The artificial supply of sediment, commonly referred to as sediment augmentation or sediment replenishment, is a method that is increasingly used to address sediment-related issues in regulated rivers. This article presents typical design approaches and assessment methods for different types of sediment augmentation measures (SAM), based on experience from the field and the literature.
1. Introduction
The sediment regime of a river is a fundamental driver of the fluvial ecosystem and is therefore a key subject of focus for its rehabilitation and management
[1]. Bedload sediment deficit is a common phenomenon downstream of dams
[2] and in areas of extensive gravel mining
[3]. It promotes progressive river bed incision
[4] and the coarsening of the bed material
[5] until a static armour layer develops
[6]. It also leads to an overall reduction of morphological channel dynamics
[7].
Sediment-starved rivers can develop a number of unwanted morphological effects, such as decreased river bank stability
[8], local scour
[9] or groundwater overdrafting
[10]. Sediment deficit also induces negative ecological effects, such as missing spawning grounds
[11] and dynamic habitat spaces for fish
[12]. With a constant low flow regime, suspended fine particles settle into the open pore spaces of the static bed layer which can lead to the clogging of the river bed
[13]. Clogging can inhibit hyporheic exchange processes between a river and the adjacent groundwater
[14] and can amplify adverse ecological effects
[15].
The artificial supply of sediment, commonly referred to as sediment augmentation or sediment replenishment, is a method that is increasingly used to address sediment-related issues in regulated rivers. The term sediment augmentation is used here as general term, because it is free of implications for its design and objectives. The design of sediment augmentation measures (SAMs) depends on the defined objectives and the morphological, hydrological, and ecological conditions of the river.
Sediment augmentation has been widely practised and documented over the last decades, particularly in Japan, the USA and in Europe
[16]. Due to the availability of data and the good representation of varying measures, this study focuses on experiences from case studies and related publications from those three regions.
The main objective of this study is to summarize case-specific restrictions, design approaches and assessment methods for SAMs.
2. Design Approaches
2.1. Sediment Properties
The properties of all types of augmented sediment mixtures should reflect the riverbed material in its natural state. This means that the material should have a relatively low content of fine particles (washed) in order to reduce clogging
[13], and it should have an abraded topography, similar to an alluvial material (rolled), in order to promote mobilization
[67][17]. Gravel is usually better for sediment-depleted river systems than fine sediments, which causes colmation of the substratum and high turbidity
[68][18]. Schälchli
[69][19] provides a rough estimation for the diameter of suspended grains that can cause internal colmation (>0.02 mm) and external colmation (>1 mm), depending on the filter medium. If a sediment mix is added for other management purposes (e.g., to manage scour issues) or material is supplied from a reservoir or the adjacent floodplain, these properties can vary. If organic material, such as earth or mud, is included, biological clogging
[70][20] or eutrophication
[71][21] can have adverse ecological effects. If the amount of organic input remains low, the adverse effects are negligible compared to the quantities of natural organic matter that are transported during a major flood event
[72][22].
If the principal objective is the creation of new spawning habitats, the grain size distribution (GSD) depends on the target fish species. Usually, the dominant fish species of the region is targeted. The dominant species can typically be identified based on the mean slope and channel width
[73][23]. Diminishing numbers of specific fish stock can lead to the selection of a different target species. Based on a review of 22 publications, it appears that the Atlantic salmon (Salmo salar) and brown trout (Salmo trutta) generally prefer pebbles that range from 16 mm to 64 mm in diameter for spawning
[74][24]. In some areas, smaller grain sizes of around 20 mm to 30 mm are identified as traditional female spawning grounds (e.g.,
[75][25]). In addition, a smaller percentage (below 20%) of fine particles (d < 0.85 mm) was found to sharply increase the survival rate of fish embryos before their emergence
[76][26]. This can be linked to the fact that the oxygen concentration may be reduced by the deposition of fine sediments
[74][24].
For the alteration of the riverbed structure or channel dynamics, a broader GSD is possible, depending on the desired morphological impact on the target reach. Laboratory experiments have shown that sediment pulses with material finer than the median diameter of the surface layer can promote its mobilization and fining
[77][27]. On the other hand, numerical investigations into the effect of grain size on bed deformation in meandering channels have shown that increasing the GSD of the bed surface (i) increases bedform height, (ii) decreases the bedform length, and (iii) gradually changes the bed configurations over time from the alternate bars shape to the ripple shape
[78][28].
