colorado River at Lees Ferry (River Mile 0) for 1921 through 2009 (modified from Topping et al, 2003) Figure 3 demonstrates on a larger scale that, after the closure of Glen Canyon Dam, intermittent high floods that occurred on a regular basis were replaced by much lower than regular flow rates that were not conducive to forcefully moving sediment downstream and churning up sand deposits located on the bed of the Colorado River main channel. The first HFE, which was conducted between March 26 and April 7, 1996, involved a 7-day steady release of 45,000 cfs that was preceded and followed by low steady flows of 8,000 cfs for four days each. Scientists coordinated their efforts to document the effects of the flood on physical, biological, cultural, and recreational resources. Improved scientific knowledge led to changes in the timing and duration of the 2004 and 2008 HFEs. The second HFE occurred between November 22 and 24, 2004, and involved a 60- hour release of about 41,700 cfs. Figure 4 shows the discharge from Glen Canyon Dam

17 9 on the Colorado River at Lees Ferry between November 15 and 29, specifically between November 22 and 24, when the 2004 HFE took place. Figure 4: 2004 Beach Building Flow Hydrograph from USGS Lee s Ferry Gage (Friends of Lake Powell, 2004) The third HFE took place between March 6 and 8, 2008, and included a 60-hour release of about 42,800 cfs. The 2004 and 2008 HFE water releases increased from base to peak flows over 30-hour periods, which is a 50% slower rate of rise than for the 1996 event (Melis, 2011). Figure 5 illustrates the discharge of the Colorado River at Lees Ferry for the 1996, 2004, and 2008 HFEs. A comparison with the 1937 discharge record shows the typical discharge record in a pre-dam year.

18 10 Figure 5: Discharge of the Colorado River at Lees Ferry for the 1937, 1996, 2004, and 2008 HFEs (Melis, 2011) All three HFEs were extensively studied by teams of scientists who coordinated their efforts to obtain a comprehensive perspective on the effects of the controlled floods. To understand the results of their analyses, it is important to be aware of various factors that play a role in the formation and restoration of sandbars. Of particular interest are the processes that lead to sandbar formation and the supply of sand before a flood. Processes of Sandbar Formation Sandbars are not only formed along the main channel of the river, but also along eddies, in which the flow of the river is separated from the main channel and creates its own circulation system in pockets along the river s edge. Such eddies are set off from the

19 11 main stem by debris, smaller rocks, or large boulders. When water levels rise during a flood, these eddies are inundated and sediment is deposited along the reattachment bar. These areas create their own flow patterns once the flood recedes, thus contributing to sandbar formation (Tucker, 1993). The sandbar areas often show substantial gain or loss, changes measurable not only in square miles, but also in their perpendicular and horizontal dimensions (called x and y axes later in this study). The images below show the change in a sandbar s dimensions near river mile 22 on the Colorado River (Melis, 2011, p. 81). Figure 6: Matched views at a sandbar near river mile 22 before and after 2004 HFE (photographs by Joseph Hazel and Matt Kaplinsky)

20 12 Figure 6. Continued Some of the sandbars have very irregular shapes, which makes it difficult to measure area changes accurately. The figure below shows a sandbar that has taken on a different shape, and parts of it are submerged under water. Figure 7: Matched views at a sandbar near river mile 30 after 2012 HFE ( Before and After Images, 2012)

21 13 After an HFE, other sandbars clearly show area loss in comparison to their pre-flow stage. Figure 8 shows an example of a sandbar near river mile 91: Figure 8: Matched views at a sandbar near river mile 91 after 2012 HFE ( Before and After Images, 2012) Scientific studies of sandbars typically focus on sandbar area and volume within three elevation zones. Only parts of the sandbars are located above water, whereas others are submerged. These submerged zones may not be important for recreational use, but they create backwater habitat for aquatic species. The elevation at which this distinction is made is termed the reference stage, defined as the elevation of the water surface at a river discharge of 8,000 cfs. About 75 % of the time, even after the closure of the Glen Canyon Dam in 1963, the Colorado River has maintained a flow of 8,000 cfs. At this discharge rate, the sand below the corresponding river stage is usually submerged

