Volume 56, Issue 7 e2020WR027822
Research Article
Free Access

Three-Dimensional Flow Structures and Morphodynamic Evolution of Microtidal Meandering Channels

Alvise Finotello

Corresponding Author

Alvise Finotello

Center for Lagoon Hydrodynamics and Morphodynamics, University of Padova, Padova, Italy

Department of Geosciences, University of Padova, Padova, Italy

Now at Department of Environmental Sciences, Informatics, and Statistics, Ca' Foscari University of Venice, Venice, Italy

Correspondence to: A. Finotello and A. D'Alpaos,

[email protected];

[email protected]

Search for more papers by this author
Massimiliano Ghinassi

Massimiliano Ghinassi

Center for Lagoon Hydrodynamics and Morphodynamics, University of Padova, Padova, Italy

Department of Geosciences, University of Padova, Padova, Italy

Search for more papers by this author
Luca Carniello

Luca Carniello

Center for Lagoon Hydrodynamics and Morphodynamics, University of Padova, Padova, Italy

Department of Civil, Environmental, and Architectural Engineering, University of Padova, Padova, Italy

Search for more papers by this author
Enrica Belluco

Enrica Belluco

Department of Civil, Environmental, and Architectural Engineering, University of Padova, Padova, Italy

Search for more papers by this author
Mattia Pivato

Mattia Pivato

Center for Lagoon Hydrodynamics and Morphodynamics, University of Padova, Padova, Italy

Department of Civil, Environmental, and Architectural Engineering, University of Padova, Padova, Italy

Search for more papers by this author
Laura Tommasini

Laura Tommasini

Department of Geosciences, University of Padova, Padova, Italy

Search for more papers by this author
Andrea D'Alpaos

Corresponding Author

Andrea D'Alpaos

Center for Lagoon Hydrodynamics and Morphodynamics, University of Padova, Padova, Italy

Department of Geosciences, University of Padova, Padova, Italy

Correspondence to: A. Finotello and A. D'Alpaos,

[email protected];

[email protected]

Search for more papers by this author
First published: 20 June 2020
Citations: 21

Abstract

The planform evolution of tidal meanders is driven by interactions between channel morphology and periodically reversing tidal flows, which feed back into the development of erosional and depositional patterns. However, the paucity of quantitative data has so far undermined detailed analyses about the geomorphic effects of tidal flows within tidal meanders. Here we aim to bond the morphodynamic evolution of tidal meanders with the structure of three-dimensional flow that shapes them. By means of an acoustic Doppler current profiler, we have surveyed the flow fields over three different tidal meandering channels in the salt marshes at San Felice (Venice lagoon, Italy), each characterized by distinct planform morphology and evolutionary dynamics. Mutually evasive paths followed by the maximum ebb and flood streamwise velocities determine periodic changes in both the position and the orientation of curvature-induced cross-stream flows. These secondary flows can be locally disrupted by patches of submerged vegetation, especially in meanders of small size, with direct implications for the morphodynamic evolution of meander bends. The latter is further affected by flow separation in sharp bends. Flow separation effectively reduces channel width, enhancing bank erosion due to increasing flow velocities. Moreover, it creates low-velocity zones of recirculating flows at both the inner and outer banks, thereby promoting the formation of point bars and concave-bank benches, respectively. By relating three-dimensional flow structure to patterns of channel change, our results provide a first step to unravel the relation between flows and forms within tidal meanders, whose planform characteristics may differ greatly from their fluvial counterparts.

Key Points

  • Acoustic measurements of 3D flow structures in tidal meanders highlight pronounced asymmetries in the ebb and flood velocities
  • Flow separation in sharp bends, controlled by the radius-width ratio, can lead to unidirectional flow in some parts of the bar deposits
  • Aquatic vegetation can influence the evolution of tidal meanders by disrupting and limiting the transfer of curvature-induced helical flow

1 Introduction

Meandering channels are widespread along coastal areas affected by the periodic action of tides and exert a prominent control on the ecomorphodynamic evolution of these landscapes by mediating water, sediment, and nutrient fluxes therein (Hughes, 2012; Marani et al., 2002). Moreover, tidal meanders influence both the sedimentology and the stratigraphy of the intertidal platforms they typically cut through, thus critically controlling the architecture of tidal sedimentary deposits (Choi, 2011; Cosma et al., 2019; D'Alpaos et al., 2017; Ghinassi, Brivio, et al., 2018).

Although it has long been recognized that flow in tidal meanders is three-dimensional (3D), direct field measurements of flow structures through tidal meander loops to date are surprisingly scarce (Leeder & Bridges, 1975; Leopold et al., 1993). This is mostly due to the assumption that flow fields in tidal meanders resemble those observed in their fluvial counterparts, despite several hydrodynamic differences suggesting otherwise. Among the latter, the periodic reversal of tidal flows, the typically asymmetric character of ebb and flood tides, and the fact that maximum velocities in tidal channels do not correspond to bankfull conditions are paramount (Bayliss-Smith et al., 1979; Fagherazzi et al., 2008; Fagherazzi & Furbish, 2001; French & Stoddart, 1992; Kearney et al., 2017).

It is well known that flow dynamics in river bends are driven by the imbalance between the centrifugal force and the lateral pressure gradient created by the curvature-induced outer bank superelevation of the water surface (Engelund, 1974; Prandtl, 1926; Rozovskii, 1957). This imbalance produces secondary (i.e., cross-stream) flows, directed inwards and outwards in the near-bed and near-surface zone, respectively. Downstream advection of cross-stream circulations operated by the main streamwise flow gives rise to a characteristic curvature-induced helical flow, a phenomenon that has been extensively documented in a variety of field (Dietrich & Smith, 1983; Dinehart & Burau, 2005a; Frothingham & Rhoads, 2003), laboratory (Blanckaert, 2011; Liaghat et al., 2014), and numerical studies (Blanckaert & de Vriend, 2003; Bridge & Jarvis, 1982; Ferguson et al., 2003).

On the one hand, the same mechanisms are likely to operate also in tidal meanders, giving rise to migration dynamics which are, on average, similar to those observed in river bends (Finotello et al., 2018). On the other hand, however, site-specific processes and meander evolution different from those of river bends might arise due to intrinsic hydrodynamic and morphological differences between tidal and fluvial landscapes, including changes in tidal asymmetries and higher spatial density of tidal channels (Finotello, Canestrelli, et al., 2019; Finotello et al., 2020; Marani et al., 2003).

Previous attempts to link the planform evolution of tidal meanders with the characteristics of tidal flows were mainly focused on the effects that bidirectional, mutually evasive pathways of ebb and flood flows have on meander planforms (Ahnert, 1960; Barwis, 1978; Fagherazzi et al., 2004; Solari et al., 2002), as well as on the meander bed morphologies (Finotello, Canestrelli, et al., 2019; Solari et al., 2002; Tambroni et al., 2017). In contrast, very few studies have attempted to relate patterns of 3D flows to the planform evolution of individual meander bends (Ghinassi, D'Alpaos, et al., 2018; Leeder & Bridges, 1975). In fact, 3D flow distributions in tidal meanders have been frequently overlooked due to a supposed, yet hardly proven, planform similarity with fluvial meander bends (Finotello et al., 2020), and a close investigation on the relationship between flow structure and different types of meander planform evolution, together with the related depositional patterns, is still missing.