If the principal objective is to target the balancing of the bedload budget, the GSD of the augmented sediments should represent the GSD of the natural bedload material of the river. Bedload, by definition, consists of “particles that spend the majority of the time on the bottom, but are periodically entrained into the turbulent water flow and carried a short distance downstream before settling again”
[79][29]. The boundary between bedload and suspended load is not sharp, and depends on the flow strength, where grains coarser than about 8 mm tend to travel as bedload
[80][30]. A common conceptualization of bedload motion involves the visualization of an active layer
[81][31], where a distinction has recently been drawn between an event active layer and a dynamical active layer
[82][32]. Both types of mechanism may be targeted by an SAM. Hereon, these mechanisms are correspondingly referred to dynamical active or event active bedload supply. The latter represents the coarser fraction of the bedload material, which is only mobilized during a channel-forming flood event (~HQ2).
2.2. Volume
The sediment volume of a single SAM in the case studies varies from small (<10³ m³) (e.g.,
[21,25][33][34]) to medium (10³–10
4 m³) (e.g.,
[34,43][35][36]) and large (10
4–5 × 10
4 m³) (e.g.,
[32,35][37][38]).
The required volume for spawning habitat rehabilitation (SHR) can be estimated based on the missing spawning substrate that might other potentially be present in the target section. The criteria for identifying potential spawning habitat space varies according to the species and is largely determined by hydraulic factors, such as water depth and flow velocity
[83][39], as well as the morphologic factors, such as spawning pit depth and bed surface structure
[84][40]. Salmon spawn mostly at a flow depth of 20 cm to 50 cm and at an average flow velocity of 0.35 ms
−1 to 0.65 ms
−1, while trout spawning areas were found at slightly shallower sites (15 cm–45 cm) with lower flow velocities (0.2 ms
−1–0.55 ms
−1)
[74][24]. The mean spawning pit depth of the brown trout, adjusted according to channel type and geomorphic unit, was assessed based on 268 randomly sampled pits, and was found to be between 6.6 cm and 9.4 cm
[84][40]. This height determines, according to the species, the minimum height of the spawning-suitable substrate that is required at potential spawning ground locations. The typical bed surface structure of the brown trout spawning area is in the upward front slope of a riffle structure
[52][41]. A similar preference of bed surface structure was reported for the Chinook salmon, where most of the spawning (73%) occurred upstream of the crest of the riffles
[85][42].
If morphological changes are the principal objectives, the required volume for SAMs can be estimated using hydro-morphological modelling
[86,87,88][43][44][45]. Single, small or medium-sized SAMs have been shown to alter riverbed structures
[25,42][34][46]. Significant channel shifting, on the other hand, has not been reported, even for large single SAMs
[32][37]. Laboratory experiments suggest that sediment supply has a significant amount of control over channel-scale bedforms
[57][47], especially if the conditions of constant discharge and non-erodible banks do not prevail
[89][48]. Channel dynamics are thus likely to be altered significantly only with recurring supplies of sufficient sediment volumes. For the morphological development of dynamic river widenings, a sufficiently large, constant sediment supply is presupposed
[55][49].
Some federal guidelines have lately produced a uniform definition and calculation of bedload budget for sediment regime rehabilitation with SAMs. The French guidelines for the measurement and modelling of bedload transport
[90][50] defines the transport capacity as the temporal mean of the bedload transport rate. The numerical calculation is based on an equilibrium state, supposing a uniform flow and bedload regime and sufficient material at one’s disposal. The Swiss guidelines for bedload regime rehabilitation states that the evaluation of the sediment augmentation volume required to balance the bedload budget should be based on a defined reference state that is determined according to the watercourse
[65][51]. Schälchli and Hunzinger
[91][52] defined five major goals (which are concerned with channel shape, sediment deposits, substratum, groundwater regimes and flood protection) that must be met, and provide empirically based evaluation methods for the corresponding bedload volume.
