Hydrodynamic and geomorphic controls on mouth bar evolution


  • Christopher R. Esposito,

    Corresponding author
    1. Department of Earth and Environmental Sciences, The University of New Orleans, New Orleans, Louisiana, USA
    2. Department of Earth and Environmental Sciences, Tulane University, New Orleans, Louisiana, USA
    • Corresponding author: C. R. Esposito, Department of Earth and Environmental Sciences, Tulane University, New Orleans, LA 70118, USA. (cesposit@tulane.edu)

    Search for more papers by this author
  • Ioannis Y. Georgiou,

    1. Department of Earth and Environmental Sciences, The University of New Orleans, New Orleans, Louisiana, USA
    Search for more papers by this author
  • Alexander S. Kolker

    1. Department of Earth and Environmental Sciences, Tulane University, New Orleans, Louisiana, USA
    2. Louisiana Universities Marine Consortium, Chauvin, Louisiana, USA
    Search for more papers by this author


[1] While river deltas are one of the major repositories for sediments and carbon on Earth, there exists a paucity of field data on the formation of distributary mouth bars—one of their key features. Here we present results from an experiment that tested a model of mouth bar development using hydroacoustic, optical, sedimentary, and geochemical tools on a mouth bar in a crevasse splay near the mouth of the Mississippi River. Our results validate an existing model for mouth bar development, which we extend to explain mouth bar stratigraphy. We propose that changes across a hydrological cycle are important for mouth bar development, resulting in a stratigraphy that has alternating fine and coarse grain sediments. Results also indicate that sand is carried up to 6 km from the main stem of the Mississippi River, despite repeated channel bifurcations, which has important implications for our interpretation of the rock record, understanding of coastal sedimentary systems, and the restoration of large deltas.

1 Introduction

[2] River mouths are among the most important interfaces on earth. They are a place where terrestrially derived material interacts with the ocean, and their dynamics place a primary constraint on the fate and transport of sediments, nutrients, and other key dissolved and suspended constituents across the Earth's surface [McKee et al., 2004]. Distributary mouth bars are among the most conspicuous features associated with the mouths of large rivers [Edmonds and Slingerland, 2007, hereinafter ES]. They occur at a transition between the predominantly channelized flow of the distributary and the largely unconfined environment of the receiving basin [Wright and Coleman, 1974] and are among the most active sites of deposition within a delta system. These transitional landforms help link subaqueous and subaerial regimes and rapidly transform shallow water into subaerial land, which makes them important to coastal planners, scientists, and engineers seeking to manage river deltas for human use. Their high preservation potential also makes mouth bars appealing targets of study for petroleum and economic geologists.

[3] While mouth bars and their associated distributary channels exert a first-order control on the plan form of the developing distributary network [Olariu and Bhattacharya, 2006; Edmonds and Slingerland, 2010; van Heerden and Roberts, 1988], the physical processes driving their evolution have not been well documented in the field using modern hydroacoustic, optical, or geochemical methods [Paola et al., 2011]. Fisk [1944] noted the prevalence of mouth bar facies in the Mississippi River Delta, as did Coleman and Gagliano [1964], who also placed mouth bar deposits into the stratigraphic sequence of prograding delta deposits. Wright and Coleman [1974], using a fluid dynamical approach, explained mouth bar formation as the deposit of an expanding, turbulent, sediment-laden jet plume, an idea that remains at the center of modern studies. Wellner et al. [2005] built upon this concept and examined delta architecture across a range of time and length scales, demonstrating that the Wax Lake Delta is the product of numerous horizontally and vertically offset jet-plume deposits, each of which consists of a complex of distributary mouth bars. These models become more insightful when viewed in the light of observational studies demonstrating punctuated sediment transport in the Mississippi River [Nittrouer et al., 2012a], suggesting that variable stratigraphic sequences will be found at the river's mouth.

