1 Introduction

Along coastal and littoral lake zones, submerged aquatic vegetation affects ambient hydrodynamics by reducing water column turbulence, leading to a reduction in sediment erosion, and thus increasing the water column clarity in lakes and saltmarshes [1,2,3]. When the water clarity is enhanced, there is greater light penetration and this creates positive feedback for the canopy [4,5,6,7].

Sediment resuspension and turbidity variations have been observed to impact plant development and hydrodynamics. For example, the construction of a large dam caused the ecosystem in the Dutch Wadden Sea to collapse from a vegetated to a bare state as a result of the increase in turbidity [8]. This then led to eutrophication, caused by a decrease in light availability, and the migration of seagrass meadows to shallower waters [7]. In Lake Taihu, Zhu et al. [9] found that under similar wind speeds, the presence of macrophytes reduced sediment resuspension rates by 29-fold. Consequently, eutrophication and cyanobacteria blooms along the calm shoreline areas of Lake Taihu negatively impact on its ecosystem [10]. Comparative data in the Mediterranean show that a canopy of Posidonia oceanica may reduce resuspension rates by three- to seven-fold compared to those in the adjacent unvegetated floor [11, 12].

Plants with different morphologies may alter the hydrodynamics differently and, therefore, the processes of erosion, suspension and deposition [1, 3, 13,14,15]. Wu et al. [10] found that the zones covered by littoral aquatic macrophytes in Lake Taihu had thicker sediment layers. The amount of sediment erosion and resuspension is known to be governed by the intensity of the external forcing event [16] and canopy properties [17]. The sediment resuspension by unidirectional flow through a simulated canopy has been found to be a function of both the flow velocity and the wakes produced by the stem scale turbulence [18]. Therefore, a threshold in the shear stress can be stablished as a function of the flow velocity and the array of the cylinders. In contrast, field studies have evidenced the role between the sediment resuspension and the presence of intermittent turbulent events [19]. Studies using emergent plants have shown that turbulence inside canopies decreases linearly with increasing stem density, and that even low densities of plants can produce substantial reductions in turbulence [20]. On the other hand, Bouma et al. [21] found that sparse canopies of rigid plants increased flow velocity, and thus sediment scouring and resuspension. The high flow velocities in sparse canopies can also impact on the distribution of seeds, nutrients and sediments [22, 23].

A great deal of research has been carried out to determine the effects emergent and submerged vegetation have on hydrodynamics [13, 14, 24,25,26,27]. Turbulence is generated in the wake of individual stems as well as in the canopy as a whole, and also by shear as a result of the velocity gradients in the mean flow field [28]. Density and plant flexibility are the key parameters that control the TKE attenuation within canopies and therefore the sediment resuspension [15]. However, most of the work has been carried out in flows dominated by waves or mean currents and not in cases where the turbulence is the main hydrodynamic force. The littoral zones of lakes and ponds are regions with limited advection and the main source of turbulence comes from wind action on the surface, or night convection [29]. In these systems, the turbulence produced at the water surface decreases with depth. Therefore, further work needs to be done to quantify the effect that both flexibility and canopy density have on the sediment resuspension produced by zero-mean flow turbulence. One way of approaching this problem is by running experiments using an oscillating grid device. Oscillating grids produce nearly isotropic zero-mean flow turbulence [30,31,32] and have been used since the 1990s to study isotropic turbulence in the absence of the mean shear associated with flowing water. The properties of the turbulence are determined by the geometry of the grid, the frequency and amplitude of the oscillations, and the distance from the grid [33, 34]. Oscillating grid turbulence devices (OGT) can be used as an analogue to open-channel flow systems by setting the operational parameters of the grid (stroke, frequency, etc.) such that the total kinetic energy of the turbulence matches that expected either at the bed or at the free surface for an open-channel flow [35].

OGTs are used to produce controlled turbulent fields allowing turbulence in physical phenomena to be understood. OGTs have been used to study the resuspension of both cohesive [36] and non-cohesive [37] sediments. Tsai and Lick [36] found that the concentration of resuspended cohesive sediment was proportional to the oscillation frequency of the grid. Huppert et al. [37] found that above a critical oscillating frequency, a given mass of non-cohesive sediment particles can be kept in suspension indefinitely. This critical frequency depends on the diameter of the sediment particles. Orlins and Gulliver [35] used OGTs to study sediment resuspension from bare beds with two different consolidation times (2 and 11 days). For the same level of TKE, less-consolidated sediment beds are subject to greater amounts of resuspension. Given than turbulence can act on sediment beds on short time scales, this study also quantifies the effects turbulence has on beds from short (hours) to long consolidation times (days), therefore covering a greater range of consolidation times than that considered by Orlins and Gulliver [35]. In canopies of aquatic vegetation, the turbulence induced by the wind affects the bottom boundary layer of the flow field in a manner that depends on the canopies’ properties and the bed’s degree of consolidation [38]. In addition, this study investigates the induced resuspension of natural cohesive partially consolidated sediment beds by turbulence in non-vegetated and vegetated environments under zero-mean flow turbulence. In this case, the entrainment of sediment particles from the interface is a result of turbulent fluctuations rather than the presence of a mean flow [39]. For this reason, an OGT has been considered suitable for studying the sediment resuspension. The canopy properties, such as the plant flexibility and canopy density, are expected to play an important role in the attenuation of pure isotropic turbulence, which has not been previously determined. Therefore, different canopy densities and plant models composed of flexible, rigid and semi-rigid plants will be considered. Furthermore, the sediment characteristics will also be explored. For this purpose, three sediments with different particle distributions will be used for the experiments.