2.3. Injection Method
In-channel injection facilitates the instantaneous creation of spawning habitats
[92][53]. Here, sediments are directly dredged out or installed in place to modify the riverbed structure in the target reach (see
Figure 31). SHR can be planned with the help of hydrodynamic
[31][54] or ecological modelling
[93][55]. In recent years, artificial riffles have also been constructed to improve water flow and sediment transportation, as well as to initiate the processes that lead to the restoration of natural riffle-pool sequences
[94][56].
Figure 31. Installation of in-channel stockpiles at the residual flow section of the Sarine River, downstream of Rossens Dam, Switzerland, in 2016. About 1000 m3 of an alluvial sediment mixture from the adjacent floodplain were injected and coupled with an artificial flood to improve the downstream riverbed structure and increase habitat diversity.
If accessibility, budget or other reasons require an upstream supply of sediment, in-channel, point bar or high-flow stockpiles are common alternatives
[47][57]. Stockpile injection can require supplementary conditions for an SAM that might decrease its efficiency, especially when the principal objective concerns SHR. A mobilization event has to occur before vegetation encroachment stabilizes the stockpiles, and the sediments therefore need to be mobilized and deposited in sufficient quantities at the location of potential spawning sites
[27][58].
To prevent vegetation encroachment from stabilizing the stockpiles, high-flow constant injection, e.g., with a heavy truck or a conveyer belt
[24][59], is another method for sediment supply. Laboratory experiments suggest that a high degree of sediment pulse dispersion occurs with high-flow injection, with only some translational transport behaviour occurring for larger hydrographs that are much greater than the entrainment threshold
[95][60]. Depending on the objective of the SAM, the dispersal behaviour of the sediment pulse might require the adaption of other design criteria, e.g., the volume.
One way of passively supplying sediments to a river is by inducing bank erosion. Rohde et al.
[51][61] describe three different types of measures, namely, self-dynamic development, self-dynamic development with initial measures and mechanical widening. One example is the Töss River in Switzerland, where the flow was divided with an artificial island consisting of large boulders
[96][62]. At the Mur River in Austria, a sidearm was dredged and bank erosion was enabled by the removal of bank protection structures. Coupled with in-channel injection, the short-term success of countering channel incision could be assessed
[97][63]. Since riverbank failure, basal residence time and the supply of material to the in-channel sediment transfer system are coupled processes and difficult to simulate
[98][64], the prediction of the morphological impact of a corresponding SAM entails a high degree of uncertainty.
Sediment augmentation through the reactivation of old side channels can increase sediment transport
[40][65] but it may take decades until a new (quasi)equilibrium of the bedload budget is reached. Under natural boundary conditions, a positive cascading effect on channel dynamics, bedload structure and interstitial habitat may develop (
Figure 2).
Figure 2. Principal rehabilitation and management foci of sediment augmentation measures.
2.4. Mobilization Event
If no direct in-channel injection is to be performed, then an SAM requires flood events to mobilize the injected sediments.
Natural flood events are often difficult to predict and might not occur at the right time or magnitude for SHR. For other principal objectives of SAM (bedload budget, channel dynamics, riverbed structure), single or several natural flood events have been successfully used to mobilize parts of mostly large augmented sediment volumes
[32,33][37][66]. In Japan, natural floods are expected in monsoon season and can be controlled to favour both reservoir flushing and downstream SAMs
[99][67].
Environmental flow releases from reservoirs target downstream ecological or management objectives
[100][68]. An environmental flow release scheme ideally contains both large channel maintenance floods that would have a morphologic impact, as well as smaller floods for habitat maintenance
[101][69]. Therefore, it can be coupled with all forms of SAMs.
As well as environmental flow releases that benefit the river, reservoir flushing operations also focus on issues inside the reservoir, such as emptying the reservoir of sediments
[102][70]. Even though a synthesis between reservoir flushing operations and downstream rehabilitation measures, such as with SAMs, has been called for
[103][71], the multiple objectives are often too far apart to permit the specification of a particular discharge and water volume
[104][72]. It is therefore assumed that reservoir flushing, as a mobilization event, is a relevant design approach for SAMs with less specific morphological target states.
2.5. Period and Frequency
The period and frequency of SAMs depend on the principal objective. In any case, in-channel construction works should fall outside the flooding season and reproduction seasons of dominant aquatic species.