[4] Edmonds and Slingerland [2007] used a numerical model and a set of observations from 11 deltas around the world to develop a conceptual framework for mouth bar evolution that serves as the starting point for this study. Their model (2007 ES), which is based on the Wright and Coleman's [1974] jet-plume concept, describes the development of a distributary mouth bar as a sequence of aggradation, progradation, and runaway aggradation. The initial aggradation phase takes place as a confined jet expands upon exiting the channel mouth, resulting in a sharp decrease of velocity along flow streamlines and a rapid accumulation of sediment on the receiving basin floor. Continued aggradation on the bar top causes flow to constrict vertically, increasing bar top velocities and initiating the progradational phase of bar development as material is scoured from the bar top and deposited toward the distal edge of the bar. Progradation ceases when the downstream pressure gradient above the bar is insufficient to overcome the frictional resistance introduced by the bar, forcing flow into the channels that flank the bar. This final phase is termed runaway aggradation, as the nearly stagnant water that now covers the bar deposits its entire sediment load on the bar top, driving the bar rapidly toward the water surface and giving the final triangular planform shape.

[5] Despite a theoretical underpinning and its ability to explain modern mouth bar morphology, the ES[2007] model has remained largely untested in the field, leaving gaps in our understanding of these important systems. Here we tested the ES[2007] model, using a field study of a mouth bar near the mouth of the Mississippi River that employed a suite of hydrological, geochemical, and sedimentological tools. Our findings support the conclusions about mouth bar development presented in ES[2007], and we show how the model can be extended to explain the development of mouth bar stratigraphy. We suggest that significant sedimentation on the mouth bar continues throughout a large fraction of the river year, which allows for the development of a stratigraphy with alternating coarse and fine material. This distinctive sedimentological signature of interbedded coarse and fine layers, noted by van Heerden and Roberts [1988], is the result of seasonally varying riverine input that drives the mouth bar rapidly through phases similar to those outlined in ES[2007].

2 Methods and Data Analysis

[6] This study was conducted in the Cubits Gap subdelta near the mouth of the Mississippi River. The system originally developed in the mid-nineteenth century following the creation of an artificial crevasse along a Mississippi River levee. This led to the development of a nearly 200 km2 system of channels, mouth bars, crevasses, wetlands, and subaqueous habitats [Wells and Coleman, 1987; Cahoon et al., 2011]. We focus our work on the Brant's Pass Splay, a largely unmanaged developing crevasse splay (Figures 1a–1c).

Figure 1.

(a) Illustration of the modern active Mississippi River Delta (Balize Delta). (b) The channel network in Cubits Gap with the location of the ADCP transects (solid lines) with numbers corresponding to those in Table 1, and the Brant's Pass Splay mouth bar marked in the dark box. The sonde location is marked as a black dot. (c) The Brant's Pass Splay mouth bar with bar and channel elevations referenced to the National Geodetic Vertical Datum of 1929. (d) Channel cross sections leading to the mouth bar. (e) The 2010 Mississippi River 2010 hydrograph at Belle Chasse, LA, obtained from the United States Geological Survey (USGS). The grey vertical solid lines show the times of the field expeditions.

Table 1. Flow Distribution in Cubits Gap Subdelta and Channel Hydraulic Characteristicsa
Location (season)Discharge (m3/s)Q (m3/s) as Percent of Cubits GapHydraulic Radius (m)Surface Slope Regime90th Percentile Velocity (m/s)Channel Depth (m)Froude Number
  1. aDischarge and hydraulic radius are both calculated from ADCP transects taken at the locations marked in Figure 1b. Surface slopes were divided into three flow regimes based on a visual inspection of GPS elevations taken continuously while surveying the system. The transitions in slope regime occur at the splay entrance (just above Transect C) and at the turnoff toward the mouth bar (just above Transect E). The 90th percentile velocities were calculated from ADCP data.
A (April)2711100.00%4-
B (April)68525.30%4.31.95E-
C (April)2619.60%2.196.79E-
D (April)1324.90%1.86.79E-051.1630.21
E (April)180.70%1.051.27E-040.9720.22
A (Nov )627100.00%-----
C (Nov)396.20%-----

[7] We conducted two field visits to the Brant's Pass Splay to document flow regimes and the splay environment during distinctly different hydraulic conditions (Figure 1e). We timed the first field visit, in April 2010, to coincide with the peak spring flood flows in the Mississippi River while the second, in November 2010, coincided with the lowest flows of the year (Figure 1e). By sampling across a hydrological cycle, we were able to partially capture the variability in sediment regimes in the Mississippi River, which is critical, as it has been observed that sand is carried in suspension in the river during high flow but not during low flow [Allison and Meselhe, 2010].

