River Ice Effects on Sediment Transport and Channel Morphology—Progress and Research Needs

Abstract

Sediment transport in alluvial channels has a long history of intensive research. River ice could affect sediment transport and channel morphology through the impact of various dynamic and thermal ice processes. However, studies on sediment transport under the influence of ice have been minimal until recent years. This phenomenon was partially due to the complicated interactions between ice, flow, and sediment dynamics, which require a good understanding of the river ice process, in addition to the difficult field data collection conditions. This paper reviews the progress and needs of river ice-related research on sediment transport and channel morphology, including the influence of ice cover and surface ice runs on sediment transport, the effects of frazil ice, anchor ice, and bank stability with freeze-thaw effects.

1. Introduction

Sediment transport in alluvial channels, a major area in river hydraulics, has a long history of intensive research. In cold regions, river ice affects sediment transport and channel morphology through flow and thermal-ice processes. The resulting changes in channel morphology will change the ice conditions in turn. Additionally, ice-related sediment transport has significant implications for river ecosystems, contaminant re-entrainment, and natural habitats through processes such as bed scour by ice jams, anchor ice formation and release, and bank erosion [1,2]. Studies on sediment transport under the influence of ice have been very limited until recent years [3,4]. This was partially due to the complicated interactions between ice, flow, and sediment dynamics, which require a good understanding of the river ice process. This paper reviews the progress of ice-related research on sediment transport and channel morphology, including the impacts of ice cover, surface ice run, frazil ice, anchor ice, and bank stability with freeze–thaw effects. In addition to these in-channel processes, the changing supply from the watershed due to the changing catchment runoff and erosion conditions needs to be considered. Numerous examples of wintertime sediment and morphological changes have been reported [5,6,7,8]. Field data collection and analysis is an important component of advancing understanding of the ice-related sediment and channel morphological processes. In situ monitoring and data collection of sediment and morphological conditions over a river reach under ice-covered conditions is difficult and time-consuming (e.g., [9]), especially during freeze-up and break-up with rapidly changing ice conditions. Advancing the current state-of-the-art techniques for observing ice-related sediment and morphological processes could benefit by coupling with the technologies of monitoring river ice and bank erosion conditions [10,11]. Numerical modeling has been widely used to complement field and laboratory studies to develop additional insights into understanding morphological characteristics and applications to engineering problems in open water conditions [12]. For the ice season, such a model requires coupling the sediment models with river ice models to enable simulation of the interaction between ice and fluvial dynamics.

2. Sediment Transport Under Ice Cover

Understanding the bed and suspended sediment transport processes is the basis for studying the sediment and channel morphology. Surface ice run or ice cover will shift a portion of the bed shear stress to the surface ice [13]. Additionally, for a given flow rate, the flow depth will increase due to the added flow resistance from surface ice, accompanied by a decrease in flow velocity [13,14]. Hence, the bed and suspended sediment transport capacities will be reduced for the same discharge. The effect of the shear stress redistribution depends on the cover roughness, which varies continuously over the ice season, and the formation and release of ice jams [15]. The effect of ice will also influence the flow characteristics and sediment transport in meandering river channels [16,17].

2.1. Bed Load Transport

Bed load transport for open water conditions can be calculated with the Einstein–Brown formula [18,19], the Meyer–Peter–Muller equation [20], or the van Rijn method [21]. Knack and Shen [22] analyzed available experimental data and showed that the bed load transport in ice-covered channels can be described by the relationship for the equivalent free-surface flow if the flow strength is expressed in terms of the bed shear stress. This relationship could be expressed as follows:

$$\phi =25 {\left(\theta -0.041\right)}^{2.1}$$

(1)

in which, the dimensionless bed load capacity $\phi =q_b/(\mathsf{\Delta }g{D}^3{)}^{{\displaystyle \frac{1}{2}}}$; dimensionless flow strength $\theta ={\tau }_b/\left(\mathsf{\Delta }\rho gD\right)$; ${\tau }_b$ = bed shear stress; $g$ = gravitational acceleration; $q_b=$ volumetric rate of sediment transport per unit width; $\mathsf{\Delta }={\displaystyle \frac{{\rho }_s}{\rho }}-1; \rho \mathrm{a}\mathrm{n}\mathrm{d} {\rho }_s=\mathrm{d}\mathrm{e}\mathrm{n}\mathrm{s}\mathrm{i}\mathrm{t}\mathrm{i}\mathrm{e}\mathrm{s} \mathrm{o}\mathrm{f} \mathrm{w}\mathrm{a}\mathrm{t}\mathrm{e}\mathrm{r} \mathrm{a}\mathrm{n}\mathrm{d} \mathrm{s}\mathrm{e}\mathrm{d}\mathrm{i}\mathrm{m}\mathrm{e}\mathrm{n}\mathrm{t}$; and D = particle diameter.