In this paper, we tackle this issue based on direct acoustic flow measurements carried out throughout three different meander bends in the microtidal lagoon of Venice (Italy). Our main goal is to determine and analyze the structures of 3D flow fields in microtidal meander bends and to link these structures to meander migration via bank erosion and point bar deposition, highlighting possible analogies and differences with their fluvial counterparts. Toward this goal, we directly examine and compare 3D velocity fields in tidal meanders characterized by different planform geometries and evolution dynamics. The focus on 3D velocity fields and channel planform evolution in tidal meanders is novel, because previous field studies have almost exclusively focused either on fluvial and tidally-influenced fluvial bends (Dinehart & Burau, 2005a2005b; Ferguson et al., 2003; Frothingham & Rhoads, 2003; Keevil et al., 2015; Nanson, 2010; Vermeulen et al., 2015) or on large-scale circulations in salt marsh channel networks (Sullivan et al., 2015).

2 Geomorphological Setting and Study Cases

The present study investigates three tidal meandering channels wandering through salt marshes in the Venice lagoon (Italy) (Figure 1a). The Venice lagoon formed over the last 7,500 years during the Holocene transgression (Tosi et al., 20122017; Zecchin et al., 2009) and with an area of about 550 km2 currently represents the largest Mediterranean brackish water body. Three inlets—namely, Lido, Malamocco, and Chioggia from north to south—connect the Adriatic Sea to the lagoon, which is subjected to a semidiurnal, microtidal regime, with an average tidal range of about 1.0 m and peak tidal amplitudes of about 0.75 m around mean sea level (MSL) (D'Alpaos et al., 2013). Over the last centuries, the Venice lagoon has experienced several anthropogenic-induced modifications. Among these modifications, the diversion of all major freshwater rivers, the excavation of navigable channels, and the stabilization of the inlets through the building of jetties have caused the most important morphodynamic effects, turning the lagoon into a sediment-starving, ebb-dominated system (Carniello et al., 2012; D'Alpaos, 2010; Ferrarin et al., 2015). This set a negative sediment budget, leading to the deepening of tidal flat areas and causing a severe loss of salt marsh surfaces (Carniello et al., 2009; Sarretta et al., 2010; Tommasini et al., 2019).

Details are in the caption following the image
Overview of the study area. (a) Landsat image of the Venice lagoon. (b) The San Felice salt marshes, located in the northern part of the Venice lagoon. The locations of the three study cases are also highlighted, together with the position of the “Treporti” tide gauge (image ©Google, Landsat/Copernicus). (c) Detailed view of the translating asymmetric (TA) and straightening cuspate (SC) study cases. (d) Detailed view of the expansional-symmetric (ES) study case. Dotted lines in (c) and (d) denote the portion of the channel along which flow measurements have been conducted. The directions of ebb and flood flows are also highlighted (image ©Google, Landsat/Copernicus).

Here we focus on three tidal meandering channels within the San Felice salt marshes (Figure 1b), located in the northern and most naturally preserved part of the Venice lagoon (Marani et al., 2003; Silvestri et al., 2018). The salt marshes in San Felice are characterized by an average elevation of 0.26 m above MSL and are typically colonized by different associations of halophytes, among which one can find Spartina maritima, Limonium narbonense, Sarcocornia fruticosa, and Juncus maritimum (Belluco et al., 2006). Sediment grain size in all the analyzed channels varies from mud to medium-fine sand, and shell fragments are typically found at the channel bottom (Brivio et al., 2016; Ghinassi, D'Alpaos, et al., 2018). Biogenically enhanced sediment cohesion, combined with a sediment transport regime dominated by isolated events of high suspended loads associated to wind wave-driven resuspension (Carniello et al., 2009; Finotello, Canestrelli, et al., 2019), produces poorly developed bedforms that mainly consist of centimeter-scale ripples (see Ghinassi, D'Alpaos, et al., 2018).

The selected study cases are characterized by different planform morphologies and styles of evolution that we reconstructed for the past 50 years based on available historical aerial photos dated at 1968, 2012, and 2018 (Figures 1c, 1d, and 2).

Details are in the caption following the image
Planform evolution of the study bends from 1968 to 2018. (a–d) Evolution of the translating asymmetric (TA) study case. (e–h) Evolution of the straightening cuspate (SC) study case. (i–l) Evolution of the expansional-symmetric (ES) study case. Green, yellow, and red lines denote the position of channel banks in 1968, 2012, and 2018, respectively. Light blue and light red colors in the right panels represent, respectively, accretional and erosional areas from 1968 to 2012, whereas dark blue and dark red colors denote areas of accretion and erosion from 2012 to 2018. Aerial image sources: Compagnia Generale Ripreseaeree Parma—Ortofoto del territorio regionale Veneto, 1968, flight survey (http://mapserver.iuav.it/website/foto_ aeree/) for (a,e,i); Google, Landsat/Copernicus for all other panels.

2.1 “TA” Meander

The first study case is a simple asymmetric meander loop (sensu Brice, 1974), skewed in the landward direction. It is characterized by a radius of curvature (R) of about 17 m and a width (B) varying between 15 and 25 m (average value urn:x-wiley:00431397:media:wrcr24721:wrcr24721-math-0001 = 20 m). Given the small ratio between radius of curvature and channel width ( urn:x-wiley:00431397:media:wrcr24721:wrcr24721-math-0002 = 0.85), this meander represents a classical example of a sharp meander bend (Blanckaert, 2011). Since 1968, this channel exhibited a translational planimetric transformation, maintaining an approximately constant sinuosity σ = 2 while the bend apex migrated seaward at a rate of 20 to 30 cm/year (Figures 2a to 2d). Hereinafter, we will refer to this translating asymmetric study case as TA.

2.2 “SC” Meander

The second study case is located close to the TA meander (Figures 2e to 2h) and consists of two similar adjacent meanders characterized by cuspate banklines, a pattern that has often been recognized as peculiar of the tidal environment (Ahnert, 1960; Dalrymple et al., 2012; Finotello et al., 2020). Such a planform cannot be ascribed to any of the classes proposed by Brice (1974). Prior to 1968, the channel was directly connected to the TA meander through a minor creek (named “MFC” in Figure 2e) that was abandoned and mud infilled when a channel piracy occurred south of the study meanders (Figures 2e and 2f). Following this piracy event, the investigated channel exhibited a peculiar planform evolution that caused a progressive decrease in the channel sinuosity, from σ = 1.44 in 1968 to σ = 1.25 nowadays, and the overall seaward shift of the study bends. During these transformations, which took place at an average migration rate of 22 cm/year, the mean curvature radius increased from R = 20 m in 1968 to R = 30 m nowadays, while the average channel width slightly increased from urn:x-wiley:00431397:media:wrcr24721:wrcr24721-math-0003 = 15 m to urn:x-wiley:00431397:media:wrcr24721:wrcr24721-math-0004 = 20 m. Hence, the current radius-width ratio is equal to urn:x-wiley:00431397:media:wrcr24721:wrcr24721-math-0005 = 1.5. In the following paragraph, we will refer to this straightening cuspate study case as SC.

2.3 “ES” Meander

The last study case consists of a symmetrical meander (sensu Brice, 1974) found along a tidal channel displaying an E-W trend (Figures 2i to 2l). The bend is about urn:x-wiley:00431397:media:wrcr24721:wrcr24721-math-0006 = 10 m wide, with a sinuosity equal to σ = 1.45 and a mean radius of curvature of about R = 29 m ( urn:x-wiley:00431397:media:wrcr24721:wrcr24721-math-0007 = 2.9). Over the last 50 years, the bend showed an expansional behavior (sensu Daniel, 1971; Jackson, 1976), characterized by a nearly symmetric growth of the point bar (Figure 2l), with the apex migrating at about 15 cm/year and following a NE-SW trajectory. We hereinafter refer to this bend as bend ES due to its expansional behavior and symmetric planform.

