Flow structure and channel morphodynamics of meander bend chute cutoffs: A case study of the Wabash River, USA

Authors

  • Jessica A. Zinger,

    Corresponding author
    1. Department of Geography and Geographic Information Science, University of Illinois at Urbana-Champaign, Champaign, Illinois, USA
    • Corresponding author: J. A. Zinger, Department of Geography and Geographic Information Science, University of Illinois at Urbana-Champaign, 605 East Springfield Avenue, Champaign, IL 61820, USA. (zinger1@illinois.edu)

    Search for more papers by this author
  • Bruce L. Rhoads,

    1. Department of Geography and Geographic Information Science, University of Illinois at Urbana-Champaign, Champaign, Illinois, USA
    2. Department of Geology, University of Illinois at Urbana-Champaign, Champaign, Illinois, USA
    3. Ven Te Chow Hydrosystems Laboratory, University of Illinois at Urbana-Champaign, Champaign, Illinois, USA
    Search for more papers by this author
  • James L. Best,

    1. Department of Geography and Geographic Information Science, University of Illinois at Urbana-Champaign, Champaign, Illinois, USA
    2. Department of Geology, University of Illinois at Urbana-Champaign, Champaign, Illinois, USA
    3. Ven Te Chow Hydrosystems Laboratory, University of Illinois at Urbana-Champaign, Champaign, Illinois, USA
    4. Department of Mechanical Science and Engineering, University of Illinois at Urbana-Champaign, Champaign, Illinois, USA
    Search for more papers by this author
  • Kevin K. Johnson

    1. US Geological Survey, Illinois Water Science Center, Urbana, Illinois, USA
    Search for more papers by this author

Abstract

[1] This paper documents the three-dimensional structure of flow and bed morphology of two developing chute cutoffs on a single meander bend on the lower Wabash River, USA, and relates the flow structure to patterns of morphologic change in the evolving cutoff channels. The upstream end of the cutoff channels is characterized by: (1) a zone of flow velocity reduction/stagnation and bar development in the main channel across from the cutoff entrance, (2) flow separation and bar development along the inner (left) bank of the cutoff channel immediately downstream from the cutoff entrance, and (3) helical motion and outward advection of flow momentum entering the cutoff channel, leading to erosion of the outer (right) bank of the cutoff channel. At the downstream end of the cutoff channels, the major hydrodynamic and morphologic features are: (1) flow stagnation along the bank of the main channel immediately upstream of the cutoff channel mouth, (2) convergence of flows from the cutoff and main channels, (3) helical motion of flow from the cutoff, (4) a zone of reduced velocity along the bank of the main channel immediately downstream from the cutoff channel mouth, and (5) development of a prominent bar complex that penetrates into the main channel and extends from the stagnation zone upstream to downstream of the cutoff mouth. These results provide the basis for a conceptual model of chute-cutoff dynamics in which the upstream and downstream ends of a cutoff channel are treated as a bifurcation and confluence, respectively.

1 Introduction

[2] The complex planform patterns of meandering rivers have long fascinated Earth scientists, engineers, and artists. As freely meandering alluvial channels migrate across their floodplains, individual bends often develop into complicated, elongate forms, such as compound loops or multilobe bends [Brice, 1974; Schumm, 1985; Frothingham and Rhoads, 2003]. Continued evolution of complex bends can eventually lead to the development of meander cutoffs, either through intersection of the migrating limbs, a process known as neck cutoff, or through erosion of a new channel across the floodplain that connects the two limbs, a process referred to as chute cutoff [Fisk, 1947; Allen, 1965; Lewis and Lewin, 1983; van Dijk et al., 2012]. Once a cutoff channel develops, flow through the original meander bend gradually diminishes as sediment plugs the entrance and exit of the bend, eventually forming an oxbow lake [Gagliano and Howard, 1984; Shields and Abt, 1989]. The production of oxbow lakes through meander cutoffs and the filling of these lakes with sediment play important roles in structuring the morphology and three-dimensional alluvial architecture of floodplains [e.g., Shields and Abt, 1989; Erskine et al., 1992; Saucier, 1994; Citterio and Piégay, 2009; Constantine et al., 2010a; Toonen et al., 2012]. These lakes also provide vital habitat for fish [Penczak et al., 2003; Miranda, 2005; Zeug et al., 2005; Zeug and Winemiller, 2008; Glińska-Lewczuk, 2009; Shoup and Wahl, 2009; Miyazono et al., 2010] and floodplain riparian communities [Amoros and Bornette, 1999, 2002; Bornette et al., 2001].

[3] Since Fisk [1947] described the formation of cutoffs on the Mississippi River, a variety of studies have documented and proposed mechanisms for the initiation of cutoffs. The progressive migration and eventual intersection of the limbs of a meander bend that occur during neck cutoff has been described in detail [e.g., Fisk, 1947; Lewis and Lewin, 1983; Gagliano and Howard, 1984; Hooke, 1995, 2004]. Chute channel formation during overbank flows can occur through deepening and widening of swales on point bars [Fisk, 1947; Bridge et al., 1986; van Dijk et al., 2012], headward erosion of gullies during overbank flows [Gay et al., 1998; Zinger et al., 2011; van Dijk et al., 2012], or downstream erosion of an embayment on the upstream limb of a bend [Constantine et al., 2010b]. Cutoffs locally shorten and straighten the river channel and thus have the potential to substantially influence the morphologic evolution of channel planform. Moreover, because meander dynamics are a spatial-convolution function of channel curvature [Güneralp and Rhoads, 2009, 2010], local changes in channel planform (curvature) induced by a cutoff can have a dynamic effect on the evolution of bends upstream and, in some cases, downstream of the cutoff [Camporeale et al., 2008].

[4] Morphologic change within a new cutoff channel is initially rapid and then slows exponentially over time [Fuller et al., 2003; Hooke, 1995]. The processes operative during this period of rapid change are crucial in determining the local pattern of planform change produced by a cutoff, the timescales of bend abandonment, and the morphology, sedimentology, and internal structure of deposits in oxbow lakes. Few systematic field studies have explored processes in a chute cutoff channel during this critical period prior to closure of either end of the bend when the cutoff channel is still actively evolving [Johnson and Paynter, 1967; Hooke, 1995; Gay et al., 1998]. Because most chute cutoffs occur suddenly, usually during floods, such events are difficult to anticipate and measure in the field. Moreover, the occurrence of major chute cutoffs on large rivers is relatively infrequent and intermittent, and may happen only once every few decades along hundreds of kilometers of river length. As a result, knowledge of the interaction between patterns of flow and changes in channel morphology within, and at the upstream and downstream ends of, incipient chute cutoff channels is currently incomplete, especially for large rivers. Thus, the treatment of chute cutoffs in even the most advanced models of meander migration is “quite schematic” [Seminara, 2006] and overlooks important process mechanisms.

[5] The present paper takes a step toward addressing this issue by documenting the flow structure and evolution of channel morphology following initiation of two natural chute cutoffs on a single bend of a large, lowland meandering river, when the cutoff channels are actively evolving and closure of the original meander bend is occurring through sedimentation at both its upstream and downstream ends. Specific aims of the paper are to (1) document the three-dimensional structure of flow within the evolving cutoff channels; (2) relate changes in bed morphology and channel form to the evolving flow structure; and (3) use these field case studies to propose a general conceptual model of chute cutoff hydrodynamics and morphodynamics that can be used to more widely interpret the characteristics and development of these common features of the alluvial landscape.

2 Study Area

[6] Two chute cutoff events on Mackey Bend, an elongated meander loop located on the Wabash River, Illinois-Indiana, USA, just upstream of its confluence with the Ohio River (Figure 1), have provided an unprecedented opportunity to document the morphologic evolution and flow structure of chute cutoffs on a large meandering river. The lower Wabash River flows southwest along the border of Illinois and Indiana (Figure 1) and is the only large meandering river in the contiguous United States that is currently not controlled by dams. Bed material consists mainly of coarse sand and fine gravel [Jackson, 1975, 1976], and the river flows through an alluvial floodplain composed of coarse sand and fine gravel capped by a thin layer of medium-coarse sand and silt, with some local outcrops of bedrock [Shaver, 1979]. At Mackey Bend, bankfull channel width ranges from 300 to 350 m, and bank heights range from 2 to 10 m. The USGS gaging station at Mt Carmel (drainage area = 74,164 km2), located ~ 70 km northeast of Mackey Bend, reports a mean annual discharge for the lower Wabash River of 825 m3 s−1 and a mean annual peak discharge of 4176 m3 s−1 (for data collected between 1928 and 2009).

Figure 1.