2 Methodology

2.1 Experimental setup

The study was conducted in an oscillating grid turbulence chamber (Fig. 1) consisting of a box made of Plexiglas® whose interior dimensions measured 0.28 m × 0.28 m × 0.33 m. This was filled with water to a depth, hw, of 0.315 m. A Plexiglas® grid was suspended from above the chamber such that its center was z0 = 0.065 m below the water surface (0.25 m above the bottom of the chamber). The oscillating grid was constructed with 1 cm wide and thick Plexiglas® square bars. Following the same technical requirements like those of De Silva and Fernando [30], the grid was composed of 5 × 5 bars, with M = 0.05 m spacing (or ‘mesh size’) between the bars giving it a 31% solidity (defined as the fractional solid area occupied by bars). Using a variable speed motor located outside the tank, with a fixed stroke s = 0.05 m, and frequencies f = 2.8, 3.3, 3.8, 4.3 and 4.8 Hz, the grid was oriented horizontally and oscillated vertically. A clearance of 2 mm between the sidewalls and the grid was maintained. We defined the vertical direction as z (positive downwards), and z = 0 cm as the mean vertical position of the oscillating grid.

Fig. 1
figure 1

Schematic diagram of the experimental OGT setup (top panel). Photograph of the grid (bottom panel)

Full size image

2.2 Vegetation models

Simulated canopies of either rigid, semi-rigid or flexible vegetation were placed in the tank prior to each experimental run. The rigid canopy models consisted of d = 6 mm wide and hs = 0.10 m long PVC cylinders (Fig. 2a). The flexible canopy models were constructed by taping flexible polyethylene blades to rigid PVC dowels 0.02 m long and 6 mm in diameter (Fig. 2b). Each simulated plant had eight 4 mm wide, 0.10 m long and 0.07 mm thick plastic blades. These flexible plant simulants were dynamically and geometrically similar to typical seagrasses, as described by Ghisalberti and Nepf [40], Folkard [41], Pujol et al. [13] and El Allaoui et al. [42]. The ratio between the thickness and the height of the plant was 7 × 10−4, similar to that used by Folkard [41] of 8 × 10−4. The aspect ratio of the plant (ratio between the width of the leaves and its height) was 0.04, the same as that used by Folkard [41] who used 0.25 m long and 0.01 m wide leaves. Therefore, the flexible plant model simulates the behavior of a Posidonia oceanica canopy under a turbulent flow. Blade density was less than that of water (as is the case for real seagrasses) so that, at rest, the flexible canopy height was the same as that of the rigid canopy. The semi-rigid canopy was made of nylon threads each 2 mm in diameter (Fig. 2c). To compare semi-rigid to flexible vegetation at d = 6 mm, eight nylon threads were stacked together at the base to mimic the equivalent number of blades (Fig. 2c) to those used for flexible plants.

Fig. 2
figure 2

Vegetation simulations: a rigid vegetation; b flexible vegetation and c semi-rigid vegetation, and the plant distribution for the range of canopy densities studied: dSPF = 1%, eSPF = 2.5%, fSPF = 5%, gSPF = 7.5% and hSPF = 10%

Full size image

Following Pujol et al. [3], the canopy density was varied and quantified between runs using the solid plant fraction SPF = 100nπ(d/2)2/A, where n is the number of plant stems, and A is the total bed surface area covered by the canopy. For the flexible canopies, d was taken as the diameter of the rigid dowels at the base of the plant (6 mm). SPFs of 1, 2.5, 5, 7.5 and 10% were used for the rigid canopy runs, SPFs of 2.5, 5, 7.5 and 10% for the flexible runs and an SPF of 2.5% was used for the semi-rigid canopy (Table 1, Fig. 2c–h). These SPFs corresponded to densities N of 354, 884, 1768, 2652 and 3536 plants m−2, which is in line with the medium to dense seagrass densities found in the field [12, 43,44,45]. To create each canopy, the plants were secured into 6 mm-diameter holes, which were arranged into a regular grid with 0.01 m center-to-center spacing on a plastic base board. The position of each plant on this grid was made using a random number generator [13, 46]. Holes left unfilled once all the plants had been positioned were covered with tape to eliminate any potential effect the hole may have had.