SAMs for SHR should be implemented before the spawning period of the target fish species so that they can provide clean and unclogged spawning substrates. In Switzerland, the Federal Office of the Environment (FOEN) recommends performing SAMs for SHR from the late summer to the autumn, between the reproduction period of cyprinids and salmonids
[105][73]. Depending on the size of the watercourse, a repetition after one or two years has been proven to be reasonable for maintaining the positive effects on the fish fauna
[52][41].
Changing the riverbed or channel structure with an upstream sediment supply requires major mobilization events
[32,34][37][35]. The SAM should therefore seek to benefit from peak annual discharge and be repeated based on recurrent assessment.
Bedload restoration with SAMs should be optimized in a way that ensures that bedload is regularly available for transport. The discharge years and discharge quantities can also be flexibly determined depending on the available bedload so that the required bedload is eventually deposited over time.
3. Assessment Methods
Different types of assessment methods exist to quantify the effectiveness of SAMs. The assessment methods should be defined in the early planning stage and be based on the dimension, effort and objective of the SAM, as well as on the ecological importance of the watercourse.
Biotic indicators, such as communities of fish
[106[74][75],
107], benthic macroinvertebrate
[108][76] or riparian vegetation
[109][77], as well as suspended particle organic matter
[58][78], have been used to investigate the effects of SAMs. The selection of biotic indicators varies according to location and should be proven to show a measurable and quantifiable relationship with broader biodiversity
[110][79]. Under this condition, biotic indicators can be used for short- and long-term assessment for all types of SAMs.
Abiotic indicators are based on the field records of hydro-morphological parameters. For example, the Hydro-Morphological Index of Diversity (HMID)
[41][80] is calculated from the records of flow velocity and water depth along predefined transects, and provides a quantification measure for the degree of flow complexity and morphological variability. It has been used to assess changes in the riverbed structure after an SAM
[111,112][81][82] and to assess the development of channel dynamics
[59][83] (
Figure 43a). Abiotic indicators can also indirectly provide insights into the effectiveness of sediment augmentation for bedload regime rehabilitation, e.g., through the assessment of the degree of colmation in the downstream reach
[113][84] (
Figure 43b).
Figure 43. Assessment of the eco-morphological effectiveness of a sediment augmentation measure: (
a) recording flow velocities and water depth along fixed cross-sections, for the calculation of the HMID abiotic indicator; (
b) mapping substrate quality for the calculation of the IRS
[113][84] abiotic indicator; (
c) RFID-pit sensor; (
d) a pebble with an RFID-pit sensor inside; (
e) post-flood bedload tracing, searching for RFID-pit sensor pebble with a mobile antenna.
Topographic surveys using satellites, aerial imagery or bathymetry can be used to assess morphodynamic development after the implementation of an SAM on a reach-
[34][35] and basin-scale
[37][85]. These methods are suited for large-scale and long-term impact assessment at reduced costs, due to the typically high degree of automatization in the data acquisition and processing procedures.
SAMs that focus on bedload regime rehabilitation can also be assessed based on bedload measurements. Direct measurement of bedload is performed with specially designed sediment traps
[2]. For indirect, continuous measurements, acoustic instruments, including geophones, hydrophones and underwater microphones
[114][86], as well as acoustic Doppler current profilers
[115][87], can be used. Passive measurements have also been performed in recent years using more sensitive seismic measurements
[116][88]. Another indirect measurement, successfully used to assess the impact of an SAM, is RFID-pit tracing
[33][66]. Sediments are marked with a tracer, placed inside the sediment augmentation deposit and are later searched for with the help of a mobile antenna
[42][46] (
Figure 43c–e).
In any case, the selection of assessment methods and the defined indicators, as well as the interpretation of their development after project implementation, must be conducted after careful consideration. Woolsey et al.
[117][89] suggest four guidelines for the project-specific indicator selection for river restoration projects. They recommend: (i) limiting the number of indicators, which together represent all project objectives; (ii) the use of direct indicators rather than indirect ones; (iii) choosing indicators that require low effort, especially where financial and time constraints are important; and (iv) the selection of survey intervals that represent both the interannual patterns and the years that elapse after restoration. In addition, for a more holistic ecological approach, the combination of different assessment methods and indicators can lead to a better representation of the interaction of communities of species and habitat properties
[118][90].