Bed level dynamics on decadal time scales

Figure 2 presents the bed level evolution at fixed locations on two intertidal flats (Fig. 1a), measured on average 7 and 14 times a year, respectively. The Zuidgors intertidal flat (mean tidal range of 4.2 m; Western Scheldt) and Galgeplaat intertidal flat (mean tidal range of 2.9 m; Eastern Scheldt) are considered here. The measurement points (Z1–Z3; G1–G4) were 40 m to 175 m apart (Fig. 2a, b). Some short-term deviations along the long-term bed level trends, which were well in excess of the centimetre measurement precision, only affected specific locations (circles in Fig. 2c, d), while other impacts affected (almost) all measurement points (squares in Fig. 2c ,d). Even then, the impact was non-uniform (e.g., the first highlighted event in Fig. 2c). The intertidal flats recovered from various events in Fig. 2c, d within the next measurements, while other events affected the long-term evolution persistently.

Both identified setbacks (i.e., bed level changes against the long-term evolution) at Zuidgors (Fig. 2c) had a vertical impact similar to approximately 4 months of evolution (impact of (sim ) 15 cm, evolution of (sim )50 cm/year). The largest bed level change that persisted at Galgeplaat occurred in 1990 simultaneously at G1–G3 (Fig. 2d). The bed level changes resulted from a major storm event (26 February–2 March), characterized by 10-min averaged wind speeds that peaked locally at 27 m/s and were above 15 m/s for 73 h. The Eastern Scheldt has a storm surge barrier at its mouth (Fig. 1), that closed for four tides within this event, two more than any other storm in the record. This implied an absence of tidal forcing for a large part of the storm while wind-driven flow and waves affected the intertidal flat for long durations of near-constant water levels. Figure 2e shows the evolution of G1 including a linear fit based on the data measured during the two years before the storm, and a linear fit based on the data measured after the storm. The bed level evolution rate did not significantly change: − 1.90 cm/year before the storm and − 1.94 cm/year after the storm (both one order of magnitude larger than the sea level rise rate), implying no sign of recovery. The 17 cm of erosion during that event was equivalent to nine years of erosion at the average rate.

Figure 2

Bed level evolution of the point-elevation measurements. (a, b) Location of the sampling points Z1–Z3 on the Zuidgors intertidal flat (Western Scheldt) and G1–G4 on the Galgeplaat intertidal flat (Eastern Scheldt). See Fig. 1 for the position of these transects. The elevation maps, measured halfway the period of the time series, are shown as a reference. (c, d) Time series of the bed level evolution at Zuidgors and Galgeplaat, respectively. The black rectangular and circular annotations indicate examples of sudden changes in evolution. (e) The time series at Galgeplaat location G1 is repeated. Here, also a linear fit based on the data between 1988 and 1990 and a linear fit based on the data after February 1990 are shown (before and after the 1990 storm). All data is shown versus a fixed reference frame (the 2014 mean sea level; MSL).

The bed level evolution at the highest location of Zuidgors Z3 (Fig. 2c) had a more gradual character than at the other two locations (Z1 and Z2). While these two lower sampling locations gained elevation in the long term, the variations in bed level (i.e., the bed level dynamics) reduced in time (i.e., their evolution became more gradual). This suggests that there may be a relation between the bed level dynamics and the bed elevation.

Figure 3

(a)–(h) Averaged bed level of two successive measurements versus the bed level rate of change between those measurements (considered as the bed level dynamics). The data is grouped per intertidal flat, see (i) for the location of intertidal flats a–h with the measurement locations indicated as black markers. The number of measurement locations for each flat (S), the total number of samples (N) and the coefficient of determination (({hbox {r}}^2)) are shown in the top right corner for each flat. The horizontal grid lines represent the boundaries of the vertical bins over which the percentiles are computed. Percentiles are only shown for vertical bins that contained at least 20 data points. All plots are vertically bounded by mean low water (MLW) and mean high water (MHW), apart from (h) that contained also a substantial amount of data points below MLW. Absence of data at the top of the tidal window in (d), (e), and (h) is due to the limited height of these flats, while absence of data at the bottom of the tidal window is solely due to absence of measurements at those elevations. All data is shown versus a fixed reference frame (the 2014 mean sea level; MSL).