Location map of cutoff site showing Mackey Bend, just upstream of the confluence of the Wabash and Ohio Rivers. Values in white on aerial image show cross-section numbers, and C1 and C2 (black) are prefixes for these numbers corresponding to the first and second cutoff, respectively. The orthophotograph was taken 12 August 2011 (Wabash discharge = 247 m3 s−1) through the USDA National Agriculture Imagery Program (available at http://datagateway.nrcs.usda.gov/). Note the presence of Mackey Island just north of the first cutoff channel.

[7] In June 2008, a chute cutoff occurred on Mackey Bend during a large flood on the Wabash River [Zinger et al., 2011] when the Ohio River, which normally produces pronounced backwater effects in Mackey Bend, was at a relatively low stage (Figure 2a). The chute cutoff in June 2008, herein referred to as the first cutoff (labeled C1 in Figure 1), occurred near the upstream end of Mackey Island, which is located on the upstream limb of the bend. During this flood, the mean daily discharge of the Wabash River exceeded 2860 m3 s−1 (1.35 year recurrence interval, 1928–2009) for a period of 12 days (9–21 June 2008) and peaked on 14 June 2008 at 7165 m3 s−1 (12 year recurrence interval, 1928–2009), with overbank flow inundating parts of the floodplain. Over these 12 days, the mean daily discharge of the Ohio River at Cannelton, Indiana, decreased from 7363 m3 s−1 to 1391 m3 s−1, reaching 2042 m3 s−1 on 14 June 2008 (Figure 2a). Thus, the ratio of the discharges between the Wabash and Ohio rivers on 14 June 2008 was 3.51 (Figure 2a). To place this value in context, a discharge ratio of 3.5 was exceeded only six times between October 1975 and October 2010, and the mean discharge ratio over this period was 0.31.

Figure 2.

(a) Discharge hydrograph for the Wabash River at Mt. Carmel, IL (blue line) and the Ohio River at Cannelton, IN (red line) during the study period. The arrow labeled C1 indicates the flood event that cut the first cutoff, and the bracket labeled C2 indicates the period of time over which the second cutoff developed. (b) The flow dominance ratio (equation (1)) since October 1940, calculated using the Wabash discharge at Mt Carmel, IL. The grey line shows Fd using discharge measured on the Ohio River at Evansville, IN (October 1940 to September 1996). The black line shows Fd using discharge measured on the Ohio River at Cannelton, IN (October 1975 to June 2012).

[8] An additional, more striking, comparison of the Wabash and Ohio discharges during the first chute cutoff event is given by the flow dominance ratio (Fd), defined as

display math(1)

[9] The 2 year flood on the Wabash River is Q2year = 3795 m3 s−1. The magnitude of Fd reflects both the discharge of the Wabash River relative to that of the Ohio River, as well as the discharge of the Wabash River relative to a moderate flood event. Because the Wabash and Ohio rivers tend to flood at the same time and the Ohio River discharge is usually much greater than that of the Wabash River, a high value of Fd indicates that the Wabash discharge is large relative to the Ohio discharge when the Wabash River is at a high flood stage. Such circumstances should prevent substantial backwater conditions, with corresponding reductions in flow velocity, from occurring in the Wabash River upstream of the Ohio-Wabash confluence.

[10] The flood in June 2008 was unusual in that a high discharge ratio occurred during a time of high discharge on the Wabash River, producing an exceptionally large value of Fd (Figure 2b). This analysis suggests that the 2008 flood was a rare event in that a large flood occurred on the Wabash River at a time when discharge on the Ohio River was low. This atypical timing of flooding on the two rivers resulted in high-velocity flow over the neck of Mackey bend, eroding a pre-existing swale into a cutoff channel and producing a system of gullies to the east of this channel [Zinger et al., 2011]. Such gullies have been noted previously in studies of cutoff formation on small meandering rivers [e.g., Keller and Swanson, 1979; Gay et al., 1998], in which flow is locally forced overbank by a natural dam; however, this mechanism for producing overbank flow differs from the present case. Flow through the gully system during subsequent overbank events resulted in the development of a second chute cutoff in June 2009, herein referred to as the second cutoff (labeled C2 in Figure 1). Although the first cutoff channel formed rapidly during a single, large flood event, the second cutoff channel was produced by a series of overbank flows with recurrence intervals on the order of 1 to 2 years. Following an initial period of incision, both cutoffs widened, realigned rapidly, and introduced large amounts of sediment into the Wabash and Ohio Rivers [Zinger et al., 2011].

3 Methods

[11] Three-dimensional velocities in the cutoff channels were measured using Teledyne RD Instruments 600 kHz and 1200 kHz acoustic Doppler current profilers (ADCP). All measurements were georeferenced using a Trimble AG132 differential Global Positioning System (dGPS) mounted above the ADCP. The Teledyne RDI ADCPs are four-beam instruments, with the beams set at a 20° angle, yielding a minimum bin size in profiling mode of 0.1 m. The velocity resolution of the ADCP is approximately 0.01 m s−1, with typical accuracies of ± 0.25% of the water and boat velocity. The instruments have a transducer draft of approximately 0.33 m, a blanking distance of approximately 0.25 m from the sensor head, and a zone of side-lobe contamination within about 6% of the total flow depth from the channel bed. No measurements were obtained in these portions of the flow. Between four and six repeat transects across the river were used to collect multiple sets of velocity measurements along georeferenced cross-sections (Figure 1), following the recommendations of Szupiany et al. [2007]. Navigation along the cross-sections was accomplished using Hypack® software linked to the dGPS. Cross-sections were generally aligned orthogonal to the local channel direction as determined by fitting cubic splines to the channel centerline. When possible, the same georeferenced cross-sections were surveyed during each measurement campaign. However, the rapidly evolving channel planform necessitated reorientation of some cross-sections between measurement campaigns in order to maintain an orthogonal orientation with respect to the channel centerline.

[12] ADCP data were collected, reviewed, and exported as ASCII files using Teledyne RDI WinRiverII® software. The data for multiple traverses were then analyzed using the Velocity Mapping Tool (VMT), a suite of MatLab® routines with a graphical user interface [Parsons et al., 2013]. VMT composites and averages ADCP velocity data from repeat transects along cross-sections and provides capabilities for plotting 3-D velocity information. VMT calculates the local depth-averaged velocity; the streamwise (u), cross stream (v), and vertical (w) components of velocity; and primary and secondary flow velocities both in the Rozovskii [1957] and zero net secondary discharge frames of reference. The Rozovskii reference frame rotates each vertical ensemble of velocity measurements such that primary and secondary velocity components are aligned parallel and perpencidular, respectively, to the orientation of the depth-averaged velocity vector and is especially useful for identifying helical motion in strongly converging or diverging flows [Rozovskii, 1957; Rhoads and Kenworthy, 1998]. VMT can also display patterns of backscatter intensity from the ADCP.

[13] All measurements were obtained during high discharges relative to the mean annual discharge (825 m3 s−1) of the Wabash River (Table 1). Flow structure through the first cutoff channel was documented at three flow events with discharges exceeding 2000 m3 s−1, whereas measurement campaigns at the second cutoff channel were conducted four times during discharges exceeding 1500 m3 s−1. Although water levels during the measured events ranged from slightly more than half bankfull to near-bankfull stage, flow stage in Mackey Bend is strongly influenced by the stage (discharge) of the Ohio River (Table 1) and is therefore not a simple function of the Wabash River discharge. The partitioning of discharge through the main channel and the two cutoff channels was measured during all five surveys (Table 1). Water surface elevations, measured using a Leica System 1200 Real Time Kinematic dGPS (RTK-dGPS), were used with depth measurements from the ADCP to compute bed elevations. The topographic survey data were analyzed in Leica GeoOffice® and integrated into ArcMap®. Additionally, repeat surveys of the positions of the cutoff channel banklines were completed using the RTK-dGPS, typically following channel-formative flow events.