3 Methods

3.1 Data Collection

We surveyed both the flow field and the channel bathymetry employing a SonTek RiverSurveyor M9 acoustic Doppler current profiler (ADCP) mounted on a “Hydroboard” floating platform operated from the channel banks. ADCPs represent the current standard for flow measurements in large-scale open water systems (Dinehart & Burau, 2005a; Parsons et al., 2013; Vermeulen et al., 2014), as the use of such technology has increased rapidly in recent years primarily driven by advances in acoustic technology and signal processing (Parsons et al., 2013). Besides providing accurate river discharge measurements (Le Coz et al., 2008; Lee et al., 2014; Nihei & Kimizu, 2008; Sassi et al., 2012), ADCPs have proved to be a useful means in studies of complex geophysical flows (Hackney et al., 2015; Kästner et al., 2017; Keevil et al., 2015; Kostaschuk et al., 2004; Lane et al., 2008; Simmons et al., 2020; Thorne & Hanes, 2002).

The M9 ADCP is a nine-beam system with two sets of four profiling beams that measure samples using different frequencies. Each sample is divided vertically into separate cells, also referred to as bins. The cell height is dynamically adjusted depending on water depth and speed in order to optimize performance and resolution. An additional 0.5-MHz vertical acoustic beam (echo sounder) provides a precise bathymetric survey. The presence of an integrated Differential Global Positioning System (DGPS) ensures a submeter precision in positioning data. Streamflow measurements carried out using the SonTek M9 have been recently tested and validated by the U.S. Geological Survey (USGS) Office of Surface Water (Boldt & Oberg, 2015).

The 3D velocity data were collected within each study case during three separate daily measurement campaigns: 20 June 2016 for study case TA; 19 July 2016 for study case SC; and 16 September 2016 for study case ES (Figure 3). All the campaigns correspond to spring tide conditions and were carried out in days characterized by low wind speeds, so that the effects of both wind setup and wave-induced sediment resuspension in the study area would be negligible. As a result, water turbidity during the measurements was very low, making the bottom of the channel mostly visible, especially in shallower areas.

Details are in the caption following the image
Planimetric view of ADCP transects and water levels measured at the “Treporti” tide gauge (see Figure 1b). (a) Translating asymmetric (TA) study case. (b) Straightening cuspate (SC) study case. (c) Expansional-symmetric (ES) study case. The cross sections are numbered in ascending order moving seaward, while the “.a” and “.b” suffixes denote the left and right channel bank according to a river-like reference system, with velocities being positive when flowing seaward (i.e., ebb flows). The number (n) of repeated surveys for each cross section is also reported. Red and blue colors in water-level plots denote the duration of the ebb and flood measuring campaigns for each study case. All the study cases were surveyed in 2016. Water levels are referred to the current mean sea level (MSL). (image ©Google, Landsat/Copernicus).

Velocity measurements were obtained at both the flood and the subsequent maximum ebb velocity stages (Figure 3) for a subset of cross sections oriented orthogonally to the local pattern of channel curvature. A total number of four and six transects were surveyed for the TA and SC bends, respectively, whereas five cross sections were measured within the ES bend. The paths of all the surveyed transects are reported in Figure 3.

In comparison with flow measurements in fluvial and tidal-fluvial settings, additional operational limitations due to flow unsteadiness need to be overcome when surveying salt marsh tidal channels. Particularly, several transects are usually needed in order to obtain robust estimates of the mean velocity fields and turbulence-averaged velocity in rivers (Dinehart & Burau, 2005a) where, however, the temporal variability of flow discharges is lower and high-velocity stages last for longer periods, compared with tidal channels. In the latter case, the maximum ebb and flood velocity stages occur when the marsh platform is just completely flooded and almost completely drained, respectively, as a consequence of the characteristic geomorphic structure of the marsh platform (Bayliss-Smith et al., 1979; Boon, 1975; Fagherazzi et al., 2008). Indeed, the level of the marsh surface provides a threshold at which both velocities and discharges can increase shortly before and shortly after high tide. This leads to two distinct discharge peaks, which occur when tidal levels are just above (during the flood phase) or below (during ebb) the elevation of the marsh platform (Pethick, 1980). Therefore, the maximum velocity stages last only for short time spans, which are typically in the order of 30 min. In view of the above, we repeated individual ADCP transects along each study case bend two to four times (see Figure 3). Since the system responds in a similar fashion to tides of different amplitudes (Kearney et al., 2017), and even though changes in tidal amplitude will somewhat affect velocity magnitudes, the patterns of flow, which are the focus of this study, should remain similar.

3.2 Data Analysis

First, echo sounder data, specifically collected on an approximately regular 2-m grid, were spatially interpolated in a Geographic Information System (GIS) environment, providing us with accurate channel bathymetries (Figure 4). Then, flow velocity data were analyzed using the USGS Velocity Mapping Toolbox (VMT) (Parsons et al., 2013) to investigate cross-sectional and depth-averaged flow fields. In order to reduce the data noise and minimize turbulent fluctuations, data from repeated transects were first projected to a mean, linear cross section and interpolated to a uniform grid, and the average velocity magnitude and direction were then recomputed at each grid node (Dinehart & Burau, 2005b). The resulting velocity field was then divided into three different components, namely, the streamwise velocity u (perpendicular to the cross section), the cross-stream velocity v (parallel to the cross section), and the vertical velocity w (Lane et al., 2000).

Details are in the caption following the image
Bathymetries of the study case bends, as obtained from the interpolation of ADCP echo sounder data. Nomenclature of point bars and concave-bank benches follows that adopted in the text. Elevation is in meters above mean sea level (MSL).

Depth-averaged velocity (DAV) plots provided us with valuable information about flow structures, flow recirculation zones, and their interactions with channel banks and allowed us to detect shifts in the position of the maximum velocity filament (MVF) between the ebb and flood phase (see Figure 5). Furthermore, VMT allowed us to explore patterns of secondary (i.e., cross-stream) velocities. We adopted the “zero-net secondary discharge” (ZSD) convention, which is generally best suited for meander bends (Lane et al., 2000; Parsons et al., 2013). According to the ZSD definition, a new linear cross section is computed so that no net secondary discharge flows through the entire cross section. The component of velocity perpendicular (i.e., primary velocity, Vp) and parallel (i.e., secondary velocity, Vs) to the new rotated cross section was finally computed and visualized in the cross-sectional plane. The use of a different rotation scheme, namely, the Rozovskii method (see Parsons et al., 2013, for a detailed description; Rozovskii, 1957), does not fundamentally change the observed patterns of primary and secondary velocities for all the study bends (see Supporting Information, Figures S1, S2, and S3), thus ensuring the validity of the results and discussions that will be presented below.

Details are in the caption following the image
Plots of depth-averaged velocities during the ebb (left) and the flood (right) for each study case. Names of individual cross sections are also reported, where the “.a” label denotes the left channel bank in the seaward direction.

No corrections were made to the ADCP data to account for possible stratification effects due to salinity and temperature gradients, given the absence of significant freshwater supply and the well-mixed character of the Venice lagoon, especially close to the inlets and in channels of small size such as those considered here (Amos et al., 2016; Pivato et al., 2020). Consistent with these observations, as well as with previous field studies (see Pivato et al., 2018), direct measurements at the study sites highlighted uniform vertical temperature profiles. Hence, we can effectively rule out the chance for the observed secondary flow dynamics to be partially driven by stratification effects (e.g., Azpiroz-Zabala et al., 2017; Sumner et al., 2014).