With Fig. 3, the existence of such a relation is tested for intertidal flats in the Eastern Scheldt and Western Scheldt. Only locations with sufficient data are considered (Fig. 3i): seven intertidal flats in the Western Scheldt and only the Galgeplaat in the Eastern Scheldt. The bed level dynamics are computed by dividing the absolute elevation change of two successive measurements (after removal of the 5 year trend) by the time interval. The bed level dynamics are shown versus the average elevation of the two successive measurements. For each intertidal flat, data from all measurement locations were merged. The average duration between successive measurements was 40 days. For each bin of 0.5 m, the 25th, 50th, 75th, and 95th percentiles were determined.

The vertical trends are relatively consistent along the different percentiles. Consistently for all sites in the Western Scheldt (i.e., excluding Galgeplaat), the bed level dynamics reduce by approximately one order of magnitude from the lower to the higher part of the flat. In contrast, such a pattern in bed level dynamics was not observed for the Galgeplaat intertidal flat, where the bed was actually more dynamic above MSL (mean sea level). In the Western Scheldt 3–22% of the variance in the logarithm of the bed level dynamics is explained by elevation alone, whereas none of this variance is explained by elevation for the Eastern Scheldt samples (see coefficients of determination in Fig. 3). Therefore, there is a relation between the bed level dynamics and bed elevation in the Western Scheldt. However, as bed elevation alone explains only part of the variance, the bed level dynamics are also affected by other aspects. The role of the hydrodynamic forcing processes is studied in the next sections.

Inhomogeneous morphological impact of a storm event

The long-term data infer a spatiotemporally inhomogeneous character in the bed level dynamics of intertidal flats. To reveal the hydrodynamic mechanisms that drive this spatiotemporal inhomogeneity, measurements of wind, water levels, flow, waves, suspended sediment concentrations, and bed levels during a single storm event (20 November 2016) were analysed in detail. Time series of these processes at three stations at the transect across the Zuidgors intertidal flat (Fig. 1a) are shown in Fig. 4b–n. As visualised in Fig. 4a, these three stations consist of an ADCP in the channel, the low elevated frame ({hbox {F}}_{mathrm{L}}) at 0.4 m above MLW (mean low water), and the high elevated frame ({hbox {F}}_{mathrm{H}}) at 1.0 m above MSL.

The peak of the storm (hourly-averaged wind speeds up to 22 m/s, Fig. 4b) coincided with low water (Fig. 4c–e). The storm surge reached 1.3 m on top of the astronomical water level at Zuidgors. ({hbox {F}}_{mathrm{L}}) was submerged during the storm peak (Fig. 4d), while it would have been emerged in absence of the storm. The water depth at this station was 0.8–1.3 m for four successive hours around low water. By contrast, ({hbox {F}}_{mathrm{H}}) was emerged during the peak of the storm.

Under mild wind conditions (e.g., the first and last tide in Fig. 4), the velocities were flood-dominant, with maximum velocities occurring just before high water. The velocities were predominantly alongshore directed (not shown here), with spring-tidal velocities near the MLW line up to 1.5 m/s. The velocity magnitudes gradually decreased towards higher bed elevations of the flat.