Table 1. Discharges (m3 s−1) Through the Wabash River, the Two Cutoffs (From ADCP Measurements), and the Ohio River (From USGS Gage at Cannelton, IN) at Each Survey
Survey DateWabash River (m3 s−1)Cutoff 1 (m3 s−1)Cutoff 1 (% Main Flow)Cutoff 2 (m3 s−1)Cutoff 2 (% Main Flow)Ohio River (m3 s−1)inline image
7 January 2009203680139%--26700.763
19 February 20092625117445%--62580.419
21 May 20093350119036%--33131.011
27 May 2010160958636%45528%40490.397
29 June 2010261878230%118945%27670.946
2 June 2011230964028%117851%52670.438
30 June 2011271861223%143854%21271.278

[14] Calculating bed shear stress from moving-boat ADCP measurements is problematic due to the lack of data collected in the zone of side-lobe contamination near the channel bed and the low signal-to-noise ratio of ADCP measurements in individual bins. The method proposed by Wilcock [1996] to determine local bed shear stress uses a logarithmic “law of the wall” function based on the depth-averaged velocity (U) and a zero-velocity height (z0) defined by a characteristic grain size. This method provides relatively precise estimates of bed shear stress for moving-boat ADCP velocity data in large gravel-bed rivers [Sime et al., 2007]. Although total flow resistance in the Wabash River, which contains abundant sand, is partly influenced by form roughness associated with bedforms, only skin friction is considered in the estimation of bed shear stress because skin friction is most relevant for sediment transport [Einstein, 1950; Einstein and Barbarossa, 1952]. The bed shear stress due to skin friction is given by

display math(2)

where τbs is bed shear stress due to skin friction, U is depth-averaged velocity, ρ is the water density, and Czs is the Chézy coefficient due to skin friction. In equation (2), Czs is defined as [Keulegan, 1938; Kamphuis, 1974; van Rijn, 1984]

display math(3)

where H is the flow depth, κ is the von Kárman coefficient, ks is the roughness height due to individual sediment grains, and D50 is the median grain size of the bed material. A constant value of D50 = 0.703 mm was used herein as derived from particle size analyses of bulk sediment samples obtained at the surface of exposed bars at low flow. The ADCP data provided information on flow depths and depth-averaged velocities. The estimated values of bed shear stress at measured cross-sections were then imported into ArcMap®, and an interpolated surface was produced using kriging for the area of interest.

4 Results

4.1 Flow Discharges

[15] The percentage of the total Wabash River discharge captured by the cutoff channels has changed over time (Table 1). Prior to development of the second cutoff channel, the percentage of discharge captured by the first cutoff from the main channel (i.e., the total discharge on the Wabash River, measured upstream of the two cutoff channels) increased from 39% to 45% between January and February 2009. This increase may be the result of widening of the entrance to the first cutoff channel by approximately 30 m during this period [Zinger et al., 2011]. Although the cutoff channel continued to widen between February and May 2009, the percentage of discharge captured by the first cutoff had decreased to 36% in May 2009. The discharge through the main channel on 21 May 2009 was significantly larger than that of 19 February 2009 (Table 1). Increased flow velocities in the main Wabash channel (Table 2) on the south side of Mackey Island in May 2009 may have led to more flow bypassing the entrance to the cutoff channel compared to February 2009. Additionally, a large percentage of the discharge in May 2009 flowed in the channel north of Mackey Island, bypassing the cutoff channel completely.

Table 2. Cross-Sectional Average Velocities (m s−1) in the Main Channel Just Upstream of the Cutoff Channel Entrance and Within the Cutoff Channel Entrancea
Survey Dateinline imageinline imageinline image
  1. aC1 and C2 denote mean velocity in the first and second cutoff channels, respectively.
7 January 20091.764 (C1)1.7870.987
19 February 20091.201 (C1)1.2570.955
21 May 20091.723 (C1)1.9930.867
27 May 20100.839 (C2)1.1330.740
29 June 20101.815 (C2)1.6221.120
2 June 20110.675 (C2)1.0500.642
30 June 20111.908 (C2)1.6611.149

[16] The incision of the second cutoff channel in June 2009 led to stabilization of the planform of the first cutoff channel [Zinger et al., 2011]. In May 2010, 28% of the total discharge of the Wabash River flowed through the second cutoff channel, with 36% moving through the first cutoff channel (both compared to the total Wabash River discharge). By 30 June 2011, following a ~2.8 year flood in March 2011 (4560 m3 s−1) and a ~14 year flood in May 2011 (7590 m3 s−1) (Figure 2a), the second cutoff had widened substantially such that it captured 54% of the total discharge, while the first cutoff channel conveyed only 23% of the total Wabash discharge.

4.2 Depth-Averaged Velocities and Bed Shear Stress

[17] Three key features of flow can be identified from the patterns of depth-averaged velocity at the entrance and exit of the cutoff channels (Figure 3; note that the data shown for the first cutoff in Figures 3a and 3c predate the development of the second cutoff channel): (1) a reduction in flow velocities in the main Wabash River channel moving from upstream to downstream past the cutoff entrance (labeled “reduced velocity” in Figures 3a and 3b), (2) large-scale flow separation and recirculation along the inner (left) bank at the upstream end of the cutoff channel (labeled “separation zone” in Figures 3a and 3b), and (3) flow stagnation and reduced flow velocities along the left bank of the main channel immediately upstream from the mouth of the cutoff channel (labeled “stagnation zone” in Figures 3c and 3d). These features were present during all of the measured flows.

Figure 3.

Depth-averaged flow velocities (black arrows) at cross-sections measured at the upstream and downstream ends of the two cutoff channels. The skin friction bed shear stress is plotted in the background as filled, colored contours. (a) Cross-sections at the upstream end of the first cutoff channel on 19 February 2009, (b) cross-sections at the upstream end of the second cutoff channel on 2 June 2011, (c) cross-sections at the downstream end of the first cutoff channel on 19 February 2009, and (d) cross-sections at the downstream end of the second cutoff channel on 2 June 2011. The principal features of the flow (stagnation zone, separation zone, zone of reduced velocity) are labeled.

[18] As flow curves into the upstream end of both cutoff channels (Figures 3a and 3b), the locus of high-velocity shifts from the left (inner) side of the cutoff channel toward the right (outer) bank over a streamwise distance of approximately one cutoff channel width. The loss of fluid mass and momentum to the cutoff channel is reflected in the reduction in depth-averaged flow velocities in the main channel, moving from upstream to downstream (right to left in Figures 3a and 3b) past the cutoff channel entrance. The depth-averaged vectors near the inner bank of the first cutoff channel show negative (upstream) streamwise velocities, indicating flow separation and recirculation (Figure 3a). Small and negative velocities in this region in the second cutoff channel (Figure 3b) also provide evidence of separation of flow at this site. The edges of these separation zones are characterized by strong cross-stream gradients in the magnitude of the depth-averaged velocity, which represent the shear layer along the boundary of the separation zones.

[19] At the downstream end of both cutoff channels (Figures 3c and 3d), flow from the cutoff channels converges with flow through the main channel and stagnates near the upstream corner of this junction. Depth-averaged velocities in the stagnation zones tend to be close to zero and in some cases are negative (upstream). Another zone of low velocity is also found along the left bank of the main channel immediately downstream from the mouth of the second cutoff, with the second cutoff showing flow separation at this location (Figure 3d). Overall, the depth-averaged velocity of flow exiting the cutoff channels is greater than that of the main channel.

[20] The patterns of bed shear stress (colored background in Figure 3) closely follow the patterns of depth-averaged velocity. High depth-averaged velocities are associated with high bed shear stress, and the zones of flow separation and stagnation overlap with areas of low and negative (upstream directed) bed shear stress. Where flow enters the cutoff channel (Figures 3a and 3b), bed shear stress decreases in the main channel, moving from upstream to downstream past the cutoff channel entrance. Zones of high bed shear stress curve from the main channel into the entrance of the cutoff channel. Where flow exits the cutoff channel (Figures 3c and 3d), bed shear stresses increase from upstream to downstream within the main channel and zones of high bed shear stress curve into the main channel from the cutoff channel.

4.3 Three-Dimensional Velocities

[21] Two representative examples from the seven sets of velocity measurements (Table 1 and Figures 47) are used to illustrate the structure of three-dimensional flow at the upstream and downstream ends of the cutoff channels: flow through the first cutoff on 19 February 2009 and flow through the second cutoff on 2 June 2011 (see location of cross-sections in Figure 1; additional cross-sectional plots are given in supporting information Figures S1–S18). Patterns of flow in these two examples are consistent with those in the other five data sets, with the velocity measurements in the first cutoff all being completed prior to initiation of the second cutoff. The secondary flow vectors plotted in the cross-sections are in the plane of the cross-section unless denoted as a Rozovskii decomposition [Rozovskii, 1957].

Figure 4.

ADCP measurements at the upstream end of the first cutoff on 19 February 2009 (looking downstream, locations plotted on inset). (a) C1-2, streamwise velocities (color) and transverse secondary vectors (arrows); (b) C1-6, streamwise velocities (colors) and transverse secondary vectors (arrows); and (c) C1-6, primary velocities (colors) and secondary velocities (arrows) in the Rozovskii reference frame.