2.2. Suspended Load

Changes in the vertical distributions of velocity, shear stress, and diffusivity have important implications for suspended sediment transport. Sayre and Song [23] performed flume experiments to investigate the effect of a floating ice cover on the suspended sediment transport by extending the Rouse equation to ice-covered conditions. Wang et al. [24] developed an analytical formulation for the vertical sediment concentration distribution and suspended sediment discharge for ice-covered channels and validated it with laboratory data. In natural river channels, the transverse mixing effect due to transverse variations in flow depth and velocity should be considered [25,26].

2.3. Bed Roughness

Smith and Ettema [27] developed a relationship for the ratio of shear stresses between the ice cover and bed, τ_ice/τ_bed. Knack and Shen [28] extended this relationship with additional flume data to a relationship for the ratio between Manning’s coefficients of the ice cover and the bed:

$${\displaystyle \frac{n_{ice}}{n_{bed}}}=1.04{\left(\eta {\displaystyle \frac{D_{50}}{k_{ice}}}\right)}^{-0.15}$$

(2)

where n_ice and n_bed are Manning’s coefficients of ice cover and bed, respectively; η = θ/θ_c; relative flow intensity; ${\theta }_c=\mathrm{c}\mathrm{r}\mathrm{i}\mathrm{t}\mathrm{i}\mathrm{c}\mathrm{a}\mathrm{l} \mathrm{d}\mathrm{i}\mathrm{m}\mathrm{e}\mathrm{n}\mathrm{s}\mathrm{i}\mathrm{o}\mathrm{n}\mathrm{l}\mathrm{e}\mathrm{s}\mathrm{s} \mathrm{f}\mathrm{l}\mathrm{o}\mathrm{w} \mathrm{s}\mathrm{t}\mathrm{r}\mathrm{e}\mathrm{n}\mathrm{g}\mathrm{t}\mathrm{h} \mathrm{o}\mathrm{f} \mathrm{i}\mathrm{n}\mathrm{c}\mathrm{i}\mathrm{p}\mathrm{i}\mathrm{e}\mathrm{n}\mathrm{t} \mathrm{m}\mathrm{o}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}$; D₅₀ = median sediment size; and k_ice = equivalent roughness height of the undersurface of the ice cover.

2.4. Water Temperature Effect

Ice formation results from the drop in water temperature to near freezing or supercooling. Field and laboratory studies [29,30,31] showed the general trend of sediment transport increases with water temperature decrease. A variation in water temperature changes the flow viscosity, which affects the thickness of the laminar sublayer on the bed to adjust the relative roughness of the bed, the force acting on the bed sediment particle, and the sheltering effect on the particle on the bed. The combined effect could enhance or weaken the bed load movement [29]. The change in viscosity will also affect the fall velocity of the sediment particle. The suspended sediment concentration profile will be more uniform but influenced by the near-bed concentration related to the bed load.

2.5. Fine Sediment Transport

For rivers on alluvial plains, the sediment load with exceptionally high concentration will be predominantly suspended fine sediment in the wash load size range. The total transport capacity can be expressed as follows [32,33]:

$$S_c=k{\left({\displaystyle \frac{U^3}{gh\omega }}\right)}^m$$

(3)

in which S_c = cross-section averaged sediment concentration in mass per unit volume, i.e., ${\rho }_s{C}_s$, kg/m³; C_s = sediment concentration; U = flow velocity; $\omega$ = fall velocity; h = flow depth; and k and m = empirical coefficients. Yang et al. [34] developed a one-dimensional model for a 276 km reach of the Lower Yellow River, with an average slope of 10⁻⁴ and channel width varying from 500 to 2000 m. They obtained empirical relationships for k and m. Both vary with $\left({\displaystyle \frac{U^3}{gh\omega }}\right)$, but are essentially the same for open water and ice-covered conditions. They obtained a relationship between the ice-covered and open-water transport capacities, S^c and S^o as:

$${\displaystyle \frac{S^c}{S^o}}={\left[0.574{\left({\displaystyle \frac{n_o}{n_c}}\right)}^{1.2}\right]}^m$$