4 Results

We will hereinafter assume a river-like reference system, with velocities being positive when flowing seaward (i.e., ebb tides) and cross-sectional views assuming a seaward-directed observation point. Based on this assumption, downstream and seaward, as well as upstream and landward, are used as synonymous. The term inner bank will be used to indicate the convex bank and the term outer bank to indicate the concave bank of a meander bend.

4.1 Bathymetric Data

The TA bend exhibits an asymmetric bottom configuration, with water depths of about 3 and 1.5 m in the pool and riffle zones, respectively (Figure 4a). A single rounded pool is observed, which is located close to the inner bank, slightly landward of the bend apex. There, a topographic high at the outer bank resembles the concave-bank benches (i.e., counter point bar) that were first described by Hickin (1979). A well-developed point bar is observed downstream of the bend apex that joins another concave-bank bench located immediately further downstream.

Along the SC meander, the channel thalweg varies in depth from 2 m in riffles to more than 3 m at pools (Figure 4a). The latter appear as distinct elongated morphologies located immediately upstream and downstream of the cuspate inner bank apex, suggesting a periodicity in the thalweg topography. Concave-bank benches are found along both banks, whereas no clear accreting point bars are observed, likely because only a limited portion of the bankline is convex in planform.

Finally, the bathymetry of the ES bend exhibits a single elongated pool zone, located in correspondence of the bend apex and characterized by a maximum depth of 1.8 m (Figure 4b). A well-developed point bar, sloped toward the channel axis, is found along the inner (convex) bank. A secluded shallow area is also observed on the landward side of the inner bank, which protrudes up to the channel axis. This bottom perturbation corresponds to a patch of submerged vegetation, namely, seagrass (Posidonia oceanica, Figure 4b). Although similar vegetation patches are present in the other study bends (see Figures 3 and 4a), this is the only case where the size of the patch is relevant compared with the channel cross section. In particular, channel depths and vegetation height are similar, causing the vegetation patch to be almost exposed at low tide.

4.2 Distributions of DAV

4.2.1 TA Meander

The TA meander exhibits complex patterns of DAVs (Figures 5a and 5b).

During the ebb phase (Figure 5a), the MVF (~40 cm/s) is steered toward the inner bank in the most landward part of the study bend (section named TA1). Moving seaward, the MVF detaches from the inner bank and moves toward the outer bank immediately downstream of the bend apex. The pronounced widening that characterizes section TA2 causes a strong flow separation (see also Figure 6), which in turn gives rise to a large outer bank DAV recirculation eddy that occupies about half of the total cross-sectional width. Downstream of s.TA2, the main streamflow continues seaward, giving rise, in s.TA3 and s.TA4, to a DAV field characterized by low velocities on the point bar top, with the MVF located at the outer bank.

Details are in the caption following the image
Flow velocities at the translating asymmetric (TA) study case. Secondary flows (Vs) are overimposed to color-coded plots of primary velocity (Vp). Left and right panels correspond to ebb and flood tides, respectively. Both Vp and Vs are computed according to the zero-net secondary discharge definition. Positive values of Vp correspond to seaward-directed flows. Cross-sectional view is from landward to seaward. Locations and names of each cross section are also shown in the corresponding panel, where the “.a” label denotes the left channel bank in the seaward direction.

In contrast, during the flood phase (Figure 5b), the MVF (~40 cm/s) is located at the inner bank (i.e., over the point bar) in the seaward portion of the bend (s.TA4). Moving landward, the MVF shifts toward the outer bank (s.TA3) and a large recirculation eddy forms in the inner portion of the channel at the bend apex (s.TA2), occupying one third of the total cross-section width. Beyond the apex, the MVF remains close to the external bank up to the meander landward endpoint (s.TA1).

4.2.2 SC Meander

This case study displays virtually symmetrical DAV fields for the ebb and the flood phase (Figures 5c and 5d).

During the ebb phase (Figure 5c), the MVF (~20 cm/s) flows along the right in sections SC1 and SC2, which are both characterized by the presence of a low-DAV core near the left bank. Downstream of cross section SC2, the MVF is located along the midchannel line, while a pronounced recirculation zone characterized by DAV lower than 5 cm/s affects the right bank (s.SC3). Further downstream (s.SC4 and s.SC5), zones characterized by low DAV match the concave-bank bench found along the right channel bank. In particular, both topographic steering and a pronounced recirculation zone are observed on the right-hand side of the SC4 cross section. Here, the MVF moves to the left bank of the channel and remains there until the outlet of the MFC is met, downstream of s.SC5. After that, the MVF shifts back to the right channel bank (s.SC6).

The DAV field of the flood phase (Figure 5d) is characterized by a nearly symmetrical velocity distribution in the SC6 cross section, where weak flow recirculation zones near the channel banks are also observed. Once past the outlet of the MFC, the MVF remains located on the right side of the channel, while low-velocity recirculation zones develop close to the left bank, immediately upstream of the MFC outlet (s.SC5, s.SC4, and s.SC3). Beyond the apex of the upstream bend, the MVF moves closer to the midchannel line (s.SC2) where it remains until the last surveyed section (s.SC1). In these last two cross sections, low DAVs are observed along the right bank, suggesting the possible presence of recirculation zones further upstream.

4.2.3 ES Meander

During the ebb (Figure 5e), the MVF is located in the middle of the ES1 cross section and displays maximum DAVs close to 17 cm/s. However, the distribution of DAVs is markedly asymmetrical, since the flow velocities at the right bank (>10 cm/s) are much higher than those observed along the left bank (<5 cm/s). Moving downstream to s.ES2, the maximum DAV values undergo a sharp reduction, and the velocity profile is now characterized by a low-velocity zone in the center of the channel (<5 cm/s) and two velocity maxima greater than 10 cm/s near both banks. Of these two maxima, only the one along the inner bank is maintained in s.ES3 and therein coincides with the MFV (~10 cm/s). On the contrary, DAVs close to zero are observed along the outer bank. The distribution of DAVs observed in s.ES3 is completely reversed in s.ES4, where the MVF (~10 cm/s) is located along the outer bank, while lower DAVs are found at the inner bank. Finally, the DAV profile in s.ES5 is pseudo-parabolic, with maximum velocities of about 15 cm/s in the middle of the channel and lower velocities along the banks, especially the inner one where DAVs are equal to 7 cm/s.

DAVs during the flood (Figure 5e) display a strongly asymmetric distribution at s.ES5, with maximum (~20 cm/s) and minimum (<2 cm/s) velocities found at the inner bank and outer bank, respectively. A similar DAV profile also characterizes s.ES4, although the maximum velocities are lower (~13 cm/s) and the MVF is slightly detached from the inner bank.

This detachment is even more evident in section ES.3, where the MVF (~12 cm/s) is positioned along the midchannel line and lower velocities are found at the inner bank. Further upstream (s.ES2), the DAV profile displays a pseudo-parabolic trend, despite being skewed toward the outer half of the channel, where the maximum velocities are higher than 15 cm/s. A much more irregular profile is shown by the ES5 cross section, where a low-DAV region (<8 cm/s) along the midchannel line separates two zones characterized by maximum DAVs of 17 and 11 cm/s along the outer and inner bank, respectively. Such an irregular profile can be ascribed to the already mentioned patch of submerged vegetation that we observed in s.ES5 (Figure 4b).