4.3.1 First Cutoff

[22] Upstream of the cutoff entrance, the core of high velocity is located on the left side of the channel and centered above the thalweg (C1-2, Figure 4a, and supporting information Figures S1 and S2). The velocity components in the plane of the cross-section show that flow is directed toward the left bank, in the direction of the cutoff channel entrance. The velocity structure just downstream of the entrance to the first cutoff channel is characterized by flow separation and bar development on the left side of the channel, similar to that along the inner bank of a sharp meander bend (C1-6, Figures 4b and 4c, and supporting information Figure S3 and S4). The channel bed at C1-6 is highly asymmetrical, with a deep thalweg on the right side of the channel and a prominent sloping bar surface on the left side. Whereas the transverse velocities depict net outward flow at downstream locations (C1-6, Figure 4b, and supporting information Figure S3 and S4), the Rozovskii decomposition shows a large, clockwise-rotating secondary circulation cell centered on the high-velocity core and a smaller, counterclockwise-rotating secondary circulation cell near the right bank (C1-6, Figure 4c, and supporting information Figure S3 and S4). These two cells converge such that fluid descends near the location of maximum channel depth.

[23] The cross-sections through the main channel at the exit of the first cutoff channel show development of a stagnation zone near the upstream junction corner (left side of C1-12, Figure 5 and left side of C1-11, and supporting information Figures S7 and S8). Secondary flow vectors on the left side of the cross-sections in the main channel are oriented toward the right bank, indicating strong penetration of flow from the cutoff channel into flow moving around Mackey Bend (C1-12 and C1-13, Figures 5b–5d and C1-14, in supporting information Figure S7 and S8). A Rozovskii decomposition of the secondary flow vectors in cross-sections C1-12 to C1-14 shows that a cell of clockwise motion develops on the left-hand side of these cross-sections, where flow curves into the original channel from the cutoff channel (Figures 5c and 5d and supporting information Figures S9 and S10). A weak counterclockwise secondary circulation cell is centered on the right side of the main channel, extending from the right bank to a zone of scour near the center of the cross-section. A strong clockwise secondary circulation cell within this zone of low bed elevation and scour may be related to local topographic effects.

Figure 5.

ADCP measurements at the downstream end of the first cutoff on 19 February 2009 (looking downstream, locations plotted on inset). (a) C1-12, streamwise velocities (colors) and transverse secondary vectors (arrows); (b) C1-13, streamwise velocities (colors) and transverse secondary vectors (arrows); (c) C1-12, primary velocities (colors) and secondary velocities (arrows) in Rozovskii reference frame; (d) C1-13, primary velocities (colors) and secondary velocities (arrows) in Rozovskii reference frame. Note different reference vectors for each panel. (e) Depth-averaged ADCP backscatter intensity at the downstream end of the first cutoff.

[24] The mixing interface between the flows from the cutoff channel and Mackey Bend can be discerned by the difference in the depth-averaged ADCP backscatter intensities (Figure 5e), which serves as an indicator of the relative concentration of suspended bed sediment in the water column [Kostaschuk et al., 2005]. Flow from the cutoff channel has a higher value of backscatter intensity compared to that from the main channel bend, which reflects the difference in suspended sediment load due to differing flow velocities between these channels and rapid erosion of the cutoff channel banks. Thus, the mixing interface between the two streams of fluid is located at the strong cross-stream gradient in depth-averaged backscatter intensity, which occurs about halfway across C1-13 (Figure 5d). This interface corresponds with the left edge of the zone of low streamwise velocities between the two cores of high primary velocity that represent the thalwegs of the two converging flows (Figure 5d).

4.3.2 Second Cutoff

[25] Results for the second cutoff primarily focus on analysis of data collected during the survey on 2 June 2011. Flow structure in the second cutoff channel (Figures 6 and 7 and supporting information Figures S11–S18) is similar to that of the first cutoff, but some differences exist because the main channel near the entrance to the second cutoff is wider than that of the first cutoff, and the offtake angle of the second cutoff is larger (98°) than that of the first cutoff (82°). Near the cutoff entrance (C2-2, Figure 6a, and supporting information Figures S11 and S12), the core of high velocity is located toward the left side of the main channel and secondary flow vectors indicate lateral movement of fluid into the cutoff channel. To the right of this zone of flow offtake, the channel bed elevation decreases into the thalweg and then increases toward the inner (right) bank.

Figure 6.

ADCP measurements at the upstream end of the second cutoff on 2 June 2011, viewed looking downstream (locations plotted on inset). (a) C2-2, streamwise velocities (color) and transverse secondary vectors (arrows); (b) C2-6, streamwise velocities (colors) and transverse secondary vectors (arrows); and (c) C2-6, primary velocities (color) and secondary velocities (arrows) in the Rozovskii reference frame. Note that reference secondary velocity vectors are different for each panel.

Figure 7.

ADCP measurements at the downstream end of the second cutoff on 2 June 2011 (looking downstream, locations plotted on inset). (a) C2-12, streamwise velocities (colors) and transverse secondary velocities (arrows) and (b) C2-12, primary velocities (colors) and secondary velocities (arrows) in the Rozovskii reference frame. Note that reference secondary velocity vectors are different for each panel. (c) Depth-averaged ADCP backscatter intensity at the downstream end of the second cutoff.

[26] Immediately downstream of the entrance to the second cutoff channel, the bed topography is asymmetrical, with the thalweg near the right bank (C2-6, Figure 6b, and supporting information Figures S13 and S14). The zone of highest streamwise velocities is located over this thalweg adjacent to the outer (right) bank. A pronounced zone of flow separation, marked by small or negative streamwise velocities, exists along the left bank of the second cutoff channel, similar to the pattern in the first cutoff channel. Cross-stream vectors have a net outward (rightward) orientation, perhaps reflecting the curvature of the flow streamlines and topographic steering of the flow by the bar along the inner bank (Figure 6b and supporting information Figure S13 and S14). Secondary vectors in the Rozovskii frame of reference indicate clockwise circulation of the fluid within this curving flow (C2-6, Figure 6c, and supporting information Figure S13 and S14).

[27] At C2-12 (Figure 7a and supporting information Figures S15–S18), the stagnation zone seen in the depth-averaged velocities of C2-10 (Figure 3d) is absent, but two separate cores of high velocity still exist, and the cross-stream velocity vectors show penetration of the flow from the cutoff into the main channel. A Rozovskii decomposition for C2-12 shows a pattern of fluid motion suggestive of dual secondary flow cells, characterized by clockwise motion on the left side of the channel and weak counterclockwise motion on the right side of the channel (Figure 7b and supporting information Figures S17 and S18). This pattern produces converging secondary flow at the water surface and diverging secondary flow near the bed at these two cross-sections. An additional counterclockwise-rotating secondary flow cell, perhaps topographically induced, is also evident within the zone of deep flow at the right bank (Figure 7b and supporting information Figures S17 and S18).

[28] The depth-averaged backscatter intensity can again be used to indicate the location of the interface between the fluid exiting the cutoff and that originating from the main bend. The absolute values of backscatter intensity differ from those presented for the first cutoff because flow conditions are different and flow from the first cutoff affects the backscatter intensities of fluid within the main channel. Nevertheless, similar patterns of relative backscatter intensity are visible (Figure 7c). A gradient in backscatter intensity extends downstream from the right bank at the cutoff exit to the center of C2-13. This mixing interface corresponds to the left side of the low-velocity zone identified in C2-12 (Figure 7b). A region of relatively low backscatter intensity is also present on the left sides of C2-12 and C2-13, which is related to a zone of reduced velocity that extends along the left bank of the cutoff channel, around the junction corner into the main channel, and along the left bank of the main channel downstream of the cutoff-channel mouth.

4.4 Morphologic Change, Flow Structure, and Backwater Effects

[29] The morphology of the cutoff channels and the flow structure in these channels coevolve over time through feedbacks that are conditioned by the discharge (stage) of the Ohio River immediately downstream, which determines the degree of backwater effects within Mackey Bend. The two cutoff channels both widened rapidly subsequent to their initiation, with this bank erosion occurring in the form of frequent cantilever block failures as large as 4–5 m high and 20–25 m in length. In both cases, the widening can be related to patterns of flow at the upstream end of the cutoff channel at the offtake point from the main Wabash River. The first cutoff widened mainly by rapid retreat of the right (west) bank (Figures 8a, 8b, 9b, and 10) adjacent to the zone of high velocities and bed shear stress (Figure 3a). Sedimentation occurred in the main Wabash channel, south of Mackey Island, and just downstream of the entrance to the first cutoff (Figures 8a, 8b, and 10), where bed shear stress decreases dramatically (Figure 3a). A prominent bar also developed in the zone of flow separation along the left (east) bank in the upstream half of the first cutoff channel. This bar is clearly visible in the bed topography maps from 19 February 2009 (Figure 8a) and 27 May 2010 (Figure 8b) and accounts for the strong asymmetrical cross-section topography at C1-7 (Figure 9b). Much of this bar is exposed at low flow stage (labeled “2” in Figure 10). Before the development of the second cutoff, which greatly reduced the amount of flow through the first cutoff, this bar aggraded and grew by lateral accretion and downstream progradation as the right bank of the cutoff channel retreated (Figures 8a, 8b, and 9b). The erosional development of the first cutoff essentially ceased after the second cutoff formed and began capturing increasing amounts of flow from the Wabash River (Table 1).

