A characterization of side channel development

Side channels are commonly constructed to reduce the flood risk or to increase the ecological value of a river. Such artificial side channels generally aggrade. We categorize the development of side channels based on the sediment that is deposited in these channels. Based on this categorization, we determine the main mechanisms that affect their development, and we propose an initial framework on how to predict the long‐term development of side channels. The results can be used to design, operate, and maintain side channel systems.


INTRODUCTION
Side channels are interventions that are important in reducing the flood risk and increasing the biodiversity of a river. It is well known that artificial side channels generally aggrade (e.g., Formann, Habersack, & Schober, 2007;Riquier, Piégay, Lamouroux, & Vaudor, 2017;Simons et al., 2001;Van Denderen, Schielen, Westerhof, Quartel, & Hulscher, 2019). Within one river branch, side channels aggrade at various rates and with various types of sediment (Riquier, Piégay, & Michalkova, 2015;Riquier et al., 2017;. There is a need for a better understanding of the mechanisms and the processes that result in the aggradation of such side channels. Our objective is to propose a categorization of side channel development that can aid in determining the mechanisms that affect the side channel development. The development of a side channel is a result of an imbalance between the sediment supply and its transport capacity. The sediment supply can be a function of several mechanisms such as, for example, a transverse bed slope (Bolla Pittaluga, Repetto, & Tubino, 2003) or spiral flow at the bifurcation (Bulle, 1926;Kleinhans, Cohen, Hoekstra, & IJmker, 2011). How these mechanisms affect the sediment supply to the downstream channel depends on the type of sediment deposited in the side channel and the way the sediment supplied to the side channel is transported in the main channel. The transverse bed slope effect is generally smaller for smaller grain sizes (Parker & Andrews, 1985), and suspended sediment transport is much less affected by bed level differences compared with bed load transport (Szupiany et al., 2012). Spiral flow changes the direction of the sediment transport that occurs close to the river bed and its effect on the sediment partitioning reduces with reducing grain size (Dutta & García, 2018;Kleinhans, de Haas, Lavooi, & Makaske, 2012). The mechanisms that determine the sediment supply to the side channel are therefore a function of the sediment type and its type of transport. We distinguish three types of sediment transport: bed load transport consisting mainly of gravel and sand, suspended bed-material load transport consisting mainly of sand, and wash load transport consisting mainly of silt and clay (e.g., Church, 2006). The sediment transported as bed load or suspended bed-material load can be found on the river bed. Wash load is defined as the portion of the transported sediment that does not affect the morphological change of the river (Einstein & Johnson, 1950) and thereby, its slope or width (Paola, 2001).
We propose a categorization of side channels that is based on the sediment deposited in the side channel in relation to the transport type of the supplied sediment from the main channel. Using the categorization, we can determine which mechanisms are important in the development of a given side channel. The categorization can, therefore, support the design, operation, and maintenance of a side channel system. We apply the categorization to a one-dimensional model that was previously used to reproduce the development of side channel systems (Van Denderen, Schielen, Blom, Hulscher, & Kleinhans, 2018) and river bifurcations in general (e.g., Kleinhans et al., 2011). We use the model to estimate the temporal scale of the side channel development for each of the categories as a function of the width/depth ratio of the side channel and its length relative to the main channel, which are both common design parameters of side channels. The temporal scale is the most interesting parameter for river managers that need to design a side channel system, because side channels are generally expected to close (Van Denderen, 2019).

CONCEPTUAL CATEGORIZATION OF SIDE CHANNELS
We propose a conceptual categorization of side channel development. We present three main categories of side channels based on the classification of the sediment transport in the main channel: predominantly (a) bed load supplied side channels, (b) suspended bed-material load supplied side channels, and (c) wash load supplied side channels ( Figure 1). Note that the sediment transport classification represents the sediment that is predominantly deposited in the side channel and that also other types of sediment can be supplied. For example, wash load is also supplied to a bed load supplied channel, but is not deposited due to, for example, a high bed shear stress. The category, to which a side channel belongs, can change over time due to an increasing bed level (Makaske, Smith, & Berendsen, 2002; or due to changes in the flow conditions (Riquier et al., 2015). Each category corresponds with mechanisms that affect the sediment supply to a side channel or the transport capacity in a side channel. Differences between the sediment supply and the transport capacity lead to morphodynamic changes. In the following subsections, we explain the characteristics and the corresponding mechanisms for each category.   (Dieras et al., 2013). The deposition of gravel occurred over the length of the channel (Dieras et al., 2013). This is in contrast to a strongly curved meander in the Ain River that shows the formation of a plug bar (Figure 2c). The meander is much longer than the cutoff. The sediment supply to the meander is large due to the small bed level difference and, due to a limited sediment mobility in the channel, a plug bar formed (Dieras et al., 2013).