4.3 Distributions of Cross-Sectional Velocity

4.3.1 TA Meander

The plot of 3D primary (Vp) and secondary (Vs) velocities for the TA1 cross section during the ebb (Figure 6a) shows almost homogeneous Vp in the order of 40 cm/s across the entire cross section. No secondary velocities of particular relevance are shown. A strongly different velocity field is observed for the TA2 cross section (Figure 6c), which is also characterized by a cross-sectional width and area almost two and three times larger than s.TA1, respectively. Here, however, only the inner half of the cross section is crossed by seaward-directed Vp, while negative Vp, corresponding to regions of separated, upstream recirculating flow, are evident at the outer channel half (s.TA2.a). Patterns of vectors of Vs illustrate that flow through the bend is markedly 3D, with pronounced counterclockwise helicity restricted to the seaward-directed freestream that occupies the inner channel half (s.TA2.b). On the contrary, very weak Vs are observed within the region of separated, landward-directed flow. The latter is divided from the high Vp flowstream by a shear zone with null Vp, wherein a weak clockwise secondary circulation is observed. Once past the bend apex, the pronounced pool-riffle morphology of the bend allows for channel narrowing to be offset without reducing Vp in s.TA3 and s.TA4 (Figures 6e and 6g). Here, the classical curvature-induced secondary circulation appears, whereby near-surface Vs move toward the outer bank while near-bed Vs are directed upbar (s.TA3.b and s.TA4.b).

During the flood, the flow field in s.TA4 is strongly asymmetrical as a result of the channel curvature downstream of the study bend (Figure 6h). The core of Vp is shifted toward the right bank (s.TA4.b), while a well-developed, clockwise secondary circulation cell affects the whole cross section. This secondary circulation cell most likely also involves the bar top, where, however, the limited water depths (<50 cm) allow only the near-bed, downbar-directed portion of the secondary cell to be observed.

Pronounced secondary circulations are observed in s.TA3 as well (Figure 6f). Here, however, the main clockwise secondary cell on the left portion of the channel (s.TA3.a) is counterbalanced by a counterclockwise-rotating cell on the right-hand side (s.TA3.b). This cell is similar to the outer bank cell typically observed in sharp river bends (Blanckaert, 2011; Nanson, 2010), and its interaction with the main cell generates plunging secondary flows in correspondence to the maximum Vp core. A main clockwise secondary cell is also observed in s.TA2 (Figure 6d), although it is restricted to the outer half of the channel where Vp are directed landward. In contrast, a weak counterclockwise secondary cell exists in the inner channel half, where separation of Vp occurs. This secondary cell rotates inwards and downwards in the distal part of the recirculation zone, but outwards and upwards in the middle of the channel, where the flow is reintroduced into the main helical streamflow. This makes the pressure distribution in s.TA2 highly nonhydrostatic. Weak Vs are finally observed in s.TA1 (Figure 6b), even though high Vp, much more sustained than in s.TA2 due to high flow confinement, are preferentially located in the left part of the cross section (s.TA1.a).

4.3.2 SC Meander

Ebb velocity distributions in both s.SC1 and s.SC2 are characterized by a clockwise-rotating secondary cell in the inner right portion of the cross section (i.e., s.SC1.b and SC2.b, close to the bend apex) corresponding to the bulk of maximum primary velocity (Figures 7a and 7c). Conversely, at the outer portion of both cross sections, regions of low to negative Vp match the positions of counterclockwise secondary cells (s.SC1.a and SC2.a). The latter produce a pronounced upwelling toward the shallow concave-bank bench found along the left bank (see also Figure 4a). Downstream of the bend apex (s.SC3), a main counterclockwise secondary cell occupies the central left portion of the cross section, and a smaller cell rotating in the opposite direction is found at the left bank (s.SC3a). No significant secondary cell exists on the right-hand side (s.SC3.b) where primary flows are recirculated upstream (Figure 7e). Similar patterns are observed in s.SC4. Here, however, a much stronger flow upwelling is observed from the main secondary cell on the left to the flow recirculation zone and eventually toward the concave-bank bench top, on the right (s.SC4.b, Figure 7g). In the SC5 cross section, the bulk of maximum Vp is found near the left bank (s.SC5. a), while the main secondary flow cell is located on the opposite side, close to the inner bend apex (s.SC5.b, Figure 7i). Finally, a well-developed, midchannel Vs cell is observed in s.SC6. This cell rotates clockwise and is flanked, on both sides, by two smaller cells rotating in the opposite direction (Figure 7k).

Details are in the caption following the image
Flow velocities at the straightening cuspate (SC) study case. Secondary flows (Vs) are overimposed to color-coded plots of primary velocity (Vp). Left and right panels correspond to ebb and flood tides, respectively. Both Vp and Vs are computed according to the zero-net secondary discharge definition. Positive values of Vp correspond to seaward-directed flows. Cross-sectional view is from landward to seaward. Locations and names of each cross section are also shown in the corresponding panel, where the “.a” label denotes the left channel bank in the seaward direction.

During the flood, convergence of Vp toward the middle of the channel produces secondary flows in the form of two counter-rotating cells directed, respectively, inwards near the water column surface and outwards near the bottom (Figure 7l). Similar patterns are observed in s.SC5, even though the Vp maximum shifts toward the right channel bank, thereby producing secondary flows oriented toward the top of the concave-bank bench (Figure 7j).

Pronounced clockwise-rotating secondary cells that occupy most of the cross-sectional area are observed both in s.SC4 and s.SC3. These secondary cells cause the advection of bulks of high Vp, located in the upper right portion of the cross section, toward the channel bottom. As a result, poor mixing and strong shear are observed between zones of high primary velocity to the right and low downstream-recirculating primary velocity to the left (Figures 7f and 7h).

Further upstream (s.SC2), primary velocities are more evenly distributed over the entire cross section. A major clockwise secondary cell occupies the midchannel region, while two counter-rotating cells are found near both the right and the left bank. While the former is characterized by sustained Vs flowing toward the center of the channel, the latter separates further near the bank, giving rise to weak Vs oriented toward the top of the concave-bank bench (Figure 7d). Finally, the SC1 cross section is divided in half by two different zones characterized by opposite secondary velocities, which are directed in both cases toward the channel banks. While a dead-Vp zone is located on the right side of the cross section, more sustained primary velocities characterize the left side near the concave-bank bench (Figure 7b).

4.3.3 ES Meander

The relatively straight reach landward of s.ES1 does not produce any curvature-induced secondary circulation therein (Figure 8a). Further downstream (s.ES2), however, secondary circulations become much more pronounced, showing a main counterclockwise secondary cell directed outwards and inwards in the near-surface and near-bed region, respectively, and a smaller outer bank cell rotating in the opposite direction (Figure 8c). The highest Vp in s.ES2 are found at the inner bank, near the point bar top (Figure 8c). A qualitatively similar flow structure is also observed in the apex cross section (s.ES3). Here, however, both Vp and Vs are significantly lower than those observed in s.ES2, in spite of limited cross-sectional area and morphology (Figure 8e). This suggests a limited downstream transfer of the curvature-induced helical flow typically observed in meandering channels. Further downstream (s.ES4 and s.ES5), however, both Vp and, above all, Vs regain strength (Figures 8g and 8i). In particular, the secondary flow field corresponds to that observed in s.ES2, with a counterclockwise secondary cell characterized by velocities directed, respectively, toward the outer bank near the water surface and toward the inner bank (i.e., upbar) near the channel bed (Figures 8g and 8i). In addition, a smaller clockwise rotating outer bank cell is clearly visible in s.ES4.

Details are in the caption following the image
Flow velocities at the expansional-symmetric (ES) study case. Secondary flows (Vs) are overimposed to color-coded plots of primary velocity (Vp). Left and right panels correspond to ebb and flood tides, respectively. Both Vp and Vs are computed according to the zero-net secondary discharge definition. Positive values of Vp correspond to seaward-directed flows. Cross-sectional view is from landward to seaward. Locations and names of each cross section are also shown in the corresponding panel, where the “.a” label denotes the left channel bank in the seaward direction.