(4)

in which n_o and n_c are the open-water bed Manning’s coefficient and the composite Manning’s coefficient for ice-covered conditions. The study showed that the sediment discharge can be reduced by 30 to 60 percent due to the presence of the ice cover.

3. In-Channel Ice-Sediment Interactions

Several in-channel ice-sediment interaction phenomena have been observed, including ice cover, surface ice jam, and hanging dam effects on sediment transport and bed changes, frazil ice entrainment, and anchor ice rafting. These phenomena will be discussed in this section.

3.1. Ice Jam Evolution and Bed Change

Ice jams usually form during a dynamic breakup of ice cover with a rapid flow increase in mid-winter or spring. However, freeze-up jams can also form during ice cover formation. Ice jam formation and release could cause significant changes in channel morphology through erosion and deposition of bed sediment. Numerical modeling has been applied to study the possible impact of ice covers on fluvial processes through simulated hydraulic conditions [35,36,37,38]. A coupled sediment and river ice model is needed to simulate the interaction between ice and fluvial dynamics. Knack and Shen [28] developed a two-dimensional sediment transport and bed change model coupled with a hydro-thermal-ice dynamics model [39]. Figure 1 demonstrates the main feature of the jam-sediment interaction in a straight, rectangular channel with a uniform bed sediment particle diameter of 2 mm, simulated with a coupled ice-sediment model [28]. The channel is 4000 m long and 60 m wide with an initial bed slope of 0.0005. The boundary conditions include a constant water discharge of 73.96 m³/s at the upstream boundary and the corresponding normal depth downstream for ice-covered or open-water conditions. Figure 1 shows the variation of longitudinal elevation profiles of ice cover and jam thickness, water level, and channel bed at different times, along with the width-averaged bed load discharge q_s,c m²/s. The channel is initially covered with an ice cover with a uniform thickness of 0.15 m and Manning’s coefficient of 0.02. The equilibrium bed load transport rate is applied at the upstream boundary. The ice cover breakup in the upstream 0~2 km portion is assumed to occur along with continuous ice supply from the upstream boundary for 12 h to enable the full development of the jam. An ice jam was initiated at the cover front located at 2 km. The ice jam constricts the flow cross-section and increases the local water velocity and sediment transport capacity under the jam. The backwater from the jam formation increases the upstream water depth and decreases the water velocity and sediment transport rate. The increased jam roughness and flow depth in the backwater zone further reduce the water velocity. The weakened flow strength leads to a decrease in bed shear stress and the sediment transport rate. The increasing difference between the sediment transport rates in the backwater zone and the jammed reach leads to the scour of the bed sediment under the jam. The sediment deposition immediately downstream of the jam is due to the sediment transport capacity being much lower than the sediment supply from the eroded sediment at the jam toe. The deposited sediment slowly migrates downstream. The scour of sediment at the ice jam toe enhanced the further thickening of the ice jam until the sediment and ice reached a quasi-equilibrium state while the ice jam and scour hole no longer grew. The downstream ice cover of 2~4 km is assumed to break up at hour 96, leading to the ice jam release. The moving surface ice shifts part of the resistance from the ice back to the riverbed, increasing the sediment transport capacity. The resulting higher velocity and sediment transport rate from the release of channel storage under the ice cover and jam erode the deposited sediment during the formation and growth of the ice jam. The water velocity returns to near-uniform flow conditions after the release of the channel storage. The deposited sediment gradually migrates downstream, and the scour hole under the jam toe gradually recovers. Figure 2 compares the modeled ice jam, bed, and water surface profiles for the cases with and without sediment transport and bed change. The jam thickness with the bed change is larger because the bed scour allows the further thickening of the ice jam. The bed scour lowers the ice jam and water surface elevations. As a result, the upstream water level due to the backwater effect is lower when the sediment transport and bed change are ignored. The channel geometry in natural rivers will change with bank erosion during the jam event. Ice jams could have a long-lasting effect on the channel morphology and will not likely return to the pre-jam condition as in the simulated result.