The wind affected the flow on the flat substantially. While the flow in the channel reversed precisely at low water (Fig. 4h), the flow at ({hbox {F}}_{mathrm{L}}) reversed 2 h before (Fig. 4g), even though the two locations were only 200 m apart. At ({hbox {F}}_{mathrm{L}}), the velocity exceeded 0.7 m/s for 2 h around low water, while the velocity around low water is normally one order of magnitude smaller or even absent (by emergence of the bed). The flow at ({hbox {F}}_{mathrm{H}}) was also modified: it was almost solely ebb-directed in the tide preceding the storm peak (second tide in Fig. 4f). The wind was in ebb direction during the largest part of the tide preceding the storm peak, only at the end of this tide preceding the storm peak (when ({hbox {F}}_{mathrm{H}}) was emerged) the wind was in flood direction (Fig. 4b). The wind remained in flood direction for 9 h in the tide following the storm peak. The largest amplification of the flow occurred at ({hbox {F}}_{mathrm{L}}) (in flood direction), as ({hbox {F}}_{mathrm{H}}) was emerged during the peak of the storm. But even with the 0.7 m/s amplification at ({hbox {F}}_{mathrm{L}}) during low water, larger velocities occurred during calmer wind conditions (e.g., first tide in Fig. 4f). However, these larger velocities occurred for larger water depths and for shorter durations.

Waves on the flat were limited by the water depth for a large part of the storm. Significant wave heights exceeded 0.60 m at both frames. However, waves at ({hbox {F}}_{mathrm{H}}) were for a large portion of time absent (by emergence of the bed) or depth-limited (at most 50% of the water depth at this intertidal flat). Even though ({hbox {F}}_{mathrm{L}}) was submerged during the peak of the storm, waves were still depth-limited for more than three successive hours around low water.

Figure 4

Process measurements for four tides along a storm event in 2016 on the Zuidgors intertidal flat (Western Scheldt). (a) Measurement stations (see Fig. 1a for the location of the transect). (b) Hourly-averaged wind speeds and directions at Vlissingen meteorological station (location in Fig. 1). (c)–(e) Water depths, including the derived astronomical water depths (the difference is the storm surge). (f)–(h) Flow velocity magnitudes (flood and ebb directions indicated in Fig. 1). (i, j) Significant wave heights (the wave logger of ({hbox {F}}_{mathrm{H}}) was positioned 50 m downslope), including the significant wave height over depth ratio. (k, l) Suspended sediment concentrations at various distances from the bed. (m, n) Bed level changes relative to the initial elevations. The vertical black dashed lines indicate the peak of the storm. All time series are in CET (UTC+1).

Suspended sediment concentrations were largely affected by the storm. Suspended sediment concentrations were measured at 0.1 m and 0.6 m from the bed at both frames, and also at 1.1 m and 1.8 m from the bed at ({hbox {F}}_{mathrm{L}}) (Fig. 4k–l). The concentrations were relatively uniform over the water column (differences smaller than 30%). Concentration peaks are observed during flow peaks and during conditions for which the wave heights were depth-limited. These two conditions did not occur simultaneously for the tides with mild wind conditions. However, during the low water at the storm peak, the amplified flow and breaking waves coincided for hours at ({hbox {F}}_{mathrm{L}}). As a result, concentrations exceeded 1 g/L for 2.5 h. This concentration peak lasted one order of magnitude longer than the other observed concentration peaks.

The morphological changes by the storm event varied across the flat. The changes in bed level at ({hbox {F}}_{mathrm{H}}) were less than half a centimetre during these four tides. The storm did not cause a distinctive impact here (Fig. 4m). Conversely, the bed level at ({hbox {F}}_{mathrm{L}}) lowered 20 cm during the storm (Fig. 4n). This drop in elevation occurred within a 3 h window around low water at the peak of the storm. This 3 h window at ({hbox {F}}_{mathrm{L}}) coincided precisely with the windows during which the water depth was limited to 0.8–1.3 m, the velocity was amplified to (sim ) 0.7 m/s, the waves experienced depth-induced breaking, and the suspended sediment concentrations were above 1 g/L. A comparable simultaneous peaking of the forcing processes for hours did not occur at ({hbox {F}}_{mathrm{H}}).