[8] Channel hydrodynamics were evaluated during the April flood conditions using a vessel-mounted acoustic Doppler current profiler (ADCP) in tandem with a differential global positioning system (DGPS). Data from the ADCP/DGPS survey were sufficient to observe water surface slopes, channel discharges, and peak velocities throughout the splay and its feeder system (Table 1 and auxiliary material), as well as channel geometries (Figure 1d). Hydrodynamic and sediment transport conditions on the bar top were assessed using a Sequoia Scientific Laser In-Situ Scattering and Transmissometry (LISST) field particle size analyzer in concert with a Nortek Vector Acoustic Doppler Velocimeter (ADV). The LISST/ADV transect shown in Figures 2b and 2d was selected to show the distribution of bar top water velocities, suspended sediment size, and concentrations along flow streamlines from the bar apex to the receiving basin.

Figure 2.

Hydrodynamic and sedimentary observations in Brant's splay. (a) Negatives of X-radiographs of push cores showing heterogeneous stratigraphy, with relatively coarse grained material appearing as darker strata and relatively fine grained material as lighter strata. (b) Downstream variation in velocity and suspended sediment size and volume along flow streamlines, along the transect location shown in Figure 2c as black solid line. (c) Down-core distribution of particle size in three sediment cores, with the median grain size in solid red line and the 10th and 90th percentiles in black dashed lines. (d) Sedimentation thickness determined from 7Be for cores collected in April 2010.

[9] To understand seasonal-scale sedimentation, 16 short (5–15 cm) sediment cores were collected during the April deployment (Figure 2d). Cores were X-radiographed to determine the shallow stratigraphy and then sampled at 1 cm intervals. These subsamples were dried, combusted at 450°C to determine percent organic content, and analyzed for 7Be on a low-energy germanium detector for ~24 h. 7Be is a naturally occurring, particle-reactive radionuclide with a half-life of 53.1 days, and as such makes an ideal tracer of short-term sedimentation [Allison et al., 2000; Kolker et al., 2012]. Here the presence of 7Be indicates deposition over the 2010 flood season under the assumption that any 7Be present in sediments deposited prior to the 2010 flood season would have decayed to undetectable levels given the short half-life of 7Be. This analysis allows us to approximate the thickness of the 2010 flood unit (Figure 2d). Five longer cores (0.5–1.0 m) were obtained during November 2010, during low-flow conditions and when the bar was exposed. Detailed information on core collection and stratigraphy is available in the auxiliary material and in Esposito [2011].

3 Results

[10] The data collected provide a synoptic view of channel/mouth bar dynamics, coupled with stratigraphic evidence that provides information on the evolution of this system over seasonal and annual time scales. The primary survey was conducted during April 2010, when flow in the Mississippi River as measured at Belle Chasse was ~23,000 m3 s−1, while the secondary survey was conducted in November 2010, when flow in the Mississippi River was ~7100 m3 s−1. These values are close to the seasonal high and low values for the Mississippi River in 2010, and the April value is about 70% of the maximum amount of water allowed to flow past New Orleans (35,500 m3 s−1). The ADCP data reveal a decrease in water fluxes with downstream bifurcations, with fluxes ranging from 2711 m3 s−1 at the mouth of Cubit's Gap to 261 m3 s−1 at the entrance to the splay and 18 m3 s−1 upstream of the mouth bar. The velocities measured in the channel at these locations showed little change from station to station. The average 90th percentile velocity for all ADCP measurements was 1.08 (±0.11) m s−1, and the 99th percentile velocity was 1.19 (±0.13) m s−1. There is little downstream variation in these velocities, which appear to be independent of the channel cross section or discharge (Table 1). Tidal velocities downstream ranged from 0.50 to 0.80 m s−1, as measured by a sonde placed distal to the mouth bar. This variability is associated with diurnal, tidally driven fluctuations in water level (0.22 m) and temperature (14.5–18.5°C). Velocity measurements on the bar itself reveal a general decrease in velocity across the bar top, ranging from 0.27 m s−1 at the apex of the bar to 0.17 m s−1 at the distal end of the bar (Figure 2b), all in a subcritical flow regime. This decrease in velocity is accompanied by changes in the volume and diameter of suspended particles. The sediment volume and particle size of sediments in the water column both increase with distance from the bar apex, reaching maxima of 430 μL/mL and 43 µm, respectively, at the bar crest and then decrease toward the bar's distal edge as the flow expands (Figure 2b).