Table 1 Characteristics of the sediment types used in the experimental work
Full size table

In addition, the vertical variation in canopy density varied from rigid to semi-rigid and to flexible canopies. Following Neumeier and Amos [47], the vertical variation in the canopy density was assessed from the lateral obstruction of the canopy by taking a lateral picture of a 2.5 cm thick canopy in front of a white background. Semi-rigid and flexible blades were painted black to increase the contrast in the image. Images of the lateral obstruction were digitized, and image analysis techniques were applied to differentiate the vegetation from the background. Finally, the lateral obstruction percentage was calculated. While rigid canopies had a lateral obstruction that remained constant with height, the lateral obstruction of the flexible plants varied with height and maximum percentages being from z = 18 cm to z = 22 cm (Fig. 3). The flexible 10% SPF canopies reached greater lateral obstruction areas (of 33%) than the rigid canopies (of 16%). For the semi-rigid canopy of 2.5% SPF, the maximum lateral obstruction area of the canopy was of 6.7%, i.e., midway between that of the rigid and flexible canopies.

Fig. 3
figure 3

Lateral obstruction area of the vegetation calculated from lateral pictures of a 2.5 cm thick canopy for a flexible plants and b rigid plants, for different SPF

Full size image

2.3 Sediment bed emplacement

Once the simulated canopy had been secured at the base of the experimental tank, and the tank had been filled with water, the bottom of the tank was then covered with sediment. Three types of sediment of different compositions were used (Table 1). Enough sediment from the marsh and lake areas was obtained in situ to perform all the experiments according to the designed experimental conditions. The sediment was cleaned to remove leaves and roots, dried and then sieved to remove particles larger than 500 μm.

The sediment particle size distribution (i.e. the sediment concentration C versus its particle size diameter d) for each sediment type used was analyzed with the Lisst-100X, (Sequoia Scientific, Inc., WA, USA) a laser particle size analyzer which has been used extensively and found to be appropriate for measuring either organic [48] or inorganic particles [12, 49]. Based on the classification from Rijn [50] and Blott and Pye [51], the sediment was divided into three ranges of particle diameter (Fig. 4). The first (2.5–6.0 μm) corresponds to very fine silts (strongly cohesive), the second (6.0–170 μm) to fine to coarse silts and small sand particles (weakly cohesive), and the third (> 170 μm) to small and medium sand particles. Considering the particle number distribution, the sediment analysis showed that ≈ 98% of the particles fell within the first range, while particles within the second range accounted for the remaining 2%. However, in considering the particle volume concentration for the three sediment types, particles in the first range accounted for 38.2% (marsh), 29.73% (lake) and 24.6% (synthetic) of the total concentration. An increase in the percentage of small particles in the sediment distribution is expected to increase the cohesive properties of the sediment.

Fig. 4
figure 4

Particle size distribution of the synthetic, lake and salt marsh sediments used in the experiments. The vertical dashed lines represent the classification by Rijn (2007)

Full size image

For the case without plants, experiments with different sediment bed thicknesses were considered to determine the effect this would have on the results obtained. The bottom of the tank was covered with a sediment layer to the uniform heights of 3.8 mm, 2.5 and 1.3 mm, which corresponded to dry mass concentrations of 300 gL−1, 200 gL−1and 100 gL−1, respectively. This seeding was performed by manually moving a tube (connected to the container) holding the homogeneous sediment mixture around the bottom of the chamber through the vegetation. The seeding resulted in a cloud of particles ≈ 1 cm in height, which was, following Ros et al. [15], then left to settle. Figure 5 shows the concentration corresponding to the resuspended bottom sediment particles versus the TKE for the three sediment layers. The greater the sediment height at the bottom was, the higher the concentration of resuspended particles. Scouring was not observed in any of experiments that had the 3.8 mm and 2.5 mm high beds. All experiments were initiated with a consolidated bottom bed height of 2.5 mm.

Fig. 5
figure 5

Particle sediment concentration within the suspension versus Turbulent Kinetic Energy for the three bed loads of 100, 200, and 300 gL−1 (Experiment with no vegetation and a time consolidation bed of 2 days for synthetic sediment)

Full size image

Once the sediment was resuspended, the particle volume distribution of the sediment for the second and third particle range was approximately constant throughout all the experiments for the three sediment types. For this reason, these larger particles were not considered in the analysis, and only particles in the smallest size range i.e., the strongly cohesive range, were analyzed.

