Upper Mississippi River Flow and Sediment Characteristics and Their Effect on a Harbor Siltation Case

Upper Mississippi River flow and sediment characteristics downstream of St. Louis, MO are presented in this study. Available and measured data were used to assess a harbor siltation case and dredging needs. Such data are also useful to researchers and engineers conducting work in the Mississippi River, and large rivers in general. Flows were characterized in terms of the mean annual hydrograph, the flow duration curve and the mean annual, dominant and effective discharges. Suspended and bed material sediments were characterized by grain size distributions (GSDs). Suspended sediment concentrations were characterized with a sediment-rating curve, a mean annual sediment-graph and a duration curve. The results of the analyses were used to assess harbor sedimentation by comparing GSDs of harbor bed samples with those observed in the river. Bathymetric surveys were used to determine rates and occurrence of sedimentation. The analyses showed that harbor siltation correlates with river conditions, and is driven by wash load in the river, which enters the harbor in suspension and deposits along the bottom due to the lack of flow-through velocities high enough to keep the fine sediments in suspension.


Introduction
Sediment characteristics and loads in the Mississippi River have been the subject of numerous studies in the past decades. Their relation with land loss (e.g. Kesel 1989Kesel , 1988van Heerden and DeRouen 1997) has been one of the key factors driving the need to better assess the sediment loads in the river and its tributaries. Some recent studies on sediment load trends in the river basin (e.g. Horowitz 2010; Meade and Moody 2010; Blevins 2006) suggest that sediment loads are declining. In spite of this, potential for building river diversions that would carry sediment to certain locations along the shoreline to prevent further land loss at the Mississippi River delta, and along coastal Louisiana, has been recognized (e.g. Paola et al. 2011;Allison and Meselhe 2010). The amount of sediment diverted is a function of the flow and sediment load in the river (Dutta et al. 2017) and proper quantification of both variables is required. As a result, research has mainly focused on the Missouri and Ohio Rivers, which are responsible for the largest tributary sediment loads (Heimann et al. 2011), or on the lower Mississippi River sediment loads (e.g. empirical relation proposed by Hwang (1989) to estimate settling velocities in each zone is given zone concentration limits as defined in Table 3; When the suspension concentration is below a value C1, free settling occurs and the settling velocity corresponds with the one each particle would have in the absence of other particles.
Particles are far away from each other and no flocculation occurs. As concentrations increase, flocculation begins to occur and therefore the original grains begin to form flocs which have higher settling velocities. This process of flocculation settling continues up to a concentration C2, which corresponds with the maximum settling velocity (wsm). Above C2, the concentration becomes so high that the flocs have trouble settling and begin to collide with each other. Settling becomes hindered and could be thought of as a condition where water is trying to escape the pore space as sediment settles down. If concentration continues to increase and reaches a value C3, the process turns into a consolidation process rather than a settling one. Zone concentration limits and coefficients are not universal and depend on the sediment type and grain size distribution, as well as the environmental conditions in which the settling process takes place, such as salinity and turbulence or the lack thereof.
The settling velocity of the material found in the harbor was determined by conducting the settling column experiments first described by McLaughlin (1959) and later improved by Ross (1988). Fig.   5 shows the settling column used. It is 0.10 m in diameter and 1.9 m high, and has 5 mm sampling tubes located at the following elevations above the bed: 0.06 m, 0.16 m, 0.31 m, 0.51 m, 0.72 m, 0.93 m, 1.13 m, 1.33 m and 1.54 m. The design of the column is based on the one developed by Lott (1987), and the experimental procedure followed the one described by Ross (1988). The following five different initial concentration conditions: Co = 1 g/L, 5 g/L, 10 g/L, 15 g/L, and 25 g/L were used. Results from the experiments are shown in Fig. 6 along with a curve fit with Eq.1; the resulting coefficients are shown in Table 4.

Available Data
Data available at USGS gaging stations 07010000 at St. Louis, MO and 07020500 at Chester, IL were used to characterize the Mississippi River in the vicinity of the harbor. A summary of the data is shown in Table 5. Given that no significant tributaries flow into the Mississippi River between St. Louis, MO and Chester, IL a preliminary analysis showed that for the matching period of record July 1942 November 2011 the flow conditions, on average, differ by less than 1% (Fernández et al. 2012). Therefore, all analyses related to river data presented hereafter only used the information recorded at St. Louis, MO.

Suspended and Bed Material Sediment Characteristics
Grain size distributions for the sediments in the Mississippi River at St. Louis, MO were available as part of U S Geological Survey field/lab water quality samples (Table 5). Fig. 9 shows a total of 108 grain size distributions of the material traveling as suspended load and Fig. 10 shows a total of 114 grain size distributions for the material found on the bed of the Mississippi River at St.
Louis, MO. The solid black line represents the median grain size distribution curve, and the dashed lines correspond to the 75 th and 25 th percentiles. The sediment size for which 50% of the grains are smaller is 0.008 mm for the material traveling as suspended load and 0.44mm for the material found on the bed of the river.