A flow structure completely mirrored to that observed during the ebb phase characterizes the ES5 cross section during the flood (Figure 8j). A counterclockwise secondary cell occupies most of the cross-sectional area, creating Vs oriented toward the thalweg (i.e., downbar) on the right-hand side of the channel, where the highest Vp are also found. Similar secondary patterns are observed in s.ES4, where both the main midchannel cell and the downbar-directed flow are much more pronounced than in s.ES5, thus causing the core of maximum Vp to plunge toward the channel thalweg (Figure 8h). At the bend apex (s.ES3), the flow structure is similar to that observed in the previous cross section, but it is characterized by much lower Vp and Vs (Figure 8f). In particular, Vp are negligible near the inner bank, where a weak flow recirculation is also observed. Further upstream (s.ES2), no traces of the main secondary cells are left (Figure 8d), pointing to a limited transfer of the secondary helical flow. Finally, a bulk of near-bed low primary velocity near the channel thalweg characterizes the flow structure in s.ES1, possibly due to the presence of the P. oceanica (see Figures 3 and 4a). A major counterclockwise secondary cell develops around this low-velocity region, while a small secondary outer bank cell and downbar-oriented flows characterize the outer and inner near-bank areas, respectively (Figure 8b).

5 Discussions

5.1 Flow Separation and 3D Flow Structures

A common feature of all case studies is the small ratio of radius of curvature to the channel width ( urn:x-wiley:00431397:media:wrcr24721:wrcr24721-math-0008), which is equal to 0.85, 1.75, and 2.9 for the TA, SC, and ES study bends, respectively.

In fluvial meanders, urn:x-wiley:00431397:media:wrcr24721:wrcr24721-math-0009 ratios lower than 2–3 are known to be associated with the growth of nonlinear hydrodynamic processes (Blanckaert, 2011; Blanckaert et al., 2010). In addition, when urn:x-wiley:00431397:media:wrcr24721:wrcr24721-math-0010 is lower than 2, flow separation may occur either at the inner or the outer bank, respectively, immediately downstream or upstream the bend apex (Blanckaert et al., 2013; Hickin, 1978; Hickin & Nanson, 1975; Hooke, 2013; Parsons et al., 2004; Rozovskii, 1957). Flow separation can create a zone of dead velocity that limits mixing and promotes deposition of fine sediment (Ferguson et al., 2003; Ferguson & Parsons, 2004), thus critically influencing meander migration (Finotello, D'Alpaos, et al., 2019; Hickin, 1974; Sylvester et al., 2019). Indeed, flow separation in our study case appears to be related to the urn:x-wiley:00431397:media:wrcr24721:wrcr24721-math-0011 ratio of the considered bend. Both the TA and the SC study bends, for which urn:x-wiley:00431397:media:wrcr24721:wrcr24721-math-0012 < 2, display large regions of flow separation during both the ebb and the flood (Figures 5a to 5d). In contrast, no clear flow separation is observed within the ES bend (Figures 5e and 5f). Limited flow separation is likely due to the relatively high urn:x-wiley:00431397:media:wrcr24721:wrcr24721-math-0013 of this bend, which is equal to 2.9. While a dead-velocity zone is observed at the inner bank of s.ES3 and ES4 during the flood (Figures 8f and 8h), possibly suggesting incipient flow separation, the lack of such zone during the ebb tide is surprising, especially in the light of the symmetrical planform of the bend and considering that the flow regime in salt marsh tidal creeks is typically ebb dominated as a result of the geomorphic structure of tidal hydrodynamics (see D'Alpaos et al., 2005; Fagherazzi et al., 2008). While separation of ebb flows might emerge for tides different from those we measured, its lack could also be due to the presence of the submerged vegetation patches in s.ES1 that can alter the inflow distribution. Both these hypotheses will be later discussed in sections 5.2 and 5.3.

Even though flow separation within fluvial and tidal meanders appears to be similarly related to the urn:x-wiley:00431397:media:wrcr24721:wrcr24721-math-0014 ratio, a unique feature of tidal meanders is that recirculation zones tend to occupy mutually exclusive portions of the channel during ebb and flood tides (Figure 5). Within the TA bend, for example, outer bank separation is observed during the ebb, when inflow located near the inner bank crosses the channel and impinges near-orthogonally against the outer bank (Figure 5a). In contrast, inner bank separation occurs during the flood near the inner bankline apex (Figure 5b) and reattaches to the main streamflow within a distance of one channel width. In both cases, a DAV recirculation eddy is formed, which is characterized by a near-vertical axis with the strongest reverse flow located close to the bank. This is by all means similar to what occurs in a sudden flow expansion (Ferguson & Parsons, 2004; Parsons et al., 2004). Analysis of 3D flow patterns reveals that the recirculation zone affects the whole cross section from the near-surface to the near-bottom region and that the curvature-induced helical flow is confined to the nonrecirculating portion of the primary flow (Figures 6c and 6d). Strong shear along flow regions oriented in opposite directions is likely responsible for the shedding of Kelvin-Helmholtz vortexes observed during the survey campaign. These vortexes are similar to those occurring at channel confluences (Sukhodolov & Rhoads, 2001) and have already been observed within sharp bends in both intertidal creeks and rivers (Leeder & Bridges, 1975; Parsons et al., 2004). Similar to the TA bend, patterns of DAVs in the SC study case are characterized by strong flow separation zones, which occur immediately downstream and upstream of the bend apexes during the ebb and the flood flow, respectively (Figures 5c and 5d). Both flow separation and the MVF operate along mutually exclusive ebb and flood streamlines, thus alternately impinging along the left and right channel banks. It is worthwhile noting that, in all the study bends, the periodic turnover of the position of recirculation zones and MVFs is often associated with curvature-induced secondary flows rotating in opposite directions within a given cross section. Notable examples are s.TA2, TA3, and TA4 (Figures 6c to 6h); s.SC3 and SC4 (Figures 7e to 7h); and s.ES4 and ES5 (Figures 8g to 8j). In some cases, however, significant secondary circulation cells are only present during either the ebb or the flood, such as in s.SC5 and SC6 (Figures 7i to 7l), as well as in s.ES2 (Figures 8c and 8d). The morphodynamic and sedimentological implications of such a periodic turnover in flow patterns will be discussed in the next section.

5.2 Implication for Tidal Meander Morphodynamics and Sedimentology

The joint analysis of the surveyed flow fields and the recent planform evolution of the study bends (i.e., from 2012 to 2018, Figure 2) has allowed us to link flow processes and form. The planform geometry of the TA bend, which widens significantly at its apex, is characteristic of fluvial meanders with flow separation at concave banks (Ferguson & Parsons, 2004; Nanson, 2010; Parsons et al., 2004). Consistent with this observation, the TA bend displays concave-bank flow separation during the ebb phase (Figure 5a). This region of separated flow protects the landward portion of the outer bank from erosion, redirecting the maximum shear stress and the MVF toward the inner bank and further downstream (Hickin, 1978). As a result, the upstream portion of the point bar is eroded, and an elongated concave-bank bench forms on the upstream limb of the channel bend (Hickin, 1979; Page & Nanson, 1982; Parker, 1976), thus causing the bends to translate seaward. Indeed, bank erosion occurs where strong ebb flows impinge along the channel banks, that is, at the inner and outer bank immediately landward and seaward of the bend apex (Figure 2d). Overall, this suggests a local ebb dominance, a hypothesis that is further reinforced by (i) the planimetric match between the concave-bank bench and the ebb flow separation zone (Figure 4, see also Hodskinson & Ferguson, 1998) and (ii) the presence of a point bar on the seaward side of the inner bank, where ebb secondary flows are directed upbar (see s.TA3 and TA4 in Figures 6e and 6g).