2.4 Turbulence measurements and analysis

The three-dimensional turbulent velocity field (u, v, w) inside the tank was measured with a three-component Acoustic Doppler Velocimeter (ADV) (Sontek/YSI16-MHzMicroADV). The ADV has an acoustic frequency of 16 MHz, a sampling volume of 90 mm3, a sampling frequency of 50 Hz and measures in the range 0–30 cm s−1. The distance between the head of the ADV and the sampling volume was 0.05 m. The ADV was mounted onto a movable vertical frame allowing it to be manually situated at working depths between z = 0.10 m and z = 0.24 m. For all experiments, the ADV was placed horizontally 0.07 m (1.4 × the mesh size) from one side wall and 0.12 m (2.4 × the mesh size) from the other side wall to avoid side-wall effects, as suggested by Orlins and Gulliver [35]. In addition, following De Silva and Fernando [30], the mesh endings were designed to reduce mean secondary circulation. To avoid any spikes in the data coming from artifacts of instrument operation rather than being representative of the flow, ADV measurements with beam correlations below 70% and signal to noise ratio (SNR) above in the range 15–30 dB. Spikes and spurious data were discarded using the method by Goring and Nikora [52]. The use of single point ADV measurements for characterizing OGT can be justified by noting that several authors [30, 53, 54] found that at a certain distance from the grid, turbulence is isotropic and the velocity fluctuations u′, v′ and w′ are proportional to 1/z. It seems, therefore, plausible to use single-point ADV measurements in this context, at least at |z| > 3 M, where M is the spacing between bars [55]. In the present study, M = 5 cm, therefore for |z| > 15 cm, the turbulence is expected to be isotropic. Furthermore, for the rigid vegetation with SPF = 1% and 2.5%, in order to test for the horizontal homogeneity of the turbulence field, vertical velocity profiles with the ADV were carried out at eight different horizontal locations. Maximum differences of 4% between the TKE measured at different positions were obtained. The Reynolds stresses at each location were calculated and no differences were obtained between locations when considering the margin of error (data not shown). Additional tests were made to guarantee the horizontal homogeneity. The exuberance, i.e. the ratio of upward (u’w’ ≥ 0) to downward (u’w’ ≤ 0) fluxes of momentum, was calculated following Rotach [56]. The exuberance was close to − 1, indicating that there was equal contribution of downward to upward flux of momentum. Consequently, single point ADV measurements were used thereafter.

To obtain valid data acquisition within the canopy for the densest canopies of flexible plants and in accordance with Neumeier and Ciavola [57], Pujol et al. [3] and [13], a few stems were removed (a maximum of 3 stems for the SPF = 10% canopy density) to avoid blocking the pathway of the ADV beams. To minimize the effect this ‘hole’ has only a few stems were repositioned. For the dense flexible canopies, a thin (0.5 mm thick) 4 cm-wide ring was situated 1 cm above the ADV sensors to avoid them being blocked by the flexible plants. This metal ring was fixed with two stems of the same material that were attached to the dowels of the plants. Measurements of the flow velocities for the SPF = 0% experiments were taken with and without the ring and no differences were observed.

For each experiment, a vertical velocity profile was taken from a z = 0.10 m to z = 0.24 m depth (see Fig. 1) at 0.01 m intervals to obtain the turbulence field. Thus, the vertical profiles covered measurements inside and above the canopy. At each depth, the instantaneous water velocity (u, v, w) was measured for 10 min (i.e. 30,000 measurements for each velocity component) and then decomposed as \(u = U + u'\), where U is the time-averaged velocity component in one horizontal direction (x) and u’ is the turbulent component in this direction. The velocity components v (speed in the y-direction—the horizontal direction orthogonal to the x-direction) and w (speed in the vertical direction) were similarly decomposed into \(V + v'\) and \(W + w^{\prime},\) respectively. The turbulent kinetic energy per unit mass (TKE) was then calculated from the mean of the square values of the three turbulent components:

$$TKE = \frac{1}{2}\left( {\overline{{u'^{2} }} + \overline{{v'^{2} }} + \overline{{w'^{2} }} } \right)$$
(1)

One of the characteristics of the zero-mean shear flow in the OGT device is that there is no recirculation in the system, i.e. the mean velocities are zero. Since the effect of the canopy is not known, the total kinetic energy (\(KE\, = \,\frac{1}{2}\left( {U^{2} + V^{2} + W^{2} } \right)\)) can be a parameter to check for the presence of zero mean currents (Fig. 6a, b). Results show that in all cases, and considering the error margin, the KE remains below the ADV noise. The other characteristic of the zero-mean shear in the OGT is that the TKE decreases with z−2 for the region of homogeneous turbulence [55]. In the present study, all experiments with and without plants presented a linear relationship between TKE and z−2 for z > 15 cm (Fig. 6c), i.e. z > 3 M in the homogeneous turbulent zone.