Mean Annual Hydrograph, and Suspended Sediment Concentrations and Duration Curves
The mean annual flow hydrograph and mean annual sediment concentrations are shown in Fig.   11. A black dashed line spike can be seen in the sediment concentration hydrograph during late February. That line corresponds to the 30-year daily average concentrations but it is significantly biased by an extreme event that occurred in February of 1985, as shown in Table 7. If the values for those days are not included in the averaging process, the curve takes the shape of the solid line, which was taken as the representative mean annual sediment concentration curve herein. In this study, the concept of characteristic discharges is adapted to assess the flows responsible for the sediment loads in the Upper Mississippi River. Specifically, the concepts of dominant and effective discharge are used due to their relation with sediment loads in the river without consideration for morphological implications. The dominant discharge is defined here as the flow that, if sustained throughout a period of time, would produce the same mean sediment discharge observed during that period under varying flow conditions. The effective discharge is defined here as the one carrying the largest volume of sediment in the river. This definition is based on the bed-generative discharge concept first proposed by Schaffernak (1916Schaffernak ( , 1922, and its computation follows the approach described by Biedenharn et al. (2000). The method has been used and described by different authors (e.g. Garde and Ranga Raju 1977;Gandolfo 1940) but other authors refer to it as the dominant discharge (e.g. Thomas and Benson 1966). It is not the objective of this study to provide clarification and comparison between available definitions; the reader is referred to Soar and Thorne (2011) for a recent review on the subject.
Using the data available for the 1981-2011 period, the mean annual suspended sediment concentration was determined and the dominant discharge was back calculated with the sediment-rating curve shown in Fig. 8 and Eq. 2. The values obtained are 0.337g/L for the mean concentration and 7,608 m³/s for the dominant discharge.
The effective discharge computation is shown in Fig. 13. The resulting value is 9,582 m³/s, which corresponds to the maximum value of the curve of weighted contributions (right panel) obtained from the product of the flow frequency curve (left panel) and the sediment rating curve (middle panel). Other local maxima may be seen in the curve. These represent the discharges responsible for carrying large sediment volumes. As is often the case, the result obtained has two distinctive peaks, indicating that a frequent discharge carrying a relatively small sediment load for a long time is almost as effective as an infrequent discharge carrying a large amount of sediment over a shorter period of time. Using the rating curve in Fig. 8, the suspended sediment concentration associated with the effective discharge was obtained. The resulting value was 0.441 g/L.

Key Findings and Discussion
What is the source of the sediment responsible for siltation inside the harbor?
Origin based on grain size distributions and sedimentation patterns The sediment size analyses from the river and the harbor are summarized in Fig. 14; median D50 values are shown in Table 8. Harbor bed sediments are slightly coarser than the material that is carried in suspension by the Upper Mississippi River at St. Louis, MO but are significantly finer than the material in the bed of the river, suggesting that the sediment source is likely to be the suspended sediment in the river. The sedimentation patterns inside the harbor also shed light on the origin of the sediment. As shown in Fig. 2, siltation blankets the entire bed of the harbor. The relatively uniform thickness of the deposited sediment observed in the March and June bathymetries is due to a combination of two factors: the fine-grained nature of the deposited sediment, and barge traffic (approximately 20 barges per day), which can under some conditions cause resuspension and redistribution due to propeller wash (Garcia et al. 1999). Although coarser materials were found close to the entrance, all sediments were significantly finer than the Upper Mississippi River bed material.

Suspended sediment dynamics within the harbor
The sediment that enters the harbor in suspension is deposited first on the perimeter of the harbor where the flow velocities and shear stresses, even in the presence of barge traffic, approach zero. Sediment deposits preferentially along these zones and then builds up uniformly from the edges towards the middle of the harbor. The siltation patterns shown in Fig. 2 show some zones that are lower in elevation in the south section close to the entrance. These areas have likely been scoured due to barge traffic going in and out of the harbor.
The settling velocities determined in the experiments (Fig. 5) and shown in Fig. 6 range between 1e-6 to 1e-3 m/s, with the largest values associated with larger suspended sediment concentrations at which flocculation occurs. Although the concentrations in the Mississippi River rarely exceed 2g/L (Fig. 8), it is possible that concentrations may exceed this value inside the harbor as the sediment settles to the bottom. This is most likely to prevail during periods when the harbor is not operating at full capacity. The presence of a bar-like feature on the east side of the harbor on the July 29 th bathymetry is also thought to be related to barge traffic redistribution of sediments, since most of the barge traffic occurs through the southern part of the harbor and towards the west and north west sections.
When are the sediments most likely to be deposited in the harbor?
Harbor siltation volumes and rates are shown in Fig. 15. The black solid line corresponds with the volumes of sediment above the design conditions of the harbor. The values are divided by 20 so as to plot this variable using the same axis limits as the flow discharge, and to clearly present the salient trends. In the three cases where the volume of sediment in the harbor increases, the period corresponds to late February or early March to late July or early August. (Decreases are caused almost solely by dredging.) This timeframe corresponds to the spring and early summer months; siltation rates within this period can be as high as 1.2 cm/m²/day, as indicated by the red dashed line.