Identifying patterns in forcing processes

The spatiotemporal variability in the forcing processes drove the inhomogeneity of the storm impact. In this section, we aim to unravel the long-term patterns of the hydrodynamic forcing processes across the Zuidgors tidal flat, which is relevant to understand the related bed level dynamics patterns. For this aim, the processes measured in the full 1-month measurement campaign at Zuidgors are analysed (Fig. 5) and relations between the processes are derived. A relation between the flow velocity and the bed elevation is not only assessed for this specific tidal flat, but is also tested for all intertidal areas in the Eastern Scheldt and Western Scheldt (Fig. 6). Finally, the distributions, relative timing, and importance of the various hydrodynamic processes across the intertidal flat are unravelled by integrally assessing the identified relations between the processes (Fig. 7).

The 1-month measurement campaign at Zuidgors shows also in the calm periods a substantial spatiotemporal variability in the bed level changes (Fig. 5; the vertical dashed lines highlight the storm discussed in Sect. 2.2). Significant wave heights were, apart from the highlighted storm event, below (sim ) 0.4 m and for most tides even below (sim ) 0.2 m. Figure 5a indicates that the bed level at ({hbox {F}}_{mathrm{H}}) was not only relatively stable during the storm, but also over the full month (less than 2 cm erosion over 30 days). Before the 20 cm erosion event, the bed level at ({hbox {F}}_{mathrm{L}}) was relatively stable (comparable evolution as at ({hbox {F}}_{mathrm{H}})). In the three days after the storm, half of the erosion (0.1 m) recovered. After two weeks, the bed returned to a relatively stable and similar evolution as at ({hbox {F}}_{mathrm{H}}). However, 25% of the erosion (5 cm) persisted.

Figure 5

Process measurements for the full 1 month campaign on the Zuidgors intertidal flat (Western Scheldt; see Figs. 1a and 4a for the locations). (a) Bed level changes for both frames relative to the initial elevations. (b) Significant wave heights at ({hbox {F}}_{mathrm{L}}) (near mean low water). (c) Water levels as measured in the channel (indicative for the full flat), the markers indicate the tidal range. (d) Flow velocity magnitudes at ({hbox {F}}_{mathrm{L}}), the markers indicate the flood and ebb peaks. (e) Suspended sediment concentrations 60 cm above the bed at ({hbox {F}}_{mathrm{L}}), the markers indicate the concentration that was exceeded for 2 h. (f)–(h) the maximum velocity of each tide versus the high water level. (i, j) the suspended sediment concentrations that were exceeded for 2 h versus the maximum velocity of each tide. The vertical black dashed lines indicate the peak of the storm of Fig. 4. All time series are in CET (UTC+1).

In the 1-month campaign, the tidal forcing was almost the smallest for the tides in the highlighted storm event (Fig. 5c; almost the smallest tidal range). The tidal range fluctuated over the month between 3.0 and 4.9 m. These fluctuations resulted mainly from spring-neap fluctuations (spring tide at the beginning, halfway, and end of the record), which were affected by storm surges (up to 1.3 m for the storm indicated with the vertical dashed lines in Figs. 4e and 5c). Spring-neap fluctuations were reflected in the peak flood velocities, with variations between 0.35 and 1.46 m/s (a ratio of maximum variation of 4.2) at ({hbox {F}}_{mathrm{L}}) (Fig. 5d). Fig. 5f–h show a proportionality between the high water level and the peak velocity for all measurement locations. On the flat, 78–88% of the variance in the peak velocities was explained by the high water level, in the channel this was 57%. The tidal range is a similar indicator for the peak velocities which explains 77–83% of the variance on the flat and 81% in the channel (not shown in these figures). The spring-neap fluctuations in the ebb peaks were much less, these varied only between 0.43 and 0.58 m/s (a ratio of maximum variation of 1.3; Fig. 5d). Nevertheless, the consequential variation in velocity asymmetry (relative magnitude of flood versus ebb velocities) across the spring-neap cycles did not affect the relation between peak velocity and high water level, as almost all tides were flood-dominant.