Fig. 6
figure 6

Relationship between the total kinetic energy (KE) at z = 22 cm and the solid plant fraction (SPF) of the canopies for oscillating frequencies, f = 2.8, 3.8 and 4.8 Hz, for a rigid and b flexible canopies. Horizontal dashed line corresponds to the ADV noise level for the KE, set at 0.44 cm2 s−1. c TKE versus (z/hs)−2 for the case WP and for RV and FV of SPF = 5%. Lines represent the linear fit between TKE and (z/hs)−2. For the WP case TKE = 7.82(z/hs)−2-11.08 (R2 = 0.9987), for the RV case TKE = 6.76(z/hs)−2-5.17 (R2 = 0.9954) and for the FV case TKE = 2.69(z/hs)−2-2.37 (R2 = 0.9476)

Full size image

2.5 Sediment entrainment measurements

The downward diffusion of grid-generated turbulence was able to erode the sediment bed and maintain a sediment load in the water column as momentum was transferred to the sediment. Within the column, sediment samples of 80 mL were obtained using a pipette introduced through the opening of the lid situated on top of the experimental tank. Samples were collected from two different depths (z = 0.1 m i.e. 0.05 m above the canopy, and z = 0.22 m i.e. 0.03 m above the bottom). For all the experimental runs, the particle volume distribution of suspended sediment was measured using the Lisst-100X laser particle size analyzer. From these measurements, the particle volume concentration in each range (Fig. 4) was obtained as the sum of the particle volume concentration of all the particles within the size range.

Given that the smaller particles in the size spectra can remain in suspension quasi-indefinitely, suspended sediment concentration (C) was calculated relatively, as the value measured at a time t (Ct) subtracted from the value measured prior to the start of the oscillations at t = 0 (C0), i.e., C = Ct− C0. C0 ranged from 0.7 to 0.9 μl l−1, representing a percentage between 9 and 2.5% of the sediment concentrations measured in the experiments. Each experimental run started at 2.8 Hz, the lowest oscillation frequency of the grid. A steady state was reached after 30 min and then after a further 30 min (at t = 60 min) the oscillation frequency was increased to 3.3 Hz. A second steady state was reached at t = 90 min, and after a further 30 min (at t = 120 min) the frequency was increased to 3.8 Hz. A third steady state was reached at t = 150 min and this continued for a final 30-minute period. Consecutive steady states were reached for frequencies of 4.3 and 4.8 Hz. The evolution of the resuspended sediment concentration Ct with time is shown in Fig. 7 for the experiments carried out with both marsh and synthetic sediments for runs with rigid vegetation of SPF = 2.5%. The dashed line in the plot represents the time evolution of the grid oscillation frequencies. Similarly, Oguz et al. [58] found that 15 min were required for sediment resuspension to reach a steady state in a wave-dominated environment. For the bare soil case, experiments with the different frequencies were also carried out separately (not in the sequence of the increasing frequencies) and the same sediment concentrations were obtained at the steady state. Therefore, all the experiments thereafter were carried out sequentially.

Fig. 7
figure 7

Time evolution of the sediment concentration for experiments carried out for rigid vegetation with SPF = 2.5%, for the synthetic sediment and the marsh sediment. The dashed line at the top panel corresponds to the evolution of the oscillation frequency (f) over the full time period of each experiment run

Full size image

Seven experiments were conducted to study the effect of the consolidation time (runs 21 and 23–28). All of them were carried out without plants, with synthetic sediment and for all the frequencies (Table 2). Three experiments were conducted to study the effect of the sediment type (runs 1, 11 and 21). All of them were carried out without plants for the 2 days of consolidation time and for all the frequencies (Table 2). Three experiments were conducted to study the effect plant flexibility, rigid plants (run13), flexible plants (run 17) and semi-rigid plants (run 22) have. All the frequencies were considered for runs 13 and 22 (Table 2) and three for run 17. All of them were carried out for SPF = 2.5%, 2 days of consolidation time and for the synthetic sediment. Ten experiments for marsh sediment (runs 1–10) and ten experiments for synthetic sediment (runs 11–20) were conducted to study the effect canopy density and type have on the sediment resuspension.

Table 2 Summary of experimental conditions and parameters
Full size table

3 Results

3.1 Vertical turbulent kinetic energy in the presence of a bottom canopy

For experiments without plants, the TKE decreased with vertical distance from the grid (Fig. 8). For experiments with rigid, semi-rigid or flexible canopies, two layers were distinguished: a transition layer and a within-canopy layer (Fig. 8). Within the canopy layer, the TKE for both the rigid, semi-rigid and flexible canopy (SPF = 2.5%) cases were below that for the run without plants. The transition layer extended up to at least 6 cm above the top of the canopy (Fig. 8). In this layer, the TKE for the cases with plants was lower than that for the without-plants case with a TKE difference that decreased from the top of the canopy (38% lower than for the without plants case) down to z = 10 cm (8.7% lower than for the without-plants case).