3.2. Frazil Jam/Hanging Dam

Frazil jams, commonly known as hanging dams, are thick ice accumulations under an existing ice cover downstream of a rapid open water reach. The ice in the hanging dam consists mainly of the frazil granules, with some ice floes, produced in the open water reach upstream of the cover throughout the winter and entrained at the cover front [40,41]. Figure 3 shows an example of the cross-sectional evolution of a hanging dam from frazil granule accumulation under the ice cover and the channel bed observed in the Hequ reach of the Yellow River downstream of a rapid reach of over 100 km [40]. The ice mass in the hanging dam accumulates and transports as a cover load, similar to the bed load sediment transport [42]. The size of a hanging dam grows and decays with the ice supply from upstream. It grows from the initial cover formation to mid-winter, then decays when the ice production in the upstream open water reach declines.

3.3. Ice Cover Effect on Alternate Bar Development

Alternative bar formation is closely related to the development of river meanders, riffles, and pools. Flume studies have been conducted to characterize the principal mechanism of alternate bar evolution [43,44]. Crosato et al. [45] conducted experimental and numerical studies on developing alternate bars by the transverse flow produced by a plate barrier extending over half of the channel width at the upstream end. Huang et al. [46] simulated alternate bar formation and development in a straight rectangular channel. The simulation results showed that the alternate bar development and migration are much slower under ice-covered conditions. The wavelength in the ice-covered case is about half of the open-water case. These differences are caused by the reduced flow strength for bed load transport by the ice cover.

3.4. Sediment Inclusion and Transport by Ice

Sediment enrichment of ice can result from several sediment inclusion processes during the freeze-up. These processes include sediment inclusion by suspended frazil, the attachment of bed sediment on ice floes in contact with the channel bottom, and anchor ice rafting. The sediment in ice will be transported by the freeze-up ice run, incorporated into the seasonal ice cover, and released with the melting of the ice cover or the breakup ice run in the spring. Figure 4 shows ice samples from the Yellow River with sediment containments from various inclusion processes [47]. The sediment included in the ice cover will not only contribute to the sediment transport during the melting and break up in the spring but also affect the material properties of the cover, especially through its influence on cover deterioration [48]. This changing material property will also affect the application of the ground-penetrating radar, a remote sensing field observation technique for surveying the ice thickness and bed profiles simultaneously along the river [49,50,51], a technique that could be used for estimating the bedload transport rate [52].

3.4.1. Frazil Scavenging

The entrainment of sediment by frazil suspension or scavenging by frazil flocs has been studied [53,54,55]. This mechanism may not produce a significant amount of sediment transport. However, the possible influence of suspended sediment transport and frazil suspension on the mixing process has yet to be examined.

3.4.2. Bed Sediment Attachment

Another sediment inclusion process that has been observed is the inclusion of bed sediment by ice accumulations along the banks during the freeze-up ice run or the attachment of bed sediment to ice floes resting on the bed in the shallow water area when water level drops during low flow conditions. Sediment in the bank ice accumulations will likely be released with the melting ice during the breakup ice run. The ice floes resting on the bed will rise and be entrained by the flow when the stage rises, either due to the backwater of the progressing ice cover from downstream or the increase in discharge. Figure 5 shows sediment-enriched ice in the Yellow River during freeze-up.

3.4.3. Anchor Ice

Anchor ice is the most studied sediment inclusion process. Sediment inclusion by rafted anchor ice formed in open water stretches, Figure 6, could be transported along the river, incorporated into the seasonal ice cover, and released in spring. Gravel, cobble, and boulders have been observed in rafted anchor ice. Localized deposition of large bed particles released from anchor ice could cause important changes in the channel geometry. Li [56] observed gravel island formation in rivers and the destruction of intake structures by a 4-ton rock released from drifting ice in the Heilongjiang River. In addition to the transport by rafted anchor ice, anchor ice can affect sediment transport by affecting the flow condition in the river [57]. Figure 7 shows an example of anchor ice in the River Tornionjoki, along the Finnish-Swedish border, causing flow distribution and bed resistance changes.