22 14 (Topping et al., 2003). Figure 9A shows a perspective view of typical eddy sandbar, the adjacent channel showing patterns of recirculating flow, the reference stage of 8,000 cfs, and a typical backwater. Figure 9B provides a cross section view of a typical eddy sandbar. Figure 9: Perspective and cross section views of a typical eddy sandbar with Sand above and below the reference stage of 8,000 cfs (modified from Grams et al., 2010)

23 15 In the process of sandbar formation, it is also important to consider the size of sediments. All sizes, from the smallest clays and silts to the largest boulders, are relevant. However, the most significant particles are fine sediment. Sand is defined as particles that are finer than 2 millimeters in diameter and coarser than mm; mud is finer than mm and can be either silt or clay. The term fine sediment refers to sediment that consists of sand and finer particles (Melis, 2011). Although sandbars along the Grand Canyon contain different layers, most of them are composed of sand (Schmidt and Graf, 1990). Therefore, sand is the grain size that is of greatest interest. Sand is typically present in eddy sandbars, along the river bank, and in patches on the river bed. During a flood, fine sand on the bottom of the river bed is churned up by the strong current. In addition, sand deposits on sandbars located upstream are scoured. Sandbars will only form if enough fine sand is suspended in the water. To build adequate sandbar beaches, sand needs to be kept entrained so that it will not settle prematurely in the channel. The discharges must also attain a sufficiently high stage so that the parked sand will not be readily eroded when the river returns to its normal flow (Lucchitta & Leopold, 1999). Figure 10 illustrates the process of sedimentation before, during, and after the flood. Figure 10: Sand on river bank before, during, and after a flood (Anderson & Graf, 1996)

24 16 Antecedent Sand Supply The successful deposition of sediments to form or replenish sandbars is highly dependent on the amount of sand supply that is available before a flood. Before the construction of Glen Canyon Dam, the Colorado River transported a large sediment load downstream past Lees Ferry, 24 miles away from the dam, through the Grand Canyon and finally to Lake Mead (see Figure 11). Figure 11: Colorado River from Lake Powell to Lake Mead (Melis, 2011) Nowadays, however, sand is trapped in Lake Powell. The only suppliers of sediment to the Colorado River in Marble Canyon and the Grand Canyon are the Paria River, which flows into the Colorado River at River Mile (RM) 1, the Little Colorado River (near RM 62), Kanab Creek (near RM 143), Havasu Creek (near RM 157), and a number of other small tributaries. Currently, the sand supply of the Colorado River is almost entirely furnished by the Paria River, but it is only approximately 6% of the pre-dam supply. Below the mouth of the Little Colorado River, the cumulative sand supply is only

25 17 about 12% to 15% of the pre-dam supply (Topping et. al., 2000; Wright et. al., 2005). As a result of reduced sand load carried by the tributaries, the total sand supply is approximately 20 % of the pre-dam sand supply (Topping et. al., 2010). Thus, there is a permanent sand deficiency, which makes it almost impossible to maintain sandbars along the stretch of the Colorado. Since the Colorado River flows in an arid environment, natural flooding events that are apt to replenish sandbars are rare. In January 1993, a major flood on the Little Colorado River transported a plentiful amount of sand downstream. Much of the sand was deposited along the banks of the river and on the channel bed, from where it could be redistributed by the manipulation of dam releases (Wiele et. al., 1996). To ensure that the supply of sand is as large as possible before a controlled flood event, it is important to make use of seasonal floods along the two main tributaries, the Paria River and the Little Colorado River. Figure 12 shows that these tributaries carry the highest average sand load in the fall. Therefore, HFEs only in the spring would be less likely than sand-triggered fall HFEs to result in sustainable increases in sandbar sizes (Topping 2010). Figure 12: Average monthly sand loads for Paria River and Little Colorado River (Topping et al., 2010)