Applicability of the dominant and effective discharge concepts
Typically, the dominant and effective discharge concepts are not meant to be used in rivers where the majority of the material transported corresponds to silt and clay sizes (i.e. wash load). The main reason for this is that wash load does not correlate with flow discharge and therefore, as long as the sediment is available, the river will transport it regardless of the flow magnitude. Fig. 14 shows that more than 80% of the material traveling in suspension in the Upper Mississippi River corresponds to wash load. However, Fig. 8 and Fig. 11 show that wash load in the Mississippi River, as defined using e.g. the 62.5 m cutoff criterion (River Research Council, 2007), does indeed positively correlate with discharge to a surprising degree. The trends shown by both variables in Fig. 11 are remarkably similar, and more than 80% of the suspended sediment concentration data shown in Fig. 8 lies within envelopes of 0.5-2.0 times the value estimated with the sediment rating curve. A possible explanation for this behavior is given below.
During late February and early March, snowmelt takes place and river flows increase. At the same time, fine sediment from bare agricultural land is carried by runoff into the river and transported as wash load. This phenomenon is sustained throughout the growing season, and is enhanced by rainfall in the spring and early summer. Once the crops are established and precipitation diminishes (late summer), fine sediment availability is reduced and both the flows and suspended sediment concentrations in the river return to their base flow patterns. The mean annual hydrograph shown in Fig. 11 reflects these processes.
Snowmelt followed by spring and early summer precipitation contribute to the flow magnitude and the availability of sediment due to bare agricultural land in the Upper Mississippi River basin, thus creating conditions in which fine sediment availability matches the period of high flows. High flows do not necessarily cause larger sediment transport, but are correlated due to the characteristics of the river basin. The dominant and effective discharge concepts may be applied in this and other river basins where sediment availability matches the period of high flows even though the relation between the two variables is not strictly causal.
How does siltation relate to the hydraulic conditions in the river? Table 9 summarizes the results obtained for the characteristic discharges, the number of days for which they are exceeded and the associated suspended sediment concentrations. Comparison of the characteristic discharges with the mean annual hydrograph and mean annual suspended sediment concentrations shown in Fig. 11 suggest that the Mississippi River carries larger sediment volumes between the end of February and early August than otherwise.
The dominant discharge is exceeded for 120 days between mid-March and mid-July, and the effective discharge is exceeded only for a few days in April and all of May. Siltation volumes and siltation rates are shown in Table 1 and Fig. 15; they are greatest in periods including these months. Although bathymetric survey dates allow assessment of the silting process over the period between February and August, lack of data for the months of April and May impede determining if harbor siltation occurs mostly during early or late spring, summer or both.  Garcia et al. 1999Garcia et al. , 1998. Barge traffic in and out of the harbor plays an important role in sediment resuspension. The harbor is directly open to the Mississippi River, but has no through-flow discharge and thus acts as a sediment trap. Towboats and barges that enter for loading and unloading operations resuspend the sediment in the harbor, but even with the small settling velocities measured in the laboratory and reported in Fig. 6, such resuspension does not seem to contribute substantially toward keeping sediment from settling inside the harbor. As shown in Table 1 and Fig. 15, between the months of July and December of 2009, the excess volume of sediment in the harbor decreased and no dredging efforts took place. This suggests that in those months in which Upper Mississippi River flow discharge and suspended sediment concentrations return to base levels, sediment resuspended by towboats and barges may leave the harbor. This observed decrease, however, corresponds to only an insignificant amount of sediment compared to the amount that comes into the harbor during the spring and summer months.

Conclusions
Flow and sediments in the Upper Mississippi River were characterized with information available at the USGS gaging station in St. Louis, MO. The most relevant results of our analysis are as follows.
1. The correlation between wash load and flow discharge in the Upper Mississippi River is due to the characteristics of the basin, namely, snowmelt followed by spring and early summer precipitation over bare agricultural land that create conditions in which fine sediment availability matches the period of high flows.                  Flocculation settling Hindered Settling Consolidation C < C1 C1 < C < C2 C2 < C < C3 C3 < C ws = wsf ws = ws(C) ws = ws(C) ws 0   Within that time period, only 3 (12, 30) values exceeded 4g/L (3 g/L, 2.5 g/L) corresponding to 0.03% (0.11%, 0.27%) of the data.