The suspended sediment concentrations varied with the spring-neap cycles (Fig. 5e; measured 0.6 m above the bed). This is especially visible in the values that were exceeded for 2 h each tide (less sensitive to outliers than the maxima). However, even though the spring tides in the middle of the record did not feature the largest tidal range (nor high water level) and peak velocities in the record, the concentrations were higher than during the preceding and following spring periods. This is expressed in Fig. 5j as well, which shows no clear relation (({hbox {r}}^2) of 0.01) between the concentrations that were exceeded for 2 h and the peak velocities at ({hbox {F}}_{mathrm{L}}). However, these concentrations did correlate well to the peak velocities at ({hbox {F}}_{mathrm{H}}) (({hbox {r}}^2) of 0.75).

Not only at Zuidgors do the peak flow velocities generally decrease for increasing bed elevations (Fig. 5f–h). For the full Western Scheldt and Eastern Scheldt (which include also tidal flats surrounded by channels) are the peak velocities generally smaller at higher bed elevations (Fig. 6; modelled average peak velocities). In the Western Scheldt 51% of the variance in the average peak velocities on the intertidal flats is explained by only the elevation, in the Eastern Scheldt (which has closed branches) this is 9%. Almost all data points in the Eastern Scheldt are below the linear fit of the Western Scheldt. There is thus an inhomogeneity in flow velocities over different elevations, within an estuary (spread across the linear fits), and between different estuaries.

Figure 6

Bed level, scaled along mean low water (MLW; 0) and mean high water (MHW; 1), versus the average peak flow velocity from a 1 month simulation. All points on the intertidal areas of the full (a) Western Scheldt (WS) and (b) Eastern Scheldt (ES) are shown. Linear fits are provided with coefficients of determination (({hbox {r}}^2)). The fits of both estuaries are shown in both graphs to allow comparisons.

Figure 7

Distributions in hydrodynamic forcing for variations in water depth and bed level. (a) Water level time series over a single tidal cycle (29 November 2016) at the Zuidgors intertidal flat (Western Scheldt). The horizontal lines indicate 20 cm bins. (b) The corresponding probability distribution of the water level across these bins. (c) The probability distribution of the water level over a full year (2016). (d) Water depth versus flow velocities at ({hbox {F}}_{mathrm{L}}) (10 min interval). The wind-induced flow during the storm event of Fig. 4 is marked with the ellipse. (e) Water levels at which the velocities peak at ({hbox {F}}_{mathrm{L}}), versus the high water of each tide. The storm of Fig. 4 is indicated. Both axes are scaled along mean low water (MLW; 0) and mean high water (MHW; 1). (f) Water depth versus the wave-induced shear stresses at ({hbox {F}}_{mathrm{L}}) (10 min interval). (g) Significant wave height versus water depth measured at ({hbox {F}}_{mathrm{L}}) (10 min interval), with the wave-breaking index (gamma ) indicated. (h) For a range in bed levels (scaled along MLW and MHW) the probability that the water depth is within a certain 20 cm bin is indicated with the colours, using the distribution of (c). The diagonal dotted region indicates the water depths at which the tidal velocities peak, using the average and standard deviation of (e). The range of depths at which the wave-induced shear stress peaks, based on (f), are indicated on the left. The vertical dashed lines show the position of the frames.

Also over long time scales, the distributions of both the occurrence and strength of the forcing processes vary across intertidal flats. The remainder of this section focusses on unravelling the relative importance of the tide, waves, and wind over different water depths and bed elevations at the Zuidgors intertidal flat (Fig. 7). The result is implemented in a schematic diagram, with the importance of the processes visualized over both bed elevation and water depth (Fig. 7h). The background colour represents the probability of occurrence of a water depth at a certain bed elevation (i.e., the relative duration). This probability of occurrence of a water depth results directly (water level–bed level) from the water level distribution (Fig. 7a, b). Here, the distribution over a full year is considered, to include surge and spring-neap fluctuations (Fig. 7c). Even with these fluctuations, the probability of occurrence of a water level near MLW and MHW (mean high water) is the largest. There is hence a variation in probability of occurrence of a certain water depth across the intertidal flat (Fig. 7h).