Figure 8.

Maps showing cutoff channel bed elevation interpolated from single-beam fathometer data (NAVD88 elevations) and bankline position for (a) the first cutoff on 19 February 2009, (b) the first cutoff on 27 May 2010, (c) the second cutoff on 27 May 2010, and (d) the second cutoff on 2 June 2011.

Figure 9.

The time evolution of the cross-sections plotted in Figure 8; (a) C1-2, (b) C1-7, (c) C1-12, (d) C1-13, (e) C2-2, (f) C2-6, (g) C2-11, and (h) C2-12.

Figure 10.

Orthophotographs of the field site from the USDA National Agriculture Imagery Program from (a) 30 September 2008, (Qwab = 186 m3 s−1, Qohio = 433 m3 s−1), (b) 5 November 2009 (Qwab = 1323 m3 s−1, Qohio = 3738 m3 s−1), and (c) 12 August 2011 (Qwab = 235 m3 s−1, Qohio = 1215 m3 s−1). Bar forms are labeled as follows: (1) stagnation zone in main Wabash River channel across from the cutoff channel entrance, (2) flow separation zone in the cutoff channel, and (3) cutoff-mouth-bar assemblage.

[30] Initial widening at the upstream end of the second cutoff channel has also played a key role in the subsequent morphological development of the second cutoff, which is characterized by a progressive increase in width along much of the total length of the cutoff channel (Figures 8c, 8d, 9f, and 10). Much like the first cutoff, flow curving into the entrance to the second cutoff channel produced high velocities and bed shear stresses near the west bank of this channel (Figure 3b), resulting in high rates of bank retreat. The widening entrance captured increasing amounts of flow from the Wabash River (Figure 11), causing overall enlargement of the entire channel over time (Figures 8c, 8d, and 10). A clear positive feedback exists between the increase in width of the cutoff channel and the proportion of the total Wabash River discharge moving through this channel.

Figure 11.

Relationship between the percentage discharge captured by cutoff 2 from the main Wabash River, the width of cutoff channel 2 and the velocity reduction factor within the Wabash River channel.

[31] Capture of flow by the cutoff has a pronounced effect on the velocity and bed shear stress in the Wabash River as flow moves past the entrance to the second cutoff channel. This effect can be expressed by the velocity reduction factor Ured:

display math(4)

where inline image is the cross-sectionally averaged streamwise velocity in the main channel upstream of the cutoff entrance and inline image is the cross-sectionally averaged streamwise velocity downstream of the cutoff entrance. As the percentage of discharge captured by the cutoff channel increases, Ured increases (Figure 11), indicating a greater reduction in the velocity of flow in the Wabash River as it moves past the cutoff channel entrance. This reduction in velocity promotes bar growth in the upstream limb of the bend (labeled “1” in Figure 10), which reinforces capture of flow by the cutoff and further increases in Ured. Between 27 May 2010 and 2 June 2011, extensive sedimentation occurred along the right bank of the main Wabash channel across from, and downstream of, the entrance to the second cutoff (Figures 8c and 8d), where bed shear stress decreases along the channel (Figure 3b). Based on the planform curvature of the main Wabash channel, some deposition associated with a point bar at this location prior to cutoff is expected, but bar growth in this location clearly has occurred since the time of the second cutoff.

[32] Patterns of depth-averaged velocities and bankline change for the second cutoff channel over four surveys between May 2010 and June 2011 illustrate how backwater effects from the Ohio River strongly influence rates of flow through this channel (Figure 12). On 27 May 2010 (Figure 12a), the cutoff channel is narrow relative to the main Wabash River channel and the backwater effect from the Ohio River results in relatively deep, low-velocity flow at the field site. By 29 June 2010 (Figure 12b), the cutoff channel has widened substantially and the nearly equal discharges of the Wabash and Ohio rivers result in shallower, high-velocity flow through the cutoff channel. The second cutoff channel widened further by 2 June 2011 but increased in width only slightly between 2 June 2011 and 30 June 2011 (Figure 12). Although the Wabash River had similar discharges during the two June 2011 surveys, high discharge on the Ohio River on 2 June 2011 resulted in relatively deep, slow-moving flow through the cutoff channel, while a decrease in the Ohio River discharge resulted in shallower, faster-moving flow on 30 June 2011. Data for these two dates in June 2011 show that the cutoff channel can convey a similar proportion of the total Wabash discharge under vastly different hydraulic conditions due to backwater effects of the Ohio River.

Figure 12.

Depth-averaged velocity vectors at cross-sections at the upstream and downstream ends of the second cutoff channel during its development from May 2010 to July 2011. The total Wabash River discharge, discharge through the second cutoff, percentage of the total Wabash River discharge in the second cutoff, and the Ohio River discharge are given for reference. All discharges are given in m3 s−1. Solid white lines indicate the bankline positions at the time of measurement and black dotted lines indicate locations where the flow depth was too shallow to collect ADCP data. Survey dates are (a) 27 May 2010, (b) 29 June 2010, (c) 2 June 2011, and (d) 30 June 2011. Aerial orthophotographs plotted in the background were collected by the USDA NAIP on the following dates: 23 June 2010 (Figures 12a and 12b) and 12 August 2011 (Figures 12c and 12d).

[33] Despite progressive widening of the second cutoff channel, the thread of high velocity at the cutoff entrance has remained close to the west bank, leading to flow separation along the east bank during the 27 May 2010 and 2 June 2011 surveys (Figures 3 and 12). Bar development occurred in the region of flow separation in both cutoff channels (Figure 8 and labeled “2” in Figure 10), where the bed shear stresses are low (Figures 3a and 3b). The bar within the zone of flow separation in the second cutoff channel grew quickly over time, in association with the rapid retreat of the right (west) bank, infilling part of the enlarging channel (Figure 9f). The growing bar deflected the incoming flow laterally toward the west bank, thereby contributing to the sustained positioning of the thread of high velocity and bed shear stress near this bank (Figure 12). The degree of flow separation at this location may also be influenced by backwater effects. The depth-averaged velocities suggest that when Ucutoff/UWabash < 1 (Table 2), flow separation at the upstream end of the second cutoff channel is large relative to the width of the cutoff channel (Figures 10a and 10c). Flow separation was not observed on 29 June 2010 or 30 June 2011, when this ratio was relatively high (Table 2 and Figures 12b and 12d); however, shallow flow depths prevented collection of ADCP data near the east bank on these dates. Values of Ucutoff/UWabash < 1 tend to occur when high flow stages on the Ohio River produce strong backwater effects at Mackey Bend (Tables 1 and 2), suggesting that these effects tend to promote flow separation. The data also suggest that the size of the separation zone increases with increasing width of the cutoff channel (Figures 12a and 12c). Increases in the size of the flow separation zone presumably are associated with episodes of enhanced bar growth.

[34] The planform of the second cutoff channel in June 2011 has resulted in a shift of the cutoff channel thalweg from the right bank in the upstream portion of the channel to the left bank at the cutoff exit (Figure 12). The kink in the right bank of the cutoff channel creates a zone of low velocity, and possible flow separation, on the right side of the downstream half of the cutoff channel (Figures 12c and 12d). Deposition within this zone has generated a prominent bar along the right bank of the cutoff channel that extends to the mouth of the cutoff channel (Figure 10c).

[35] At the mouths of both cutoff channels, substantial sedimentation has occurred in the main channel of the Wabash River where flow from the two cutoffs rejoins the downstream limb of Mackey Bend and where the thalweg of the river was probably located prior to cutoff (Figures 8, 9c, 9d, 9g, and 9h). Instead of an outer-bank pool, large volumes of sediment have accumulated within the Wabash River at the mouths of the cutoff channels. These bodies of sediment, exposed at low flow (labeled “3” in Figure 10), occupy the left side of the Wabash River channel, such that the deepest portion of the channel is in the center of the river (Figures 8a, 8b, 9c, and 9d) or along the right bank (Figures 8c, 8d, 9g, and 9h). The bar fronts marking the edge of these wedges of sediment (Figures 8, 9c, 9d, 9g, and 9h) correspond to the pattern of decreasing cross-stream bed shear stress, and thus sediment-transport capacity, in the Wabash River at the mouth of the cutoff channel (Figure 3). During periods of limited backwater, the extensive deposits downstream of the mouth of the second cutoff produce extremely rapid and shallow flow. On 29 June 2010, flow attained supercritical conditions at this location as indicated by standing waves on the water surface (Figures 12b and 13).