Fig. 8
figure 8

TKE profiles for experimental runs without plants (WP), and with flexible (FV), rigid (RV) and semi-rigid vegetation (SMRV), all with SPF = 2.5%. Grid oscillation frequency was f = 4.8 Hz in all cases shown

Full size image

To compare between the runs, the TKE at z = 22 cm was chosen to represent the TKE within the canopy. In Fig. 9, the TKE is plotted for both rigid (left panel) and flexible (right panel) plants for all the canopy densities studied, and also for the without-plants case. In all cases, the TKE was found to increase with increasing grid oscillation frequency. In both rigid and flexible canopies, the TKE was below that of the without-plants case (SPF = 0%). In the rigid canopy the TKE reached a minimum at an intermediate value (of SPF = 5%), remaining constant afterwards for SPF > 5%. In contrast, for flexible canopies the TKE decreased gradually with increasing SPF. It is important to notice that for SPF < 2.5%, flexible and rigid canopies present similar TKE for the same oscillating frequency. However, for SPF > 2.5%, the TKE for flexible plant is smaller than that for rigid plants.

Fig. 9
figure 9

Relationship between the turbulent kinetic energy (TKE) at z = 22 cm and the solid plant fraction (SPF) of the canopies for different oscillating grid frequencies, f, for a rigid and b flexible canopies

Full size image

3.2 Sediment re-suspension in the presence of a canopy: the effect of plant flexibility

Within the canopy, the behavior of the suspended sediment concentration at the steady state (Css) with SPF was different for rigid and flexible canopies (Fig. 10a, b, respectively). Css for the without-plants experiments was greater than for all the experiments with rigid plants. The greater the oscillating frequency, the higher the Css was. For rigid canopy models, Css was nearly constant with SPF for all the frequencies tested. In contrast, Css decreased markedly with SPF for flexible canopies, attaining smaller Css for the denser flexible canopies than that of the denser rigid canopies of the same SPF. Similar results were obtained for the synthetic sediments for both rigid and flexible plants (Fig. 10c, d, respectively).

Fig. 10
figure 10

Relationship between the suspended sediment concentration at the steady state (Css) measured at z = 0.22 m and the solid plant fraction (SPF) for different oscillating frequencies (f) for (a and c) rigid, (b and d) flexible canopies, for the marsh (top) and synthetic sediment (bottom)

Full size image

Css was found to follow an exponential relationship with TKE with different exponents for the different vegetation types (Fig. 11). For the same TKE, the highest Css (and the highest coefficient of the exponential) was found for the flexible vegetation model, while the lowest Css was found for the rigid vegetation model.

Fig. 11
figure 11

Dependence of the sediment concentration on the suspension at z = 22 cm (i.e. z/hs= 0.7) and the turbulent kinetic energy, for the three types of canopies (rigid, semi-rigid and flexible) for a solid plant fraction of 2.5%. For all runs, a two-day synthetic consolidated bed was used. Vertical error bars are calculated from the standard deviation of different measurements of the same run. Solid lines represent the exponential best fit curve through the data obtained in each case. The equations of the exponential fitting are Css= 1.46e7448TKE (r2 = 0.9968) for FV, Css= 0.87e7085TKE (r2 = 0.9932) for SMRV and Css= 1.49e2733TKE (r2 = 0.9622) for RV

Full size image

3.3 Sediment resuspension related to sediment bottom consolidation

In all the experiments, the longer the consolidating time, the lower the Css was for all the TKE studied (Fig. 12). Two behaviors were observed based on the evolution of Css with TKE that depended on the consolidation time. The first for the long consolidation time (> 12 h) and the second for the short consolidation time (< 12 h). For long consolidating times above 12 h, Css increased with TKE, following an exponential dependence. On the other hand, and considering the uncertainties, for bed consolidation times between 1 and 6 h, Css was approximately constant with TKE.

Fig. 12
figure 12

Relationship between the sediment concentration of the suspension at z = 22 cm (i.e. z/hs= 0.7) and the turbulent kinetic energy, for the seven bed consolidation times, varying from 1 h to 3 days. For all runs, the synthetic type sediment was used. Vertical error bars are calculated from the standard deviation of different measurements of the same run

Full size image

3.4 Sediment re-suspension related to sediment bottom characteristics

The suspended sediment concentration Css increased exponentially with the TKE for all the sediments tested (Fig. 13). For TKE < 4 × 10−4 m2 s−2, no differences were obtained between the Css obtained for the different sediments. In contrast, for TKE > 4 × 10−4 m2 s−2, the behavior between Css and the TKE depended on the nature of the sediment. The greatest Css corresponded to the marsh sediment and the lowest to the synthetic sediment.