In spite of less pronounced channel widening and higher urn:x-wiley:00431397:media:wrcr24721:wrcr24721-math-0015 ratio compared with the TA bend, the SC study case still displays flow separation. Ebb and flood flows operating along mutually evasive streamlines impinge along channel banks upstream and downstream of the bend apexes, respectively. This process carves banklines in a cuspate fashion, a unique feature of tidal meanders (Dalrymple et al., 2012; Finotello et al., 2020; Hughes, 2012). While historical erosion patterns (i.e., from 1968 to 2012, Figure 2h) suggest the dominance of ebb flows, with preferential erosion to happen upstream of the bend apexes, the most recent bankline evolution (i.e., from 2012 to 2018) displays a similar erosion trend both upstream and downstream of the apexes, thus pointing to a more symmetric tidal regime. This might allow one to unravel the signatures of tidal asymmetry, or the lack thereof, based on the analysis of erosion and deposition patterns, and vice versa (Finotello, Canestrelli, et al., 2019; Tambroni et al., 2017). The cuspate shape of the banks produces flow separation immediately beyond the bend apex (Figures 5c and 5d). Here, recirculation eddies occupy an expansion in the concave bank created by previous bank erosion upstream of the apex (Figure 2). This generates a zone of dead velocity and promotes the formation of concave-bank benches by deposition of fine sediment (Hickin, 1979; Page & Nanson, 1982; Parker, 1976). Even when the concave-bank bench is directly subjected to the main Vp streamflow, such as in s.SC1, s.SC3, and s.SC4 during the flood (Figures 7b, 7f, and 7h), the growth of a Vs outer bank cell (Blanckaert, 2011; Dietrich & Smith, 1983; Einstein & Harder, 1954; Rozovskii, 1957) limits the growth of the main Vs cell and prevents the erosion of the bench. Secondary circulation patterns reveal that sediment delivery to the bench is generally operated by ebb flows (s.SC1, SC2, SC4; Figures 7a, 7c, and 7g), even though upbar-oriented secondary flows are observed in the most landward portion of the bench (s.SC1, SC2, SC5; Figures 7a, 7c, and 7k). This is consistent with the typical ebb flow dominance in tidal creeks, as well as with the site-specific symmetric tidal regime suggested by the recent evolution of channel banks.

Besides reducing the effective width of the main streamflow, an interesting effect of flow separation observed in the TA and SC study cases is that some areas of the tidal channels can hardly experience periodic flow reversal, a counterintuitive occurrence that challenges the paradigm of tidal channels always experiencing reversing flows. The absence of flow reversal for some portions of the considered cross sections is especially noticeable in the concave-bank benches (Figures 5a to 5d) and might hamper the formation of sedimentological proxies for bidirectional flow in related deposits (Barwis, 1978; La Croix & Dashtgard, 2014). Consistent with this observation, Leeder and Bridges (1975) analyzed flow separation in intertidal meanders and observed the development of ripple bedforms oriented in the opposite direction to the main stream. They also pointed out that a further consequence of flow separation is that the classical model of point bar deposition, which assumes a bar-hugging helical flow, cannot be valid in the case of tightly curving meander bends. Accordingly, our results suggest in most cases a limited transfer of secondary helical flow beyond the meander apex zone. These observations are particularly relevant, from sedimentological perspectives, considering that more extreme curvatures (i.e., sharp bends), and therefore more frequent flow separations, are observed in tidal meandering channels compared with meandering rivers (Finotello et al., 2020). Moreover, mutually evasive patterns of ebb and flood maximum velocity could further prevent a ubiquitous distribution of bedforms recording the signature of bidirectional flows. This effect could give rise to a suite of bedforms and grain-size distributions different from those predicted by classical facies models, as exemplified by the description of sedimentological features of the TA provided by Ghinassi, Brivio, et al. (2018).

Mutually evasive patterns of ebb and flood flows are also found in the ES bend, where less variable channel widths and a higher urn:x-wiley:00431397:media:wrcr24721:wrcr24721-math-0016 ratio prevent flow from separating. The patterns of bank erosion and point bar accretion are symmetrical to the bend apex (Figure 2l). Secondary ebb flows move sediment upbar both landward (s.ES2) and seaward (s.ES4 and ES5) of the bend apex, whereas secondary flood flows are oriented upbar only in the most landward part of the bend (s. ES2, see Figure 8). This would suggest a slight dominance of ebb flows. The latter, however, display maximum DAVs at the apex inner bank and seem therefore to be out of phase with the bar-pool pattern of the channel (Figure 4b). In addition, a double peak in DAVs and high turbulence are found landward of the bend apex (s.ES1 and ES2, see Figures 8a, 8b, and 5e). Nevertheless, pronounced turbulence and perturbed DAVs characterize the flows in the s.ES1 cross section even during the flood (Figures 5f and 8b). This is arguably due to the presence of a submerged patch of P. oceanica at s.ES1 (Figures 3c and 4b). Even though submerged vegetation is quite common within the study bends (see Figure 3), the patch observed near s.ES1 is likely to have stronger hydrodynamic effects because of its large dimension relative to the dimension of the channel cross section.

Aquatic vegetation is known to affect the flow field in small channels by enhancing bottom roughness and thereby increasing turbulence (Folkard, 2005). This hypothesis is further supported by the alternating patterns displayed by streamwise vorticity (i.e., vortexes with axes parallel to the streamwise primary velocity, Figure 9), which resemble those observed by Stoesser et al. (2015) in channels with nonuniformly roughed bed. Analysis of the available photos reveals that the patch size changed over time and was at its maximum in 2016 when the flow was measured (Figures 9c to 9f). On the one hand, the presence of this vegetation patch justifies the perturbed flow fields that we observe in our study cases. On the other hand, its hydrodynamic effects suggest that submerged vegetation could potentially alter the hydrodynamics, and therefore the planform evolution, of meandering salt marsh creeks. In general, aquatic vegetation can modify flow and transport within open channels (e.g., Nepf et al., 2013; Nepf & Ghisalberti, 2008). In particular, submerged flexible vegetation within channel bends (i) suppresses curvature-induced secondary flows; (ii) alters turbulent structures, thus possibly reducing sediment transport at the banks; and (iii) alters the position of DAV MVF (Schnauder & Sukhodolov, 2012; Termini, 20132016). Even though macrophyte cover may change over time and depend on the season (Schnauder & Sukhodolov, 2012; Figures 9c to 9f), limited transport capacity of secondary circulations by aquatic plants might have direct consequences on point bar growth and meander migration rates. This would ultimately affect the meander planform morphodynamics, especially in tidal creeks no wider than few meters, which are incidentally the most widespread across salt marshes.

Details are in the caption following the image
Effects of submerged vegetation. (a, b) Streamwise vorticity observed in s.ES1 during the ebb (a) and the flood (b). Vorticity is computed according to the zero-net secondary discharge definition. Negative (positive) values correspond to counterclockwise (clockwise) rotating vortexes with axes parallel to the streamwise primary velocity. The position of the Posidonia oceanica patch is also highlighted. (c–f) Evolution of the P. oceanica patch in s.ES1 as observed from satellite images from 2012 to 2018 (images ©Google, Landsat/Copernicus).