26 18 Increasing the antecedent sand budget by timing the release of flood according to tributary sand loads is the most promising strategy for sustainably restoring sandbars. A direct alternative, transporting sand via a slurry pipeline from Lake Powell to the Lees Ferry area, would be extremely costly and thus probably unfeasible (Patten et. al., 2001). In the HFEs conducted in 1996, 2004, and 2008, sand supply and timing varied considerably. While the 1996 and 2008 floods occurred in the spring, the 2004 flood took place in the fall. In the following, results of the three HFEs will be discussed in greater detail. Effects of 1996, 2004, and 2008 High-Flow Experiments As explained above, sediment deposition is closely linked to the available sand supply before the controlled floods. The three floods took place under varying antecedent sand supply conditions. Table 1 provides a comparison of discharge and sand supply pre-flood taken upstream from RM 87. Conditions that are least likely to lead to large sand accumulation are shown in red. Table 1: Comparison of discharge and sand supply before 1996, 2004, and 2008 HFEs (Topping et. al., 2010)

27 19 The measurements presented in the table above indicate that, in all likelihood, the 1996 flood had less of a restoring effect than the 2004 and 2008 floods. However, the 1996 event was still highly effectual in creating and restoring sandbars above the reference stage. Studies show that, in the Grand Canyon, the 1996 flood caused widespread deposition of sand at relatively high elevations on the river banks where it would not be immediately washed away by typical daytime dam releases of ~20,000 cfs. The average increase in volume of high-elevation sand was 164%, the average increase in area was 67%, and the average increase in thickness was 0.64 m. Most of the sand was deposited in eddies, and large amounts of sand were also deposited along the channel-margin levees. The total area of new deposits was estimated to be between 10,500 and 21,000 square meters per kilometer (Schmidt, 1999). Studies suggest that the 1996 controlled flood that led to deposits in the Grand Canyon scoured a large amount of sand from the sandbars located in the upper reaches of the Colorado River in Marble Canyon (Melis, 2011). However, the loss of total sand volume in Marble Canyon was balanced by the volume of sand supplied by the tributaries, mainly the Paria River, so that there was no net decrease in total sand volume at the study sites in Marble Canyon. A sizable portion of the sand likely also came from the river bed. In 1996, more sand was probably available on the bed of the river in the Grand Canyon than there was in 2004 and 2008 (Topping, 2010). The 1996 HFE provided a great deal of insights into the effects of a high flow. Scientists learned that the concentration of suspended sand was greatest on the first day of the HFE and quickly decreased after that. Therefore, the duration of the 2004 HFE was only three and a half days instead of 7 days for the 1996 HFE. It was also timed to follow a naturally occurring flood of the Paria River in the fall of 2004 that had washed a large volume of fine sediment into the main channel. As a result, sandbar deposits in

28 20 Marble Canyon were larger in 2004 than in 1996, but in the Grand Canyon, they were very similar to the deposits seen after the 1996 HFE (Melis, 2011). Since the antecedent sand supply, specifically the amount of suspended sand, is critical for the deposition of sand in sandbars, measurements for suspended sand were taken at five gages in the main stem of the river. Data for the 2004 HFE were collected at Lees Ferry (river mile 0), at the 30-mile gage, the 61-mile gage, the Grand Canyon gage (river mile 88) and Diamond Creek gage (river mile 166). Figure 13 shows the suspended-sand concentration in milligrams per liter during the 2004 HFE. The left vertical gray line is the beginning of the high, steady discharge during the HFE, and the right vertical indicates the end of the high, steady discharge part of the HFE (Melis, 2011). Figure 13: Suspended-sand concentration collected during 2004 HFE (Melis, 2011)

29 21 The graph above shows that suspended-sand concentration increased significantly between the river mile 0 and river mile 30 study sites. Between the river mile 30 and 61 study sites, the concentration increased again, but to a lesser extent. It remained relatively constant between the river mile 61 and 166 study sites. A similar pattern can be seen in measurements taken during the 2008 HFE, although more sand was available during the 2008 HFE. The March 2008 HFE took place after above-average sand inputs from the Paria River in the fall of 2006 and The antecedent sand supply was abnormally large because above-average multiyear sand inputs are relatively rare in an arid environment. Figure 14 shows suspended-sand concentration measurements collected during the 2008 HFE (Melis, 2011). Figure 14: Suspended-sand concentration collected during 2008 HFE (Melis, 2011)