Figure 13.

Standing waves observed on 29 June 2010 at the downstream end of cutoff 2 (location indicated in Figure 12b), viewed looking: (a) upstream toward the mouth of cutoff 2, (b) northeast toward the left bank of the main channel.

5 Discussion

5.1 A Conceptual Model for Chute Cutoff Channel Dynamics

[36] The results of this study provide insight into the interaction between flow and morphological change in the development of chute cutoffs after an initial channel has formed across the neck of a meander bend, but before the formation of an oxbow lake. The occurrence of two cutoffs on a single bend is somewhat unusual, but because the cutoffs developed sequentially, with the second cutoff forming upstream from the first one, the two cutoffs on Mackey Bend essentially evolved independently of one another. The similarity of these cutoffs to one another, to chute cutoffs on meandering rivers described previously in the literature, and to chute cutoffs readily observable from satellite imagery all suggest that the findings from this study can be extended to other chute cutoffs. Based on these findings, a conceptual model of flow structure and morphodynamics of chute cutoff channels can be proposed. In this model, the processes operative at a chute cutoff channel, prior to plugging of the entrance and exit of the main river channel with sediment, are the same as those found in a bifurcation (upstream end) and confluence (downstream end) with an intervening section of straight or curved channel (Figure 14). Thus, an understanding of channel bifurcations and confluences provides a framework for interpreting and predicting the spatial patterns of flow and morphologic change in developing chute cutoff channels.

[37] The flow structure at the upstream end of a cutoff channel (i.e., the cutoff entrance) is similar to that at flow diversions from straight or curved channels [e.g., Fares and Herbertson, 1993; Neary and Odgaard, 1993; McLelland et al., 1995; Fares, 1995; Ramamurthy et al., 2007] and channel bifurcations [e.g., Bolla Pittaluga et al., 2003; Kleinhans et al., 2008, 2011; Hardy et al., 2011; Thomas et al., 2011]. Flow structure near the entrance of the cutoff channel is characterized by (1) a reduction in velocity in the main channel across from the cutoff entrance, (2) flow separation along the left bank of the cutoff channel immediately downstream of where flow enters this channel, (3) the development of a secondary flow cell within the cutoff channel as flow curves into it from the main channel, and (4) an abrupt shift in the location of the zone of highest velocity from the left to the right bank as flow enters the cutoff channel (Figure 14). This flow structure leads to development of a bar in the main channel where velocities are reduced (Figures 10 (labeled “1”) and 14) and a bar in the separation zone within the cutoff channel (Figures 10 (labeled “2”) and 14). The high flow velocities at the outer bank of the cutoff channel also lead to substantial erosion of the right bank of the cutoff, which promotes rapid widening and realignment of this channel during its initial development (Figures 11 and 12).

Figure 14.

Conceptual model for the morphodynamics of chute channels showing the bifurcation and confluence, the primary features of flow structure, and the associated morphology

[38] The effects of secondary circulation and topographic steering of flow on the spatial patterns of velocity and bed shear stress account for the rapid erosion of the right bank of the cutoff channel immediately downstream from the offtake (Figures 812). Channel widening associated with rapid bank erosion, combined with formation of a bar across the entrance to the meander bend, increases the proportion of flow captured by the cutoff channel (Table 1 and Figures 812), which in turn promotes enlargement of this channel, mainly through further widening. Thus, the morphologic evolution of the upstream end of chute cutoff channels is driven by positive feedbacks between the channel morphology, both through bar growth and changes in channel dimensions, and the flow structure associated with the offtake of flow from the main channel (Figure 14).

[39] The downstream end of a chute cutoff channel, where flow through the cutoff channel merges with flow moving around the meander bend, shares many characteristics with asymmetrical river confluences [Best and Reid, 1984; Best 1986, 1987, 1988, Reid et al., 1989; Rhoads and Kenworthy, 1995; 1998; Rhoads and Sukhodolov, 2001; Sukhodolov and Rhoads, 2001; Weber et al., 2001; Boyer et al., 2006; Best and Rhoads, 2008; Rhoads et al., 2009]. The confluences at these cutoffs are characterized by (1) flow stagnation at the upstream junction corner, (2) penetration of flow from the cutoff channel into flow that has moved around the original meander bend, (3) deceleration of flow along the left bank of the main channel downstream of the cutoff mouth, (4) helical motion of the confluent flows, (5) the development of a mixing interface between flow exiting the cutoff and flow in the main channel, and (6) deposition of an assemblage of bars in the main channel (Figure 14).

5.2 Chute Bifurcations

[40] The reduction in flow velocity in the main channel across from the cutoff offtake (Figures 3, 11, and 12) is the result of fluid mass and momentum transfer across the main channel, as flow curves into the cutoff channel. This same pattern has been observed in studies of flow diversions from curved channels [Fares and Herbertson, 1993; Neary and Odgaard, 1993; Fares, 1995; Ramamurthy et al., 2007]. The reduction in flow velocity is accompanied by a decrease in bed shear stress (Figures 3a and 3b), which can be related to a decrease in sediment-transport capacity. Fares [1995] found that the distribution of bed shear stress in the main channel is such that bar development is expected to begin downstream of the diversion entrance on the bank opposite the diversion and extend across the channel. This pattern of bed development downstream of offtake entrances is identical to that reported herein (Figures 8 and 9 and labeled 1 in Figure 10) and has also been observed previously in natural cutoffs [Fisk, 1947; Shields and Abt, 1989; Hooke, 1995; Fuller et al., 2003] and human-induced cutoffs [Tiron et al., 2009].

[41] Growth of a bar across the entrance to the meander bend (see feature labeled “1” in Figure 10) likely produces a positive feedback by directing increasing amounts of flow into the cutoff channel, thereby increasing Ured (Figure 11), which promotes further bar development. Eventually, this bar should become large enough to completely prevent water from moving into the original bend when flow is below bankfull stage, effectively sealing off the bend entrance and initiating formation of an oxbow lake [Gagliano and Howard, 1984]. Observations at Mackey Bend in August 2012 indicate that bar development may be enhanced by vegetation growth on the bar surface during low stages, which subsequently promotes deposition at high flow stages when the bar is partly or fully submerged.

[42] Nodal point relationships and three-dimensional models for the division of flow and sediment at bifurcations [e.g., Bolla Pittaluga et al., 2003; Kleinhans et al., 2008, 2011] also provide insight into the processes occurring at the upstream end of chute cutoff channels. The tendency for one bifurcate channel to become dominant over the other is a function of the Shields parameter of the upstream channel [Bolla Pittaluga et al., 2003], the slopes of the two bifurcate channels, the curvature and length of a bend upstream of the bifurcation, the width:depth ratio of the upstream channel, local topographic effects, and changing boundary conditions [Kleinhans et al., 2008; Thomas et al., 2011]. Chute cutoffs are clear examples in which one bifurcate channel (the chute channel) has a steeper gradient than the other channel (in this case, the original bend). At Mackey Bend, changes in backwater from the Ohio River modify this gradient advantage, thereby possibly delaying the rate of cutoff development compared to bends uninfluenced by backwater effects.

[43] Additionally, the bifurcation at a chute cutoff can be affected by channel curvature upstream of the cutoff, which influences the division of flow and sediment at the bifurcation through the inherited secondary flows and the gravitational influence of transverse and longitudinal bed slopes on sediment transport. Three-dimensional modeling has shown that the dominant bifurcate channel (i.e., outer bank or inner bank) is controlled by the width:depth ratio of the upstream channel and the length of the upstream bend [Kleinhans et al., 2008]. The second cutoff at Mackey Bend is located on the outer bank of a relatively tight bend upstream (the ratio of the radius of curvature: channel width, R/W, ~ 2.3). The observed deepening and widening of this cutoff channel, and deposition in the main channel downstream of the bifurcation, show that cutoff development is dominated by flow processes near the outer bank. The division of discharge at the second cutoff channel bifurcation (QC2/Qtotal = 0.54, Table 1) on 30 June 2011 indicates that the second cutoff channel is still in the beginning of phase 2 of bifurcation development, as defined by Kleinhans et al. [2008], at the time of this survey. The development of the bifurcation at the second cutoff at Mackey Bend may diverge from the stable third phase of development described by Kleinhans et al. [2008] due to the gradient advantage of the cutoff channel that should eventually result in complete abandonment of Mackey Bend.