Suspended bed-material load supplied side channels
Suspended bed-material load supplied side channels are filled with sediment that can be found on the bed of the main channel and that is partly transported in suspension (Figure 1). Suspended bed-material load consists primarily of sand (Church, 2006). The sand is mainly

Wash load supplied side channels
Wash load supplied side channels are filled with sediment that is transported as wash load in the main channel and that is generally not found on the bed of the main channel (Figure 1). Secondary channels filled with wash load deposits are often blocked at the bifurcation with a logjam or a plug bar (e.g., Makaske et al., 2002)  increasing discharge from the main channel (Middelkoop & Asselman, 1998), and therefore, the distance between the upstream end of the side channel and the main channel becomes important in determining the sediment supply. Inside the side channel, the flow velocity is sufficiently low such that wash load is deposited. Therefore, the deposition processes within wash load supplied side channels are expected to be similar to deposition processes in floodplains. This means that nonequilibrium sediment transport is important (Asselman & Van Wijngaarden, 2002), and the sediment deposition is lagged in time and space compared with a change in transport capacity as a function of the settling velocity.
The Mackey bend in the Wabash River is an example of a wash load supplied side channel (Figure 2b). The large meander was cutoff twice, and the East channel is now the main channel. Measurements showed that in the other two channels, mainly silt and some clay is deposited (data from USACE and Zinger, 2016). The discharge in the channels is limited allowing fines to settle. Wash load supplied side channels were also found in the Columbia River in Canada (Makaske et al., 2002) and in the Rhône River in France (Riquier et al., 2015).

IMPLICATIONS FOR THE MODELING OF SIDE CHANNEL SYSTEMS
Each category of side channels (Figure 1) develops differently. Here, we propose a preliminary method to estimate the time scale of side channel development for each category. The method is based on a previously published 1D model (e.g., Kleinhans et al., 2011). The development of bed load and suspended bed-material load supplied side channels can be estimated using a simple backwater model. Van Denderen et al. (2018) apply such a model to bed load supplied side channel systems. The transport capacity in such channels is best represented using a transport relation that includes the initiation of Transport capacity e.g., Meyer-Peter and Müller (1948) e.g., Engelund and Hansen (1967) e.g., Asselman and Van Wijngaarden (2002) Grain size Similar to main channel Based on grain size measurements -

Roughness
Similar to main channel Based on ripple/dune height in side channel Based on ripple/dune height or vegetation in side channel motion (Bolla Pittaluga et al., 2003;Bolla Pittaluga, Coco, & Kleinhans, 2015). The roughness and grain size in the side channel are assumed similar to the main channel (Table 1). For suspended bed-material load supplied channels, the sediment sorting at the bifurcation is significant, and in order to compute the transport capacity in the side channel, a smaller grain size and bed roughness should be taken into account compared with the main channel (Table 1). In addition, we use a sediment transport relation that includes both bed load and suspended bed-material load transport (Bolla Pittaluga et al., 2015). The sediment supply is assumed to be equal to the discharge partitioning, which is reasonable because the sediment is transported near the bed in As an example, we apply our modeling approach to the three categories for various initial geometries. We assume that the side channels are connected to the main channel during bankfull conditions. This allows for the usage of a simple one-dimensional model (Supporting Information). We vary the initial width/depth ratio and the length of the side channel while keeping the initial discharge partitioning constant. The initial discharge is assumed 10%, 1%, and 0.1% for a bed load, a suspended bed-material load and a wash load supplied side channel, respectively. Using these conditions, we compute the time scale needed to reach an equilibrium state where the discharge in the side channel does not change more than 0.01% of the upstream dis-

CONCLUDING REMARKS
We present a categorization for side channel systems such that the main mechanisms for its development are easily identified (Figure 1).
In addition, we propose an initial framework to easily estimate the side channel development ( Table 1) that will help to optimize the design, operation and maintenance of side channels. For a combined assessment of planimetric forcing and mixtures of sediment, it would be essential to run a two or three-dimensional flow and morphodynamic model with multiple grain-size classes.