Although many laboratory and field studies have been conducted on anchor ice development and associated bed material rafting [58,59,60,61,62,63,64], the formation and release mechanisms are still unclear. Hammar et al. [58] pointed out the need for a better understanding of the flow and thermal processes in the hyporheic zone, including the thermal-ice process influenced by solar radiation, the bed heat flux [65,66], and the bonding strength between anchor ice and bed material. Research on the near-bed flow and turbulence conditions [67,68], which affect the deposition of frazil, in-situ growth, decay, and release of anchor ice, is needed.

4. Sediment Load in Glacial-Fed Rivers

With the impact of climate change, the sediment load in streams in glacierized drainage basins has been increased by glacial ice ablation and runoff [69,70,71]. Hodgkins et al. [69] showed that the sediment influx to glacier-fed rivers occurs during discrete episodes of enhanced meltwater discharge. The in-channel sediment storage varies with the influx of sediment and water. This influx will cease during the winter ice-covered period. The fine sediment input to the ocean from these streams could affect the seawater quality. Wada et al. [72] studied the discharge and sediment load of the Tana River in the subarctic Yukon River basin and showed that the glacier-melt discharge accounted for 26–57% of the discharge, while the sediment load from the glacierized regions accounted for 76–94% of its sediment load. The remaining contribution (6–24%) of the sediment load was probably due to the fluvial resuspension of glacial sediment deposited previously in the river channels. Diodato et al. [73] developed a model to predict climate-induced sediment load based on seasonal precipitation and temperature and applied it to a catchment in western Norway. Zajaczkowski and Włodarska-Kowalczuk [74] conducted a field study on the sediment dynamics in a glacier-fed river estuary with detailed measurements. They showed the patterns of sediment storage and sedimentary dynamics in a glacier-fed river estuary, which are driven by gravity flows and turbidity currents rather than by the patterns of the vertical sedimentation of suspensions in the water column. Although these studies were based on glacierized drainage basins in Arctic and subarctic basins, the same phenomenon will impact rivers in glacierized basins in the Qinghai-Tibet Plateau.

5. Bank Erosion and Failure

It has been shown that ice cover can cause larger bank retreats and bed depositions with slower channel cross-sectional change due to the reduced bed shear stress and the increased flow depth in non-cohesive alluvial channels [75]. However, bank erosion and failure during the spring breakup are major factors affecting channel morphology. A few field studies exist on the ice effect on bank erosion. Uunila and Church [76] examined the effects of river ice on bank morphology and the downstream reach of the Peace River. They showed that shoving and gouging by moving ice have created distinctive local landforms and features along the riverbanks and nearshore channel bed, but there is no overall geometry change of the channel. However, sedimentation along the channel margin has created an inner shelf one to two meters below the matured floodplain level. Vandermause et al. [77] conducted field studies with aerial photography records on bank erosion caused by breakup events in the Susitna River, Alaska, and indicated the high breakup discharge and the dynamics of breakup ice runs are the dominant factors. They also mentioned the protection of abrading by vegetation root mats and shear walls of ice rubble, similar to the one shown in Figure 8. In addition to the breakup ice run effects, the ice load on banks, as shown in Figure 9, should be considered. Prowse [78] and Beltaos and Burrell [79] discussed the effects of river-ice breakup on sediment transport, emphasizing suspended-sediment concentration increase due to bank erosion resulting from changing flow conditions related to ice jam formation and release and the impact of the surface ice run abrasion and gouging on banks. It should be noted that the changes in sediment discharge, especially the suspended load, are not only controlled by the changes in the river channel itself but also by sediment supply from the surrounding catchment area [80,81,82,83]. This is especially true for rivers in glacierized catchments, as discussed in Section 4.