[44] The separation zone observed along the left (inner) bank of the cutoff channel (Figures 3a, 3b, 4, and 6 and supporting information Figures S1, S2, S11, and S12) is also a common feature of flow diversions with angular planform geometries [Neary and Odgaard, 1993; McLelland et al., 1995; Ramamurthy et al., 2007] and is analogous to the flow separation zone that forms at the inner bank of sharp bends [e.g., Bagnold, 1960; Leeder and Bridges, 1975; Bathurst et al., 1977; Thorne et al., 1985; Blanckaert, 2011]. Due to the high inertia of the flow as it curves into the cutoff channel, it is unable to remain attached to the left channel bank downstream of the angular corner at the cutoff entrance and forms a region of separated flow. The size of the separation zone is expected to increase as (i) the momentum of the flow in the main channel increases relative to that moving into the cutoff channel and (ii) the angle of entry into the cutoff channel increases [Neary and Odgaard, 1993; McLelland et al., 1995; Ramamurthy et al., 2007]. The rapid widening of the cutoff channels and development of bars in the separation zones (labeled “2” in Figure 10) in both cutoff channels at Mackey Bend altered the flow structure, and it was thus difficult to confirm the relationship between momentum ratio and the size of the flow separation zone with the field data presented herein. However, the depth-averaged velocities from the second cutoff suggest that the size of the separation zone does increase with increasing momentum of the main channel flow relative to the cutoff channel flow (Table 2 and Figure 12), which is consistent with the conclusions of earlier studies [Neary and Odgaard, 1993; McLelland et al., 1995; Ramamurthy et al., 2007].

[45] The separation zone is associated with low values of bed shear stress (Figures 3a and 3b), indicating that flow within this zone has a small sediment-transport capacity. Sediment moving into the separation zone from the adjacent freestream is likely deposited, resulting in extensive bar development (Figures 810). Sedimentation in the separation zone of chute cutoff channels has been documented previously [Hooke, 1995; Fuller et al., 2003] and can produce key feedbacks between flow structure, sediment transport, and the morphology of the cutoff channel [Neary and Odgaard, 1993; McLelland et al., 1995].

[46] The curving of flow streamlines into a diversion has been compared to flow through a bend [Neary and Odgaard, 1993]. Similar to flow through meander bends, the streamline curvature induces helical motion in the flow through imbalances of the pressure gradient force with centrifugal force over the flow depth [e.g., Leeder and Bridges, 1957; Bathurst et al., 1977; Thorne et al., 1985]. Additionally, entrainment of flow into the separation zone may enhance skew-induced vorticity and the lateral across-channel transfer of fluid [McLelland et al., 1995]. Neary and Odgaard [1993] show that the strength of secondary circulation, defined as the difference between flow velocity at the surface and flow velocity near the bed at a given point, is proportional to the log of the ratio of the mean flow velocity in the diversion to the mean flow velocity in the main channel (Ucutoff/Umain). Based on their experimental data, Neary and Odgaard [1993] showed that the threshold velocity ratio for incipient secondary circulation in the cutoff is Ucutoff/Umain = 0.03. During the events measured at the Mackey Bend cutoffs, this stability threshold was greatly exceeded (Table 2), producing strong secondary circulation in the upstream portion of the chute cutoff channels. Secondary circulation in both cutoff channels tended to be confined to the thalweg in transects where the left-bank flow separation zone was extensive (Figures 4 and 6 and supporting information Figures S3, S4, S12, and S13).

[47] The development of strong helical motion has a pronounced influence on the morphologic evolution of the cutoff channels. Secondary circulation results in a transfer of momentum toward the bed, leading to the development of a submerged core of high velocity (Figures 4 and 6 and supporting information Figures S3, S4, S12, and S13) and high bed shear stresses at the bank toe. This effect promotes erosion of the bank toe and subsequent bank failures [e.g., Frothingham and Rhoads, 2003; Engel and Rhoads, 2012]. The transfer of momentum toward the right bank of the cutoff channel is also likely reinforced by the strong topographic steering of flow by the bar formed in the flow separation zone on the left side of the channel. Growth of the bar in the flow separation zone effectively reduces the flow width, forcing flow toward the right bank and increasing the velocity in the channel thalweg [Blanckaert, 2011].

5.3 Chute Confluences

[48] The ADCP measurements from Mackey Bend show that fluid is essentially stagnant at the upstream junction corner of the cutoff confluence (Figures 3c, 3d, and 5 and supporting information Figures S5, S7, S13, and S15) and characterized by low bed shear stress (Figures 3c and 3d). Flow stagnation, a well-documented feature of channel confluences [Best, 1987, 1988; Rhoads and Kenworthy, 1995, 1998; Rhoads and Sukhodolov, 2001; Best and Rhoads, 2008], develops as superelevation of the water surface at the upstream junction corner creates an adverse pressure gradient, causing the converging fluid from the two channels to stall [Best, 1987; Rhoads and Kenworthy, 1998; Rhoads and Sukhodolov, 2001]. Low velocities or recirculating fluid in this area can lead to sediment accumulation [Best, 1987; Best and Rhoads, 2008], as observed at the upstream junction corners of the two chute cutoffs at Mackey bend (Figures 810). As the junction angle and momentum ratio at a confluence increase, flow from the tributary penetrates farther into the main channel flow, increasing the size of the stagnation zone and pushing it farther into the main channel [Best, 1987; Rhoads and Kenworthy, 1998; Rhoads and Sukhodolov, 2001].

[49] A region of low flow velocities was observed between the two thalwegs of the converging flows (Figures 5 and 7). A similar low-velocity region has been observed in field studies of flow structure at small confluences [Rhoads and Kenworthy, 1998; Rhoads and Sukhodolov, 2001]. This low-velocity region is most likely the result of the advection of near-bank, low-velocity flow into the confluence from the upstream channels and may be linked to the region of flow stagnation at the junction corner. The low-velocity zone becomes smaller and less pronounced as flow momentum is redistributed downstream through the confluence (Figures 3, 5, and 7 and supporting information Figures S7 and S15) [Rhoads and Sukhodolov, 2001]. For the first cutoff, the mixing interface, defined by a strong cross-stream gradient in the ADCP backscatter intensity, corresponds spatially with the left edge of the low-velocity zone (Figure 5). The location of the mixing interface for the second cutoff is more difficult to define based on the backscatter intensity (Figure 7). The variability in backscatter intensity may be due to the influence of flow exiting the first cutoff channel, as well as sediment suspension due to the effect of bed topography on the flow structure. However, even given these caveats, the approximate location of the mixing interface, as based on the backscatter intensity, corresponds well with the left edge of the low-velocity zone in C2-12 (Figure 7).

[50] Surface-convergent helical cells were identified in the cutoff confluence in the Rozovskii frame of reference (Figures 5 and 7 and supporting information Figures S9, S10, S17, and S18), which has previously been used to detect secondary circulation at junctions [Rhoads and Kenworthy, 1999]. This helical motion is likely produced by curvature of the streamlines as the two flows enter the junction and realign with the downstream channel [Rhoads, 1996], as well as the influence of the bed topography [Best, 1988]. In the present study, curvature of the main channel aids in the production of secondary circulation in the fluid moving into the confluence from Mackey Bend—a mechanism that has been observed previously at confluent meander bends [Roberts, 2004; Riley and Rhoads, 2012]. At the downstream end of the first cutoff, the counter-rotating cells are centered on the cores of high streamwise velocity from the two converging flows and are separated by a zone of low primary velocity (C1-13, Figure 5d, and supporting information Figures S9 and S10). Clockwise secondary circulation extends beyond the position of the mixing interface, as defined on the basis of the backscatter intensity data, over most of the zone of low velocity. This fluid motion may represent an extension of the main secondary flow cell that is centered on the cutoff flow or, alternatively, may arise independently from the locally complex bed topography associated with this region of deeper flow/scour. Counter-rotating helical cells are also centered on the two cores of high streamwise velocity at the downstream end of the second cutoff (C2-12 and C2-13, Figures 5c and 5d, and supporting information Figures S17 and S18). In this case, the zone of low flow velocity is smaller than that for the first cutoff and does not show any local secondary circulation. Previous studies have associated the boundary between surface-convergent helical cells in a confluence with the location of the mixing interface [Mosley, 1976; Ashmore and Parker, 1983; Best, 1986, 1987; Rhoads and Kenworthy, 1995, 1998; Rhoads, 1996; Rhoads and Sukhodolov, 2001; Sukhodolov and Rhoads, 2001]. The location of the mixing interface is positioned close to the boundary between the main cells for the second cutoff (Figure 7c) but is shifted slightly toward the clockwise cell in the cutoff flow in the case of the first cutoff (Figure 5e). This discrepancy may reflect imprecision in the delineation of the mixing interface on this date or possible extension of the main cell in the cutoff flow beyond the mixing interface.