5.3 Relationship Between Bend Tightness and Flow Separation

The growth of flow separation zones in tidal channels hundreds of meters wide (Finotello, Canestrelli, et al., 2019; Ghinassi, D'Alpaos, et al., 2018) suggests that they are not restricted to small salt marsh creeks. However, it should be questioned whether or not flow recirculation would exist for flow conditions different from those associated with spring tides. Previous research on river meanders highlighted that patterns of flow recirculation do not necessarily remain similar as hydrological conditions change through time (Andrle, 1994; Ferguson & Parsons, 2004; Keevil et al., 2015; Markham & Thorne, 1992; Parsons et al., 2004). Although the survey of a complete neap-spring tidal cycle would be necessary to settle the dispute, we could speculate upon this based on the results proposed by Leeder and Bridges (1975) for intertidal meanders in the Solway Firth (Scotland).

They suggested that flow separation in meander bends can be expressed as a function of bend tightness (R/B) and Froude number (Fr). Assuming that R does not vary significantly with changing water elevation, we computed the cross-sectional width BD corresponding to different water depths (D) for one cross section of each study case bend, namely, s.TA2, s.SC3, and s.ES3. Therefore, we were able to calculate how the R-BD ratio changes according to varying water levels. Similarly, we also calculated the cross-sectional area (AD) and the wetted cross-section perimeter (CD) corresponding to different values of water surface elevation. Then, assuming uniform flow conditions, we estimated a characteristic flow velocity urn:x-wiley:00431397:media:wrcr24721:wrcr24721-math-0017, where the Gaukler-Strickler coefficient was assumed equal to ks = 30 m1/3·s−1 for all the study cases, to a first approximation, and i = 10−5 represents a characteristic gradient of the water free surface that drives flow motion in salt marsh settings (D'Alpaos et al., 2005; Marani et al., 2003; Rinaldo et al., 1999). Finally, we computed the Froude number for different water depths as Fr = U/(gD)1/2, where g denotes the gravitational acceleration.

Plotting of R/BD against Fr shows that our results are in good agreement with Leeder and Bridges' (1975) data (Figure 10). In particular, data points for the TA bend plot consistently below the curve discriminating between meanders with and without flow separation, with the R-BD ratio never exceeding 1.5. In contrast, flow separation in the SC study case can occur only for near-bankfull flow depths. This is likely due to the peculiar morphology of the bend, whereby BD drops from 25 m for D/Db = 1, where Db denotes bankfull depth, to 12 m for D/Db = 0.2, thus causing an increase in R/BD from 1.32 to 5.28. In addition, channel banklines of the SC bends are much more sinuous compared with the channel thalweg, so that R would significantly increase as D/Db decreases. This would effectively increase R/BD, thus preventing flow separation even further. Finally, all the data from the ES bend fall above the discriminant curve, suggesting that flow separation is unlikely to occur in this bend even for flow conditions different from bankfull.

Details are in the caption following the image
Dimensionless graph showing the occurrence of flow separation in meander bends (adapted from Leeder and Bridges, 1975). Values of radius of curvature (R) normalized by the effective channel widths (B) corresponding to different water depth (D) are reported along the vertical axis, whereas the horizontal axis contains values of the Froude number (Fr) calculated for different water depths (D) and the related uniform flow velocity (U). Data from this study are color coded according to the ratio between the considered channel depth (D) and the bankfull depth (Db). Original data points from Leeder and Bridges (1975) are also reported.

These results corroborate the general validity of our measurements and confirm that flow separation in tidal meandering creeks is essentially related to the tightness of the bends, thus further supporting the conclusions we drew in the previous sections. Moreover, the nondimensional treatment we employed could potentially allow one to upscale our results to tidal channels much larger than those we analyzed, owing to the approximately linear relationships that both B and R bear with tidal meander wavelengths (Finotello et al., 2020; Marani et al., 2002). Nonetheless, care has to be taken in linearly upscaling the results, since hydraulic geometry is essentially nonlinear (Kästner et al., 2017; Leopold & Maddock, 1953). Given that channel depth typically varies less rapidly than channel width as channel size changes, the relative intensities of secondary flows will be likely reduced in meanders larger than those analyzed here. Moreover, more detailed analyses will be required in order to refine and extend our results to water depth much higher than the bankfull, such as those that are established, for example, during storm surge conditions. In this case, the confinement of the flow within the channel is progressively reduced as water depth increases, thus possibly modifying the stage-velocity relationship that normally characterizes the study case channels (Kearney et al., 2017).

6 Conclusions

This paper contributes to the understanding of the dynamics of meandering tidal channels by documenting 3D velocity fields and patterns of erosion and deposition in three different meander loops found within the microtidal Venice lagoon (Italy). The main conclusions of this paper can be summarized as follows.
  1. Mutually evasive paths of the ebb and flood maximum velocity streams during a tidal cycle are a common feature of tidal meanders, regardless of the considered meander planform morphology and evolution dynamics.
  2. Similar to fluvial meanders, secondary circulations arise from the unbalancing between centrifugal force and the lateral pressure gradient driven by the curvature-induced superelevation of the free surface. However, in most tidal meanders, the orientation of secondary circulations changes periodically in response to ebb and flood tides. In addition, smaller outer bank secondary cells are also observed, which keep the main secondary cell far from channel banks, thus inhibiting bank erosion.
  3. Flow separation is controlled by the radius-width ratio of the bend and is to be expected in many tidal meanders owing to the high curvature values that they typically display. The effects of flow separation are (i) a decrease in the effective channel width and therefore a significant increase in local flow velocities and the associated bank erosion and (ii) the formation of dead-velocity zones with upstream-recirculating flows, wherein enhanced deposition of fine sediment forms point bar and concave-bank bench deposits if flow separation occurs at the inner or outer bank, respectively.
  4. Recirculation zones can locally inhibit flow reversal between ebb and flood flow, giving rise to unidirectional sedimentary features such as asymmetric bedforms and grain-size distributions that are different from those predicted by classical sedimentary facies models for bidirectional flows.
  5. Submerged vegetation patches of a size comparable with the channel cross section can significantly enhance bottom roughness and increase local turbulence, thereby disrupting and limiting the transfer of curvature-induced helical flow, with direct implications for the morphodynamic evolution of meander bends.

Even though velocity measurements obtained in this study only monitored flow fields associated with spring tides, the potential for flow separation has been estimated also for sub-bankfull stage conditions. A comparison of our results with literature data suggests that possible differences between bankfull and sub-bankfull stages may emerge as a consequence of the 3D meander morphology, but overall supports the idea that flow separation is a function of bend tightness and Froude number. Patterns of channel migration and flow fields within our study bends also indicate that meander evolution is controlled by both local and nonlocal channel morphologies and dynamics (Furbish, 1991; Güneralp & Rhoads, 2010; Stolum, 1998). Additional field and modeling effort would be required to investigate how different inflow conditions, and changes in local hydrodynamics due to spatial interactions among adjacent meander loops, can alter the planform evolution of individual meander bends found within complex tidal channel networks.

Acknowledgments

This work was supported by the University of Padova project “From channels to rock record: Morphodynamic evolution of tidal meanders and related sedimentary products” (BIRD168939) and by the “HYDROSEM: Fluvial and tidal meanders of the venetian-Po plain: From hydrodynamics to stratigraphy” project (Progetto di Eccellenza Cariparo 2017) which are gratefully acknowledged. We acknowledge thorough and insightful comments from two anonymous reviewers.

    Data Availability Statement

    The data sets generated and/or analyzed during the current study are freely available at https://doi.org/10.5281/zenodo.3774454.