For banks affected by freeze-thaw actions, the strength of the layer of soil thawed from the heaved ice lens will be reduced from its original strength. In cold regions with seasonally frozen ground and permafrost regions, progressive thawing of riverbanks in contact with flowing water results in heat transfer between water and ice, producing easily removable sediments that can be destabilized and removed by terrestrial and fluvial processes. Anderson et al. [84] showed that ice in the soil voids could weaken even in non-saturated soil. Moreover, as the water draining from the thawed soils reaches the still-frozen ice layer, the water will flow downslope, which may cause a seepage flow and piping to the river [85,86]. Ferrick and Gatto [87] conducted a laboratory study on the freeze-thaw cycle on erosion in small, straight, rectangular channels constructed with the silt of different moisture contents. Yumoto et al. [88] monitored riverbank profile and freeze-thaw processes in a small mountain stream in Japan. They observed freeze and thaw-triggered notch development and mass failure, causing significant subaerial erosion and bank retreats. These studies provided the basis for further research on physical and numerical models to gain better insights into bank erosion and stability processes. Li et al. [89] extended the Darby and Thorne [90] bank stability model with a limit equilibrium bank stability model [91], considering the infiltration water pressure effect and seasonal variations of the water levels. They showed that the freeze-thaw of the bank material, seepage pressure, and bank slope are the key factors affecting bank stability, and the slope stability safety factor increases with the river stage but decreases with the groundwater level. This study confirmed the findings of Lotsari et al. [92] that bank erosion in sub-arctic meandering rivers occurs during the low flow period after the spring flood and the effect of groundwater seepage. Using time-lapse analysis of satellite imagery, Ielpi et al. [93] showed that the lateral migration of large Arctic sinuous rivers has decreased by about 20% over the last half-century, impacting the residence time of sediment and organic matter in floodplains. Douglas et al. [94] showed that the sediment entrainment and riverbank armoring of the failed slump block from thawing effects limit permafrost bank erosion. This indicates the need to combine the in-channel sediment transport with a slump failure model.

Modeling Thermal-Ice Processes Related to Bank Failure

Thermal-ice process modeling is an essential element in studying bank failure and erosion in seasonal and permafrost rivers. Zheng et al. [95] developed a one-dimensional model by adding river water temperature to the control volume permafrost model [96,97] to model the thermal condition of a shallow permafrost riverbank. Pan and Shen [75] developed a two-dimensional numerical model to simulate the seasonal freeze-thaw process in riverbanks with changing ice cover conditions from fall to spring. Douglas and Lamb [98] developed a numerical model for permafrost thaw, thawed sediment entrainment, and heat transfer within the thawed and frozen portions of the riverbank. They showed that thin layers of thawed sediment formed along banks insulate permafrost and dramatically slow thaw and erosion rates. When the thawed sediment was unstable and failed past a threshold thickness, the thawed layer remained thin, and bank erosion was sensitive to warm water temperature. These studies advanced the understanding of the frost effect on the bank failure processes. Additional factors, such as groundwater seepage and shear stress at the soil-ice interface, need to be considered [99,100,101,102].

6. Conclusions

This paper reviews the progress of research on sediment transport, bank stability, and channel morphology affected by ice cover, surface ice run, frazil ice, and anchor ice. Ice formation in river channels will alter the channel hydraulics and sediment transport capacity. Developing the sediment transport capacity theories on the bed and suspended loads was the first step in extending the non-ice-related knowledge of sediment transport to ice conditions, followed by developing physically-based numerical models. Ice cover and ice jams could affect the bed change and bank erosion. In addition to the ice cover, other sediment-related phenomena during the freeze-up, including anchor ice rafting and sediment inclusion processes such as sediment entrainment by frazil suspension and bed sediment enrichment of ice floes deposited on the bed, are discussed. Bank erosion and failure during the breakup are other important sediment transport phenomena that impact channel morphology. Bank erosion and failure include mechanical erosion caused by breakup ice runs and failure induced by the changing thermal-ice condition in the banks and the effects of changing river and groundwater levels. Climate change impacts, including permafrost thaw, glacier melting, and catchment erosion, have become important factors for sediment and channel dynamics research. Fluvial processes are studied through field research using direct measurements to increase the understanding of channel hydrodynamics under various conditions with different morphological characteristics. However, wintertime field measurements on sediment transport and morphological changes over a river reach are challenging, especially during rapidly changing ice conditions due to the need for more suitable instruments and safety issues. Coupling theoretical analysis, mathematical models, and well-designed laboratory and field studies with advanced monitoring techniques will enhance research on ice effects on sediment transport and channel morphology.