30 22 In addition to the total sand supply and the amount of suspended-sand concentration, the geomorphic organization of the river plays a major role in the deposition of sand. Depending on the width or the gradient of the channel, deposition will vary. In the wide parts of Marble and Grand Canyons, where there are more eddies and the channel gradient is relatively flat, more sand will accumulate and thus re-build sandbars. Figure 15 shows downstream changes in suspended-sand concentration measured during the 2004 and 2008 HFEs. Tan shading indicates reaches of net sand erosion, and blue shading indicates reaches of net sand deposition. Parcel 1 indicates sampling on the first day of the steady high flow, and parcel 2 shows sampling on the second day of the steady high flow (Melis, 2011). Figure 15: Downstream changes in suspended-sand concentrations measured during 2004 and 2008 HFEs

31 23 The HFEs had a significant impact on sandbar development. Figure 16 summarizes the findings pertaining to sandbar changes above the reference stage caused by the three HFEs. Comparing and contrasting sandbar formation in the Marble Canyon (A) and the Grand Canyon (B) between 1996 and 2008 (data for 2004 in the Grand Canyon were not collected), the graph shows sandbar volume that is normalized (by dividing the volume for each indicated date by the volume measured before the 1996 HFE). This initial data point is reflected by the thick blue line. In each plot, the shaded region shows the upper and lower quartiles, the line shows the median value, and the whiskers show the range of data within 1.5 times the distance between the bounds of the upper and lower quartiles. Outliers outside this range are shown by dots. Pre-flood measurements are shown in blue, post-flood measurements in red, 6-month post-flow measurements in green, and measurements between controlled floods in brown (Melis, 2011). The graph demonstrates that major sandbar development occurred after each flood, both in Marble Canyon and in Grand Canyon. However, six months after the floods, erosion had occurred in both study areas. Losses were more extensive in the upper reaches of the Colorado River (Marble Canyon) than in the lower reaches of the river in Grand Canyon.

32 24 Figure 16: Normalized sandbar volume above reference stage in Marble Canyon and Grand Canyon from 1996 to 2008 (Melis, 2011) Altogether, the HFEs have resulted in net increases of sandbars in the downstream parts of Marble Canyon and in most of Grand Canyon. Long-term monitoring in lower Marble Canyon showed that 83% of sandbars were larger in 2008 than in All of the sites in central and western Grand Canyon were larger in October 2008 than in In upper Marble Canyon and in eastern Grand Canyon, though, only 33% of the sites were larger than they were in February 1996 before the first HFE; 66% were even smaller. The collected data point to the scouring of existing sandbars at the upper reaches of the river (Melis, 2011). As can be seen from Table 2, sandbars newly formed by HFEs are subject to erosion. Immediately after the April 2008 HFE, the number of sites that were larger than February 1996 measurements was significantly higher than in October The eastern, central, and western regions of the Colorado River showed larger benefits than the Marble Canyon sites.

33 25 Table 2: Summary of changes in sandbar size between February 1996 and October 2008 (Melis, 2011) Reseach indicates that, between February 1996 and October 2008 (one month before the 1996 HFE and six months after the 2008 HFE), 75% of sandbars at long-term sites in Grand Canyon experienced net increases in volume, despite ongoing sandbar erosion (Melis et. al., 2011). Data reflecting the erosion that is continuously occurring along the Colorado River suggest that, to manage the riparian environment sustainably, it is important to be flexible with the timing of controlled floods. Typically, HFEs should be coordinated with the antecedent sand supply. If the sand input from the tributaries is small, HFEs should be short and have a low discharge magnitude. The goal is to facilitate sand deposition along the river instead of flushing it quickly downstream where it will land in Lake Mead. If the sand load of tributaries is large, HFEs should be of longer duration and occur at higher discharge magnitudes. The strategy of scheduling most HFEs in the fall, when the Paria River typically carries the largest amount of sand as a result of mid-summer through early-fall floods, would be helpful in most years. In the rare case that the Paria River floods between December and April, a spring-time flood would be advisable. Additionally, after intermittent floods, extremely short HFEs could take place at low magnitudes (Melis et. al., 2011).

34 26 The above information about the rationale for high-flow floods, the processes of sandbar formation, the significance of antecedent sand supplies, as well as the results of the three HFEs conducted between 1996 and 2008 forms the background for the hypothesis outlined in the following section.