Fig. 13
figure 13

Relationship between the sediment concentration Css at z = 22 cm at the steady state and the turbulent kinetic energy, for the three types of sediments (synthetic, lake and marsh) for the without-plants experiments. For all runs, a two-day consolidated bed was used. Vertical error bars are calculated from the standard deviation of different measurements of the same run. Solid lines represent the exponential best fit curve through the data obtained in each case. The equations of the exponential fitting are Css= 0.56e5937TKE (r2 = 0.9798) for the marsh sediment, Css= 0.67e5213TKE (r2 = 0.9644) for the lake sediment and Css= 0.94e4139TKE (r2 = 0.9398) for the synthetic sediment

Full size image

4 Discussion

The bed sediment within non-vegetated and vegetated model canopies were resuspended due to the turbulence generated by the oscillating grid. The resuspension of particles from the sediment beds was found to depend on the characteristics of the structure of the canopy (both plant density and plant flexibility) and the characteristics of the sediment bed (both consolidation time and sediment composition).

4.1 The effect sediment cohesiveness had on sediment resuspension

The three cohesive sediments studied were resuspended, due to the turbulence generated by the oscillating grid, producing a homogeneous vertical suspended sediment concentration for all the experiments carried out. This homogeneous vertical distribution of sediment is in accordance with the results found by other authors when the suspended sediment concentration was below 80 mg L−1 [59]. In the present study, the maximum concentration of suspended sediment was 30 μl L−1, which corresponds to a mass sediment concentration of 75 mg L−1.

The total suspended solids was found to depend on the degree of TKE near the bottom of the bed, as was also found by Tsai and Lick [36]. The turbulent energy dissipation produced by the oscillating grid for the oscillating frequencies studied ranged from 1.02 × 10−4 to 5.13 × 10−4 m2 s−3. This range of turbulence is characteristic of mean turbulence intensities in the shallow littoral zones in lakes, with mean values of 2.41 × 10−4 m2 s−3 and 3.97 × 10−5 m2 s−3 for water depths of 0.5 m and 1.5 m, respectively [60, 61]. The particle volume concentration was found to exponentially increase with TKE (Fig. 14). The greatest resuspension was found for the marsh sediment, which was 22% higher than that of the synthetic sediment. Given that the sediment mass was the same for both sediments, it is likely that the higher resuspension rates are associated to the greater concentrations of fine particles in the bed. Then, turbulent events acting on muddy bed substrates produce bed erosion resulting in higher water turbidities than sandier regions under the same hydrodynamic forcing [62]. Therefore, our data show that the greater the concentration of fine particles is in the bottom of the bed, the greater the resuspension of particles in the water column. The increase of fine particles in the water column might cause an increase in water turbidity (i.e. a reduction in water clarity) that may have a negative feedback for the ecosystem, especially for organisms that require light to survive.

Fig. 14
figure 14

Relationship between the sediment concentration of the suspension at z = 22 cm (i.e. z/hs= 0.7) and the turbulent kinetic energy, for the rigid vegetation runs, no vegetation runs and for flexible vegetation, for both the synthetic and marsh sediment. For all runs, a two-day consolidated bed was used. Solid lines represent the exponential best fit curve through the obtained data in each case. The equations of the exponential fitting are Css= 0.7e5444TKE (r2 = 0.9073) for RV, and Css= 1.09e10012TKE (r2 = 0.8770) for FV

Full size image

4.2 The effect the structural characteristics of the model canopy had on the resuspension of sediments

Sediment resuspension depended on the characteristics of the vegetation, which is in accordance with Tinoco and Coco [18]. In the SPF range studied, rigid canopies produced less sediment resuspension than bare soils. This result can be attributed to the reduction of the turbulent kinetic energy by the canopy. However, flexible canopies produce a wide range of resuspended sediment concentrations, expanding from smaller to greater concentrations than those obtained for the rigid canopy and the without-plants case. This behavior can be explained by the movement of the flexible plants’ leaves in the water column, because as the leaves are able to capture sediment particles these can be washed off as the flexible plants move. This can explain why, for the same TKE, flexible plant models produce greater resuspension than rigid models that do not move with the flow. The lower values of the suspended sediment concentration obtained by the flexible canopies compared to the rigid ones, corresponds to the cases with high SPF, where the TKE is greater for rigid plants than for flexible plants. Therefore, once sediment particles are resuspended from the bottom their settling in a flexible canopy is lower than it would be in a rigid canopy. Therefore, beds covered with flexible plants in the field might present a greater erosion of the finer particles once resuspended, as they are potentially transported to other regions by waves and currents. In such cases, unlike the beds in rigid canopies, the beds with flexible canopies would result in sandier compositions.