Tidal flow velocities generally decrease with water depth (Fig. 7d). Outliers are part of the wind event of Fig. 4, illustrating the effect of the wind. At large water depths, smaller velocities occur because of the reversal of the tidal flow. The average water level at which the tidal flow velocities peak equals (0.88{pm } 0.10) (scaled between MLW and MHW) at ({hbox {F}}_{mathrm{L}}) and almost identically (0.93 {pm } 0.10) at ({hbox {F}}_{mathrm{H}}) (Fig. 7e). Therefore, the water depth at which tidal velocities peak decreases almost linearly with an increase in bed elevation (the dotted region in Fig. 7h), which goes together with a decrease in peak velocity magnitude (Figs. 5f–h, 6). Furthermore, the higher the water level at which tidal flow velocities peak (i.e., larger water depths), the larger the magnitude of these peak flow velocities. This is a consequence of (1) an increase of the water level at which the tidal flow velocity peaks for an increase of the high water level of the tide (({hbox {r}}^2 = 0.85) at ({hbox {F}}_{mathrm{L}}); Fig. 7e) and (2) larger peak flow velocities for higher high water levels (especially on the flat, Fig. 5f–g). Peak flow velocities decrease hence both for higher bed elevations and smaller water depths (illustrated by the dot size in Fig. 7h).

Waves are limited by the water depth. At this location, the significant wave height does not exceed half the water depth (Fig. 7g), with the same wave-breaking index at ({hbox {F}}_{mathrm{H}}) (not shown in the figure). Within the 16 months of wave measurements, the maximum wave-induced shear stress occurred for water depths smaller than 1.2 m (Fig. 7f), which is roughly twice the maximum observed significant wave height (Fig. 5b). These limited water depths for which waves are most important occur the longest at the lowest part of the intertidal flat (Fig. 7h).

Inhomogeneous impact on benthic macrofauna

As bed level dynamics may affect benthic macrofauna (and the other way around), we investigate whether the spatiotemporal inhomogeneity of storm impacts has consequences for the benthic macrofauna on the Zuidgors intertidal flat. Figure 8 presents the logarithm of the abundance (mean and standard deviation) of benthic macrofauna before and after two storms in 2017 (in October and November, respectively; i.e., different storms than the one of Fig. 4) over six stations along a cross-shore transect of 210 m on the Zuidgors intertidal flat (1 km west of the frames of Fig. 4). During both storm events, the hourly-averaged wind speed exceeded 15 m/s for several hours.

Figure 8

The abundance of benthic macrofauna (number of individuals per ({hbox {m}}^2)) across six stations equally spaced in emersion time on the Zuidgors intertidal flat (Western Scheldt). Benthic macrofauna were sampled around the 5 October 2017 and the 23 November 2017 storm events, with at most 10 days between the before storm and the after storm measurements. The error bars (one standard deviation above and below) are based on three replicates.

In general, the abundance increased along the cross-shore transect with a higher emersion time (i.e., higher bed elevation). Storm events imposed substantial reductions in abundance. These changes in abundance occurred spatially inhomogeneous, just as observed for the morphological changes (e.g., Fig. 4m–n). For example, the highest two stations (66% and 80% emersion time) had an almost equal quantity of benthic macrofauna before the first storm, whereas the reduction by the storm was four times larger at the 66% emersion time station than at the 80% emersion time station. In contrast to the morphological observations, almost no ecological recovery was observed over the time scale of a month (abundance after October storm was similar to the situation before November storm). Furthermore, the decrease in abundance due to the second storm was smaller compared to the decrease due to the first storm.

Source link

Leave a Reply

Your email address will not be published. Required fields are marked *