35 27 TESTABLE HYPOTHESIS AND EXPECTED RESULTS After the HFEs in 1996 and 2004, monitoring of sandbar formation was conducted at a limited number of locations judged to be most likely to have the greatest amount of erosion and deposition. Accurate evaluation of sand resources by direct measurements is not feasible for extended areas due to the sheer length of Glen Canyon Dam Adaptive Management Program study area (380 km). Many areas along the Colorado River are also inaccessible to field researchers, and the cost of making field measurements in vast stretches of the river would be exorbitant. Additionally, there is a high degree of variability in channel geometry and sand storage that is difficult to account for in field measurements. Consequently, conventional ground-based methods have their limitations that can be, at least partially, overcome by aerial photography. Before and after the 2004 HFE, aerial photographs were taken along wide stretches of the river and compiled for GIS analysis (Unfortunately, these data are not available or accessible for the 1996 and 2008 HFEs.). These images provide a snapshot in time of sandbar formations along wide stretches of the Colorado River that would be difficult to analyze with field-based monitoring. Focusing on the 2004 HFE, for which extensive aerial photography is available and accessible, this study tests the hypothesis that existing sandbars have been enlarged in major areas along the banks of the Colorado River. Since the amount of sand available before the 2004 flood was larger than before the 1996 flood, the hypothesis that sandbars will show a positive change is reasonable. The degree of increase will vary from area to area due to the distance from the Paria River, the most significant source of sand, and the Little Colorado River, the second most important tributary, which flows into the Colorado below Marble Canyon. The segments considered in this study are located between river mile 0 and 88, spanning Upper Marble Canyon, Lower Marble Canyon, and Eastern Grand Canyon. The Paria

36 28 River joins the main stem of the Colorado at river mile 1. The high flow will mobilize sand transported by the Paria River and churn up sand from eddy sandbars and the channel bed. As can be seen in Figures 11 and 12, measurements taken during the 2004 HFE indicate that suspended sand is relatively low in Upper Marble Canyon below mile 30 and increases significantly in the Lower Marble Canyon between the river mile 30 and 61study sites. Therefore, it is to be expected that scouring of sand in Upper Marble Canyon and maybe portions of Lower Marble Canyon will take place. Sandbars will probalby experience some enrichment in Lower Marble Canyon, and definitely more than in Upper Marble Canyon. This finding is corroborated by measurements shown in Figure 13, which indicates net sand erosion in Upper Marble Canyon and net sand deposition in Lower Marble Canyon. After the juncture of the Little Colorado River, suspended-sand concentration is the highest, which will probably lead to substantial replenishment of sandbars in Eastern Grand Canyon. Variations may occur due to channel structure, sedimentation, and countless other attributes. In general, however, there should be a significant difference between sandbar area and volume before and after the 2004 HFE.

37 29 METHODS In the assessment of HFEs, field-based methods have produced important insights into the development of sandbars. Before and after the 1996, 2004, and 2008 events, aerial photographs were taken to complement the conventional ground-based methods. Unfortunately, only the data taken before and after the 2004 HFE are accessible to researchers who do not possess a special permit. Therefore, this study will focus on data generated by 2004 aerial photography. Study Sites Before and after the HFEs, the USGS took a series of aerial photographs along the Colorado River that were assembled into GIS. The software ArcGIS by ESRI allows for the interpretation of the images generated by aerial photography, specifically the quantification of the distribution of sandbars, the dimensions of the channel, and other aspects of the riparian environment. The GIS data were collected at several sites along the Colorado River (see Figure 17). They are located just below Glen Canyon Dam at the confluence of the Paria River and the Colorado River at Lees Ferry (River Mile 0) and extend all the way to River Mile 225 at the confluence of Diamond Creek and the Colorado River (River Mile 225). The red circles and labels designate the study sites, and the green labels define the reaches between the study sites. RM is the abbreviation of River Mile.