[51] The bed morphology associated with this characteristic flow structure includes development of a bar complex extending from the stagnation zone along the left bank upstream of the cutoff mouth to the zone of reduced velocity along the left bank downstream of the cutoff mouth and is herein referred to as the cutoff-mouth-bar assemblage (Figures 10 (labeled “3”) and 14). Most of the sediment deposited in the cutoff-mouth-bar assemblage at Mackey Bend was likely supplied by the rapid widening of the cutoff channel. Tributary-mouth bars are commonly observed in confluences where one channel dominates the flow structure [Petts and Thoms, 1987; Best, 1988; Reid et al., 1989; Mosher and Martini, 2002; Best and Rhoads, 2008; Rhoads et al., 2009]. Bars with distinct fronts developed at the confluence of both cutoff channels (Figures 5 and 710), and portions of these bars are subaerially exposed at low flow stage (Figure 10). Deposition of sediment at the mouths of the cutoff channels reflects abrupt deceleration of high-velocity flow moving through the relatively steep cutoff channel as it enters the comparatively low-gradient main channel. The abrupt deceleration of flow at the cutoff channel mouth is reflected in the patterns of bed shear stress (Figures 3c and 3d). The extension of the mouth-bar assemblage along the left bank downstream of the cutoff mouth (Figures 810) indicates that these locations are characterized by decreasing bed shear stress and a loss of transport capacity during transport-effective flows—a mechanism that can produce bar formation at confluences [Rhoads and Kenworthy, 1995; Rhoads, 1996; Rhoads et al., 2009]. While the data collected in this study do not provide direct evidence of large-scale flow separation, a phenomenon documented at laboratory confluences [Best and Reid, 1984; Best, 1987, 1988], this effect may occur at the downstream junction corner of the cutoffs under suitable conditions.

5.4 Future Development of the Cutoff Channel

[52] Now that the second cutoff has captured a majority of flow of the Wabash River, the first cutoff has become erosionally dormant [Zinger et al., 2011]. Eventually, the second cutoff should completely capture the flow of the Wabash River, transforming Mackey Bend and the first cutoff into a lake. Long-term observations of cutoffs indicate that closure of the upstream limb of a cutoff bend typically occurs prior to closure of the downstream limb [Hooke, 1995]. Once the upstream limb of the bend becomes plugged with sediment and the cutoff channel captures all of the flow in the upstream river channel, the abandoned bend undergoes a transition to stagnant, lentic conditions [Gagliano and Howard, 1984; Glińska-Lewczuk, 2009]. Flow at the downstream end of the abandoned bend, which is still partially open to the new river channel path, is analogous to flow in a local slackwater embayment along the river [Le Coz et al., 2010]. Progressive deposition may cause the downstream end of the bend to be closed gradually, as sediment is entrained into the embayment from fluid shear between flow in the river and stagnant water at the open downstream end of the bend. During this phase of oxbow lake formation, bar growth in the downstream limb progresses from the bank across from the cutoff channel mouth toward the upstream junction corner [Le Coz et al., 2010] until it coalesces with the zone of deposition associated with the cutoff-mouth-bar assemblage.

[53] Thus, the transformation of Mackey Bend into an oxbow lake likely will progress according to the conceptual model, in which bar development across the entrance to the bend eventually prevents flow from entering the meander, the cutoff channel widens, and deposition at the downstream end of the cutoff channel continues to occur. Once flow into the bend is blocked by sedimentation at the upstream end of the cutoff, deposition into slackwater in the downstream limb of the abandoned meander will proceed until the cutoff channel is completely integrated into the path of the main river. Continued realignment of the cutoff-channel planform should occur during this integration, although the width of the cutoff channel will stabilize. The timescales of blockage and abandonment of bends at a variety of scales of channel size are not well known, but reported values range from 2 to 10 years on the lower Mississippi River [Gagliano and Howard, 1984], <1 to 7 years on the Rivers Dane and Bollin [Hooke, 1995] to 15 years on a bend of the Wabash River at Grayville (see Jackson [1975, 1976] for details of this bend when active). At Mackey Bend, completion of the cutoff process, as well as the initiation of the cutoff itself, is delayed by backwater effects from the Ohio River, which diminish the erosive potential of flows through the cutoff channel when the Ohio River is at a high stage. The flow conditions that occurred when the first cutoff channel was initiated and gullies were carved in the location of the second cutoff (i.e., the Wabash River is at a high flood stage and the Ohio River is not) occur rarely according to historical data (Figure 2).

[54] Rapid lateral erosion downstream of cutoffs has been observed on several rivers [Kulemina, 1973; Brice, 1974; Bridge et al., 1986], and it has been argued that these high lateral erosion rates result from the increased sediment supply associated with cutoff channel incision [Nanson and Hicken, 1983]. Following the cutoffs at Mackey Bend, extreme sedimentation occurred downstream at the nearby confluence of the Wabash and Ohio rivers [Zinger et al., 2011]. Upstream effects may also be produced by formation of a local knickpoint in the longitudinal profile that translates along the river, but in some cases cutoffs do not appear to affect the morphologic evolution of adjacent bends [Hooke, 1995]. Because the cutoff is incomplete and the channel is still actively evolving, the long-term upstream and downstream morphodynamic effects of the Mackey Bend cutoffs have yet to be revealed.

6 Conclusions

[55] This study has examined the hydrodynamics and morphodynamics of two chute cutoff channels formed along a large river immediately following initiation of the cutoffs. The results shed light on the fluvial processes and forms that characterize chute cutoff channel dynamics after initiation of a cutoff but prior to the formation of an oxbow lake. A general conceptual model of chute cutoff hydrodynamics and morphodynamics is developed to encapsulate insights gained from the study. The conceptual model places the results of the present field study within the context of relevant extant literature, such that the processes illustrated by the model are broadly applicable to chute cutoffs in a variety of environments over a wide range of scales.

[56] A key aspect of this model is that it views bifurcations and confluences as suitable analogs for depicting flow structure, patterns of erosion and deposition, and short-term morphodynamics at the upstream and downstream ends, respectively, of cutoff channels. Insights from previous work on bifurcation stability suggest that chute cutoffs “succeed” due to a combination of factors, including a gradient advantage and the influence of the bend upstream. Capture of flow by an incipient cutoff channel triggers positive feedbacks between flow and evolving channel form. These responses include a feedback between the reduction in velocity in the upstream limb of the main channel and bar formation across the entrance to the meander bend, and a feedback between the increase in the percentage of discharge captured by the cutoff channel and widening of the cutoff channel. Feedbacks between the curvature of flow into the cutoff channel, the development of a flow separation zone, and bar formation in the flow separation zone also drive cutoff channel widening. Furthermore, deposition of sediment excavated from the floodplain by erosional enlargement of the cutoff channel generates a cutoff-mouth-bar assemblage, which partially blocks the downstream limb of the bend. Spatial patterns of deposition in the cutoff-mouth-bar assemblage respond to, and develop because of, the same processes that occur in channel confluences.

[57] The results of the present study and previous research on oxbow lake formation [e.g., Hooke, 1995; Le Coz et al., 2010] indicate that the transformation from a chute cutoff to an oxbow lake is a two-stage process. The first stage of this transformation is the period of time following initiation of the chute channel and prior to closure of the upstream limb of the bend. The conceptual model presented herein encompasses the first stage of development, which includes the period during which the chute channel rapidly widens, sediment plugs the upstream limb of the bend, and the cutoff-mouth-bar assemblage is deposited (Figure 14). However, the model does not address how chute cutoff channels are initially carved across the necks of meander bends [e.g., Fisk, 1947; Bridge et al., 1986; Gay et al., 1998; Constantine et al., 2010b; van Dijk et al., 2012], and additional research is needed to examine this issue. Although the second stage of oxbow lake formation, which takes place after the upstream limb of the bend is plugged, has been described in detail by Hooke [1995] and Le Coz et al. [2010] and is encapsulated in a conceptual model by Saucier [1994], the exact mechanisms by which the entrance and exit to a meander bend are sealed also must be explored further to connect the model in this study to the formation of oxbow lakes. Additional field investigations of chute cutoffs over a range of spatiotemporal scales are required to fully evaluate the validity of the conceptual model of chute cutoff dynamics presented herein. Future work should include experimental and numerical modeling to better identify the process-form interactions involved in the initial development of chute cutoff channels, the evolution of these channels once formed, and the formation of oxbow lakes.

Acknowledgments

[58] This research was supported by NSF SGER Grant (BCS-0852865) and NSF Grant 112554. We thank Kevin Collier, Bill Norfleet, and Bill Floyd for access to the field site. We also thank Frank Engel, Jim Riley, and Kory Konsoer for assistance with data collection in the field and Ross Jackson for discussions on Wabash River morphodynamics. This manuscript was greatly improved by the comments of A. Densmore, J. Buffington, W. van Dijk, J. Hooke, and one anonymous reviewer.