The finding that dense canopies of flexible plants reduces sediment resuspension more than the sparse canopies of flexible plants do, is in accordance with the findings from field [12, 62] and laboratory experiments [63]. The presence of macrophytes in shallow lakes effectively abates sediment resuspension as a result of a reduction in bed shear stress or turbulent kinetic energy above the bed [64, 65]. In experiments conducted in lake enclosures, Li et al. [66] found that macrophytes reach their maximum effectiveness in reducing resuspension at a certain species-specific biomass threshold, beyond which the biomass effects on resuspension are negligible. This result is in accordance with the findings in the present study. For example, flexible canopies with SPF lower than SPF = 7.5% substantially reduce sediment resuspension, whereas canopies with densities over SPF = 7.5% do not produce any further decrease in sediment resuspension. In the coastal Mediterranean, canopies of Posidonia oceanica have been found to reduce resuspension rates by three- (medium dense canopies) to seven-fold (dense canopies) compared to those in the adjacent unvegetated floor [11, 12].

4.3 The effect sediment bottom bed consolidation had on sediment resuspension

Different sediment resuspension dynamics have been found depending on whether the sediment is consolidated for a short or long period. Sediments that have a long consolidation time will require a greater critical turbulent kinetic energy to initiate resuspension from a bed. These results are in accordance with Orlins and Gulliver [35] who found that for TKE < 10−3 m2 s−2, the same level of TKE produced a greater resuspension for low consolidation times. Orlins and Gulliver [35] found that for TKE = 10−3m2s−2, resuspension did not depend on the consolidation times studied (2 and 11 days). Mud erodibility was tested by Lo et al. [67] on cores containing suspensions of coastal lake sediments that were consolidated for 1, 2 and 4 weeks, and found that the strengthening of the beds could be attributed to the bed’s time consolidation, and inversely on initial suspension concentration over concentrations ranging from fluid mud to hydraulic dredge effluent.

For high TKE of 2 × 10−3 m2 s−2, Orlins and Gulliver [35] found that the total suspended solids concentration was independent of the consolidation times of the 2 and 11 days they studied. Our experiments were extended to shorter consolidation times than those studied by Orlins and Gulliver [35] but the highest TKE studied was 5.5 × 10−4 m2 s−2, lower than the threshold found by Orlins and Gulliver [35]. Our results show that the shorter the consolidation time is, the greater the suspended sediment concentration (Fig. 11). Furthermore, for consolidation times below 6 h, and considering the uncertainty in the data, the concentration of suspended solids was independent of the TKE for the range of TKE studied. However, for consolidation times above 6 h, the concentration of suspended solids increased with the TKE, especially for TKE > 4 × 10−4 m2 s−2. For these ranges of consolidation times above 6 h, the difference in the suspended sediment concentration between the different consolidation times decreases with TKE but, contrary to the findings by Orlins and Gulliver [35], still remained different for the highest TKE studied, which was probably due to the fact that the TKE in the present study was below the threshold of Orlins and Gulliver [35]. The results found in our study, agree with those of James et al. [68] where, for sediments located at canopy-forming and meadow-forming beds, the concentration of suspended solids increased markedly as a function of increasing bottom shear stress.

5 Conclusions

The resuspension of sediment by zero-mean turbulence depends on the consolidation time of the bed, the composition of the sediment and the characteristics of the bed (vegetated or bare soil). For vegetated beds, the characteristics of the canopy, in terms of its plant flexibility, is crucial in determining sediment resuspension. We found that the degree to which the sediment bed was consolidated played a crucial role in determining the magnitude of the sediment resuspension. Sediments that have a long consolidation time will require a greater critical turbulent kinetic energy to initiate resuspension from a bed. As such, for beds with consolidation times lower than 6 h, the suspended solids were independent of the turbulent kinetic energy. However, for consolidation times above 6 h, the concentration of the resuspended sediment increased markedly with the turbulent kinetic energy, especially for turbulent kinetic energies greater than 4 × 10−4 m2 s−2. For these ranges of consolidation times, the suspended sediment concentrations increased with the turbulent kinetic energies.

In the simulated vegetated experiments, rigid, semi-rigid and flexible plant canopies were found to reduce the turbulent kinetic energy in shear-free conditions compared to without-plants cases. Dense flexible canopies of SPF = 5% reduced the turbulent kinetic energy more than the rigid canopies, thus reducing sediment resuspension in the water column. In contrast, sparse canopies of flexible stems produced similar turbulent kinetic energies to those of the rigid canopies of the same density For the same level of turbulent kinetic energy the resuspended sediment in the flexible canopies was higher than in the rigid canopies as a result of the movement of the plant leaves. Assuming that stable substrates play a vital role for plant survival, this suggests a mechanism that may lead to dense distributions of flexible vegetation being better able to survive than sparse flexible canopies.