38 30 Figure 17: Six study sites along the Colorado River in Lower Glen, Marble, and Grand Canyons (Topping, 2010) The six study sites are (1) RM 0 = Colorado River at Lees Ferry, Arizona, gaging station (2) RM 30 = River Mile 30 sediment station (3) RM 61 = former Colorado River above Little Colorado River near Desert View, Arizona gaging station (4) RM 87 = Colorado River near Grand Canyon, Arizona, gaging station (5) RM 166 = former Colorado River above National Canyon near Supai, Arizona, gaging station (6) RM 225 = Colorado River above Diamond Creek near Peach Springs, Arizona, gaging station The seven reaches are (1) LGC = Lower Glen Canyon (2) UMC = Upper Marble Canyon (3) LMC = Lower Marble Canyon (4) EGC = Eastern Grand Canyon

39 31 (5) ECGC = East-Central Grand Canyon (6) WCGC = West-Central Grand Canyon (7) WGC = Western Grand Canyon These study reaches were chosen for three reasons. They were (1) the locations of key sediment-supplying tributaries, (2) locations of already existing or former USGS gaging stations, and (3) locations where the USGS had collected substantial pre- and post-dam sediment-transport data that would provide a context for subsequent studies (Howard, 1947; Topping et al., 2000). Aerial Photography Analysis with ArcGIS Software The GIS data available for the 2004 HFE are the main point of interest in the following section. Aerial photos before and after this HFE were taken in reaches in Upper and Lower Marble Canyon as well as eastern Grand Canyon (see Figure 18). Figure 18: Area chosen for aerial photography before and after 2004 HFE

40 32 GIS allows topographic areas to be segmented into a grid of Fundamental Assessment Units (FAUs) of equal size. These grids can be laid over a map in order to analyze each of the units individually, with a greater image resolution than might be available for a single contiguous image, while at the same time retaining the geographical attributes and positioning of the data. Figure 19 shows this grid pattern, which divides the topographic area between the upper Marble Canyon and eastern Grand Canyon into a grid of uniform squares. Figure 19: Grid pattern laid over topographic area between upper Marble Canyon and Eastern Grand Canyon Due to the large actual extent of the river and the size of the aerial photos generated by USGS, it is necessary to determine a suitable ratio that will allow a sufficiently detailed view of sandbars. In this study, a ratio of 1:1000 was chosen so that it would be possible to

41 33 see the extent of the sandbar development pre- and post-flow and to generate consistent images that could be displayed in a Word or PowerPoint document. The data provided by USGS for the 2004 HFE contain 30 images suitable for analysis. Depending on the size of the sandbar and the visibility conditions, it is necessary to adjust the zoom feature contained in ArcGIS. With these adjustments, it is possible to obtain a visual impression of the pre- and post- flow size of sandbars. The ArcGIS measuring tool allows for relatively accurate measurements of the sandbars before and after the HFE. Since the sandbars have inconsistent shapes, expressing the changes in square meters seems less exact than measuring the horizontal and vertical distances between the points on the sandbar that are the furthest removed from each other. These distances can be expressed on an x/y axis, with x showing the distance in meters between points perpendicular to the flow of the river and y showing the distance in meters between points parallel with the flow of the river. Based on the data that the image contains, ArcGIS automatically calculates the linear distance between these points. This method has the advantage of providing a consistent frame of reference for all analyzed sandbars. The ArcGIS measurement tool was also used to collect a set of measurements for each of the images by tracing around the full perimeter of each sandbar to obtain its area. This combination of measurement methods enabled the comparison of length, width, and area of each sandbar prior to and following the flood event. Figures 20 serves as an example of a clear image of a sandbar generated by ArcGIS and one of the measurement techniques used.

42 34 Figure 20: Pre-flow sandbar in Colorado River with red lines showing the x axis (perpendicular to the flow of the river) and y axis (parallel to the flow of the river) In many cases, though, it is not easy to determine where a sandbar begins or ends because parts of the sandbar that appear to be submerged under the water level may in actuality be above the water. Figure 21 shows the same sandbar as Figure 20, this time after the 2004 HFE, but the edges do not appear equally sharp, especially at the lower end of the sandbar. It is reasonable to assume, however, that the margin of error is no more than 5 %.