[11] A series of shallow cores reveal sediment dynamics over seasonal and annual time scales. Figure 2d shows the depth of 7Be penetration. The distal fringe of the bar is marked by the most intense deposition (≥3 cm) and also by a decline in the amount of suspended material (Figure 2b). The central bar is free of any detectable depositional activity but is notable for the aforementioned peak in the suspended sediments (Figure 2b). Core data reveal that the bar consists of fine sand, interbedded with consolidated and unconsolidated fine materials that we will simply refer to as mud (Figure 2a). These cores consist of approximately three stratigraphic units, each centimeter to decimeter in scale, and alternate between consolidated and unconsolidated sands and mud. Patterns of alternating sand and mud are observable in the X-radiographs and the particles' size data (Figure 2c) and are further presented in the auxiliary material.

4 Discussion

[12] The present study builds on previous work [van Heerden and Roberts, 1988; Wellner et al., 2005; Allison and Meselhe, 2010; ES] to indicate how channel hydraulics and seasonal fluctuations in water and sediment discharge can lead to the variable stratigraphy observed in this mouth bar. Hydraulic and sedimentological data presented here (Figures 2b and 2d) show that ES[2007] correctly describe the hydraulics and resulting sediment transport over a mouth bar in flood. However, core data (Figure 2a) taken in this study and others [Wellner et al., 2005; van Heerden and Roberts, 1988] show a more varied stratigraphic product than a direct application of the [2007]ES model would suggest. Instead of a three-stage process of aggradation, progradation, and runaway aggradation at the mouth bar (ES[2007]), we see evidence of cyclically shifting depositional regimes that alternately deposit sand and mud in the same location (Figures 2a and 2c, and auxiliary material). We suggest that depositional processes in this mouth bar are significant throughout a substantial portion of the year, rather than only during floods (Figure 3). Furthermore, mud is deposited and partially preserved during transitional or low-flow conditions in this setting. We propose that the ES[2007] model be extended to account for shifting patterns of sediment supply and delivery within an efficient channel network by defining the following three flow regimes:

  1. [13] During periods of high flow, the mouth bar can be in an aggradational, progradational, or runaway aggradational phase, as described in ES[2007]. Sand is transported as suspended load in the splay channels at higher river stages and then deposited on the bar (Figure 3a).

  2. [14] As the floodwaters recede, the depth of water over the bar decreases driving the bar into a runaway aggradational phase. Sand in the main stem of the Mississippi River is not in suspension during these conditions; hence, only mud is deposited on the bar (Figure 3b).

  3. [15] At low flows, the bar is exposed subaerially, allowing mud to consolidate and resulting in a stratigraphic record that alternates between fine- and coarse-grained materials.

Figure 3.

Extended conceptual mouth bar model, showing (a–d) different depositional regimes based on varying riverine input and flow over the bar.

[16] This results in a substrate that is able to resist the erosive force of the next spring flood and can be expected to be at least partially preserved under a new sand layer that is deposited on top (Figures 3c and 3d). Evidence of this preservation is found in core X-radiographs (Figure 2a), core descriptions (auxiliary material), and grain size analysis (Figure 2c).

[17] Our model provides an explanation for the interbedded sands and mud that are commonly observed in mouth bars [van Heerden and Roberts, 1988], and ties the cycling depositional environment to a river-dominated hydrological cycle as well as to the longer-term aggradation of the mouth bar. The thickness of the flood deposit (Figure 2d) is greatest around the fringe of the mouth bar, and the flow appears to be scouring sediment from the bar top (Figure 2b), placing this mouth bar in a progradational stage while the river is in flood. In November, the bar was completely exposed and topped with a thick layer of mud, indicating that it had progressed through stages 2 and 3 of our proposed model. While more study is clearly needed across seasonal gradients [Paola et al., 2011], we suggest that the regular transition between phases is critical to understanding the stratigraphic evolution of mouth bars and deltaic sediment budgets.