43 Figure 21: Post-flow sandbar in Colorado River with red lines showing the x axis (perpendicular to the flow of the river) and y axis (parallel to the flow of the river) 35

44 36 RESULTS The data provided by USGS for the 2004 HFE cover the following reaches of the Colorado River: Upper Marble Canyon, Lower Marble Canyon, and Eastern Grand Canyon. Figures 11 and 17, topographic maps of the Colorado River reaches, provide a geographic reference for these reaches. Within these reaches, several sandbars clearly show significant changes between the 2004 pre- and post-flow periods. Images of Pre- and Post-Flow Sandbars Figure 22 illustrates a sandbar flowing from North to South near the Upper Marble Canyon. The red line on the left indicates the location of the sandbar. Obviously, the images were taken under different weather conditions. The pre-flow image on the left is clearer than the post-flow image on the right. While the left image even shows the waves created by the movement of the water, the corresponding right image basically reveals black space around the sandbar in the middle of the river. The post-flow image on the right seems smaller horizontally, but longer vertically. Figure 22: Pre- and post-flow sandbar in Upper Marble Canyon

45 37 Exact measurements in ArcGIS show that the sandbar covers a smaller area after than before the flood, a loss of 345 square meters. The exact measurements on the x and y axes can be seen in the following chart: Widest x Widest y Widest x Widest y Difference x Difference y Pre-flow Pre-flow Post-flow Post-flow Table 3: Measurement of sandbar in Figure 22 By contrast, a sandbar with Northwest to Southeast direction in the Lower Marble Canyon shows positive growth. At a glance, one can see that the sandbar is elongated, yet it seems to be slightly narrower than in its pre-flow stage. GIS measurements show an expansion by 303 square meters. Figure 23: Pre- and post-flow sandbar in Lower Marble Canyon

46 38 The following measurements in ArcGIS provide details for the change in dimensions on the x and y axes: Widest x Widest y Widest x Widest y Difference x Difference y Pre-flow Pre-flow Post-flow Post-flow Table 4: Measurement of sandbar in Figure 23 The third set of images is taken from the Eastern Central Canyon. The sandbar shown in Figure 24 is found in Southwest to Northwest direction. It shows substantially increased size both horizontally and vertically, a change of 870 square meters. Figure 24: Pre- and post-flow sandbar in Eastern Central Canyon

47 39 Again, a chart with measurements in meters provides exact measurements in on the x and y axes in meters: Widest x Widest y Widest x Widest y Difference x Difference y Pre-flow Pre-flow Post-flow Post-flow Table 5: Measurement of sandbar in Figure 24 The sceenshots of sandbars in the Upper Marble Canyon, Lower Marble Canyon, and Eastern Grand Canyon demonstrate clearly visible pre- and post-flow changes, either positive or negative. Unfortunately, however, not all images are of the same high quality and therefore do not lend themselves to the kind of detailed analysis possible for the sandbars shown above. Sandbars Unsuitable for Analysis While USGS aerial photography provides an extensive overview of wide reaches along the Colorado River, the clarity and therefore the usability of images for detailed analysis varies according to light conditions. Figure 25 is a good example of this phenomenon. The image shows portions of a sandbar, but since a dark shadow falls over the area that would be critical for measurements, the image does not lend itself to analysis by ArcGIS methods.

48 40 Figure 25: Sandbar obscured by shadow In other cases, the image quality seems to shift at random points, due to however the different photos taken are connected. Figure 26 shows this shift. Figure 26: Image of sandbar showing shift at random points

49 41 As can be seen in Figure 27, this shift may occur even in the middle of a single photo, an issue that may be due to various factors in the way in which the photographs are taken. Figure 27: Image with shift in middle of photo Due to the varying quality of images provided by USGS for the 2004 HFE, only about 43 % of images are of such good quality that measurements of sandbars for pre- and postflow conditions can be taken with relatively great accuracy. Sandbar Measurements in Upper Marble Canyon, Lower Marble Canyon, and Eastern Central Canyon Reaches Analysis of pre- and post-flow sizes of sandbars demonstrates that, depending on the location along the Colorado River, sandbar areas show significant differences in size. The following charts illustrate measurements in Upper Marble Canyon, Lower Marble Canyon, and Eastern Central Canyon reaches.