[18] To further support this model, we examine the ability of the Cubits Gap channel network to deliver sediment from the river to the mouth bar. While it is known that the Mississippi River at flood can carry a substantial fraction of its sand in suspension in certain portions of the river, it is striking that a coarse mouth bar feature could develop nearly 6 km from the river's main channel. A critical feature of the channel network of this splay is that its sediment transport capacity is undiminished with distance from the river. The key observation is that while the hydraulic radius and the depth of the channel decrease with each successive split in the channel network, the maximum observed velocities (expressed as 90th percentile velocity) do not significantly change, implying that the bed shear stress in the channels increases downstream (Table 1). This means that the competency of this system to transport sand remains largely unaltered in the downstream direction despite numerous bifurcations where one might expect a loss in stream power. Water surface slopes and Froude numbers both increase with distance from the river, corroborating the observations made from the ADCP data. We also note that the position of the channels in this system did not change substantially over the 10 years preceding this study, despite substantial land growth in this local complex. These results suggest that the flow characteristics did not change during this 10 year period, an observation which is supported by persistent deposition of fine to medium-grained sand at the apex of the mouth bar recorded in core CB-4 (Figure 2c).

5 Implications for Coastal Restoration and Interpretation of the Rock Record

[19] Many deltas worldwide are in a state of degradation, driven by high rates of subsidence, altered hydrological regimes, and global climate change [Syvitski and Saito, 2007]. The Mississippi River Delta exemplifies these types of impacts, which resulted in the loss of nearly 4800 km2 of land over the past century, as well as efforts to restore deltas [Day et al., 2007]. Plans exist to rebuild land in the delta by partially diverting the flow of the river, thereby reinitiating the natural land-building processes that originally built this system [Kim et al., 2008; Paola et al., 2011; Edmonds, 2012; Day et al., 2007; Coastal Protection and Restoration Authority of Louisiana 2012; Kolker et al., 2012]. Our results suggest that a diversion with an efficient channel network can deliver sand far into a receiving basin. While several recent studies have pointed out how extreme events such as the Great Mississippi/Atchafalaya River Flood of 2011 can be used as models for coastal restoration [Nittrouer et al., 2012b; Falcini et al., 2012], our findings point out that the normal seasonal cycle of sediment deposition also plays a critically important role in wetland development.

[20] To effectively build land, sediment deposition rates must exceed rates of relative sea level rise (RSLR). RSLR rates vary both spatially and temporally in the lower Mississippi River Delta [Tornqvist et al., 2008], ranging from <1 to >5 cm yr−1; rates of 1–3.5 cm yr−1 are most likely for this region [Kolker et al., 2011, 2012]. Our deposition rates can be compared to measurements from similar systems in the Mississippi River Delta collected during the 2011 flood (1.4 cm) [Falcini et al., 2012], the 2008 flood (0–10 cm) [Rosenheim et al., 2013], and the summer of 2009 (0–4.2 cm) [Kolker et al., 2012]. Taken holistically, these results suggest that well-managed diversions can deliver enough sediment to locally offset the high rates of observed RSLR.

[21] Preservation of flood deposits, such as the one we measured, must remain an active area for research. It plays a critical role in sediment budget calculations, with recent studies suggesting that sediment retention rates in the Mississippi River Delta range between 30% and 70% [Blum and Roberts, 2009]. Our model has implications for the interpretation of the rock record, as it suggests that the stratigraphy of a mouth bar can be influenced by seasonal variations in flow conditions and sediment input, in addition to previously mentioned controls [Wellner et al., 2005;ES].

[22] Future research on mouth bar dynamics should focus on identifying the thresholds at which different classes of sediment are transported in channel networks and from the channel environment onto the mouth bar. These thresholds, which will change as the bar develops, can be used to identify the three flow regimes presented here. The relative proportion of the river year that is spent in each flow regime contributes significantly to the thickness and sedimentological characteristics of the deposit, so studying mouth bars at seasonal time scales across a gradient of ages will connect the hydrologic cycle to the stratigraphic product.


[23] Financial support for this research was made possible by grants from the USGS Northern Gulf of Mexico (NGOM) Ecosystem Change and Hazard Susceptibility Program, and the Long-Term Estuary Assessment Group (LEAG) at Tulane University. The authors gratefully acknowledge Denise Reed for helpful comments and insights, Duncan FitzGerald for comments on an early version of the manuscript, and Chris Paola, Doug Edmonds, and one anonymous reviewer for their constructive reviews that improved the manuscript. Field and lab assistance from Mike Brown, Dallon Weathers, Duncan FitzGerald, Kevin Trosclair, Vallerie Cruz, and Alex Ameen is greatly appreciated.